id sid tid token lemma pos fbem-26173 1 1 frontiers frontier NOUN fbem-26173 1 2 in in ADP fbem-26173 1 3 business business NOUN fbem-26173 1 4 , , PUNCT fbem-26173 1 5 economics economic NOUN fbem-26173 1 6 and and CCONJ fbem-26173 1 7 management management NOUN fbem-26173 1 8 issn issn PROPN fbem-26173 1 9 : : PUNCT fbem-26173 1 10 2766 2766 NUM fbem-26173 1 11 - - PUNCT fbem-26173 1 12 824x 824x NUM fbem-26173 1 13 | | ADJ fbem-26173 1 14 vol vol NOUN fbem-26173 1 15 . . PUNCT fbem-26173 2 1 16 16 NUM fbem-26173 2 2 , , PUNCT fbem-26173 2 3 no no INTJ fbem-26173 2 4 . . NOUN fbem-26173 2 5 3 3 NUM fbem-26173 2 6 , , PUNCT fbem-26173 2 7 2024 2024 NUM fbem-26173 2 8 76 76 NUM fbem-26173 3 1 cnn cnn PROPN fbem-26173 3 2 - - PUNCT fbem-26173 3 3 bilstm bilstm NOUN fbem-26173 3 4 - - PUNCT fbem-26173 3 5 attention attention NOUN fbem-26173 3 6 based base VERB fbem-26173 3 7 stock stock NOUN fbem-26173 3 8 price price NOUN fbem-26173 3 9 prediction prediction NOUN fbem-26173 3 10 and and CCONJ fbem-26173 3 11 quantitative quantitative ADJ fbem-26173 3 12 investment investment NOUN fbem-26173 3 13 strategy strategy NOUN fbem-26173 3 14 chunzhong chunzhong NOUN fbem-26173 3 15 li li PROPN fbem-26173 4 1 * * PUNCT fbem-26173 4 2 , , PUNCT fbem-26173 4 3 weiqi weiqi PROPN fbem-26173 4 4 hua hua PROPN fbem-26173 4 5 college college PROPN fbem-26173 4 6 of of ADP fbem-26173 4 7 statistics statistic NOUN fbem-26173 4 8 and and CCONJ fbem-26173 4 9 applied applied ADJ fbem-26173 4 10 mathematics mathematic NOUN fbem-26173 4 11 , , PUNCT fbem-26173 4 12 anhui anhui PROPN fbem-26173 4 13 university university PROPN fbem-26173 4 14 of of ADP fbem-26173 4 15 finance finance NOUN fbem-26173 4 16 and and CCONJ fbem-26173 4 17 economics economic NOUN fbem-26173 4 18 , , PUNCT fbem-26173 4 19 bengbu bengbu NOUN fbem-26173 4 20 233030 233030 NUM fbem-26173 4 21 , , PUNCT fbem-26173 4 22 china china PROPN fbem-26173 4 23 * * PUNCT fbem-26173 4 24 corresponding correspond VERB fbem-26173 4 25 author author NOUN fbem-26173 4 26 : : PUNCT fbem-26173 4 27 chunzhong chunzhong PROPN fbem-26173 4 28 li li PROPN fbem-26173 4 29 ( ( PUNCT fbem-26173 4 30 email email NOUN fbem-26173 4 31 : : PUNCT fbem-26173 4 32 czhongli@163.com czhongli@163.com NOUN fbem-26173 4 33 ) ) PUNCT fbem-26173 4 34 abstract abstract NOUN fbem-26173 4 35 : : PUNCT fbem-26173 4 36 with with ADP fbem-26173 4 37 the the DET fbem-26173 4 38 rapid rapid ADJ fbem-26173 4 39 development development NOUN fbem-26173 4 40 of of ADP fbem-26173 4 41 technology technology NOUN fbem-26173 4 42 and and CCONJ fbem-26173 4 43 the the DET fbem-26173 4 44 increasing increase VERB fbem-26173 4 45 maturity maturity NOUN fbem-26173 4 46 of of ADP fbem-26173 4 47 financial financial ADJ fbem-26173 4 48 markets market NOUN fbem-26173 4 49 , , PUNCT fbem-26173 4 50 stock stock NOUN fbem-26173 4 51 price price NOUN fbem-26173 4 52 prediction prediction NOUN fbem-26173 4 53 has have AUX fbem-26173 4 54 become become VERB fbem-26173 4 55 a a DET fbem-26173 4 56 hot hot ADJ fbem-26173 4 57 topic topic NOUN fbem-26173 4 58 and and CCONJ fbem-26173 4 59 an an DET fbem-26173 4 60 important important ADJ fbem-26173 4 61 trend trend NOUN fbem-26173 4 62 in in ADP fbem-26173 4 63 the the DET fbem-26173 4 64 financial financial ADJ fbem-26173 4 65 field field NOUN fbem-26173 4 66 . . PUNCT fbem-26173 5 1 however however ADV fbem-26173 5 2 , , PUNCT fbem-26173 5 3 the the DET fbem-26173 5 4 stock stock NOUN fbem-26173 5 5 market market NOUN fbem-26173 5 6 has have VERB fbem-26173 5 7 complexity complexity NOUN fbem-26173 5 8 and and CCONJ fbem-26173 5 9 volatility volatility NOUN fbem-26173 5 10 , , PUNCT fbem-26173 5 11 in in ADP fbem-26173 5 12 order order NOUN fbem-26173 5 13 to to PART fbem-26173 5 14 reduce reduce VERB fbem-26173 5 15 the the DET fbem-26173 5 16 investment investment NOUN fbem-26173 5 17 risk risk NOUN fbem-26173 5 18 and and CCONJ fbem-26173 5 19 ensure ensure VERB fbem-26173 5 20 the the DET fbem-26173 5 21 maximization maximization NOUN fbem-26173 5 22 of of ADP fbem-26173 5 23 benefits benefit NOUN fbem-26173 5 24 , , PUNCT fbem-26173 5 25 selecting select VERB fbem-26173 5 26 the the DET fbem-26173 5 27 optimal optimal ADJ fbem-26173 5 28 stock stock NOUN fbem-26173 5 29 to to PART fbem-26173 5 30 predict predict VERB fbem-26173 5 31 the the DET fbem-26173 5 32 stock stock NOUN fbem-26173 5 33 price price NOUN fbem-26173 5 34 and and CCONJ fbem-26173 5 35 formulating formulate VERB fbem-26173 5 36 the the DET fbem-26173 5 37 quantitative quantitative ADJ fbem-26173 5 38 trading trading NOUN fbem-26173 5 39 strategy strategy NOUN fbem-26173 5 40 are be AUX fbem-26173 5 41 extremely extremely ADV fbem-26173 5 42 important important ADJ fbem-26173 5 43 issues issue NOUN fbem-26173 5 44 for for ADP fbem-26173 5 45 financial financial ADJ fbem-26173 5 46 academics academic NOUN fbem-26173 5 47 and and CCONJ fbem-26173 5 48 investors investor NOUN fbem-26173 5 49 . . PUNCT fbem-26173 6 1 for for ADP fbem-26173 6 2 the the DET fbem-26173 6 3 research research NOUN fbem-26173 6 4 of of ADP fbem-26173 6 5 stock stock NOUN fbem-26173 6 6 price price NOUN fbem-26173 6 7 prediction prediction NOUN fbem-26173 6 8 and and CCONJ fbem-26173 6 9 quantitative quantitative ADJ fbem-26173 6 10 trading trading NOUN fbem-26173 6 11 , , PUNCT fbem-26173 6 12 firstly firstly ADV fbem-26173 6 13 , , PUNCT fbem-26173 6 14 the the DET fbem-26173 6 15 quantitative quantitative ADJ fbem-26173 6 16 stock stock NOUN fbem-26173 6 17 selection selection NOUN fbem-26173 6 18 model model NOUN fbem-26173 6 19 combining combine VERB fbem-26173 6 20 xgboos xgboos PROPN fbem-26173 6 21 t t PROPN fbem-26173 6 22 and and CCONJ fbem-26173 6 23 multi multi ADJ fbem-26173 6 24 - - ADJ fbem-26173 6 25 factor factor NOUN fbem-26173 6 26 stock stock NOUN fbem-26173 6 27 selection selection NOUN fbem-26173 6 28 model model NOUN fbem-26173 6 29 is be AUX fbem-26173 6 30 constructed construct VERB fbem-26173 6 31 , , PUNCT fbem-26173 6 32 and and CCONJ fbem-26173 6 33 four four NUM fbem-26173 6 34 stocks stock NOUN fbem-26173 6 35 are be AUX fbem-26173 6 36 screened screen VERB fbem-26173 6 37 out out ADP fbem-26173 6 38 , , PUNCT fbem-26173 6 39 and and CCONJ fbem-26173 6 40 then then ADV fbem-26173 6 41 the the DET fbem-26173 6 42 cnn cnn PROPN fbem-26173 6 43 - - PUNCT fbem-26173 6 44 bilstm bilstm NOUN fbem-26173 6 45 - - PUNCT fbem-26173 6 46 attention attention NOUN fbem-26173 6 47 model model NOUN fbem-26173 6 48 is be AUX fbem-26173 6 49 proposed propose VERB fbem-26173 6 50 to to PART fbem-26173 6 51 predict predict VERB fbem-26173 6 52 the the DET fbem-26173 6 53 stock stock NOUN fbem-26173 6 54 price price NOUN fbem-26173 6 55 trend trend NOUN fbem-26173 6 56 of of ADP fbem-26173 6 57 the the DET fbem-26173 6 58 selected select VERB fbem-26173 6 59 stocks stock NOUN fbem-26173 6 60 , , PUNCT fbem-26173 6 61 and and CCONJ fbem-26173 6 62 it it PRON fbem-26173 6 63 is be AUX fbem-26173 6 64 found find VERB fbem-26173 6 65 from from ADP fbem-26173 6 66 the the DET fbem-26173 6 67 prediction prediction NOUN fbem-26173 6 68 results result NOUN fbem-26173 6 69 of of ADP fbem-26173 6 70 the the DET fbem-26173 6 71 four four NUM fbem-26173 6 72 stocks stock NOUN fbem-26173 6 73 that that PRON fbem-26173 6 74 the the DET fbem-26173 6 75 prediction prediction NOUN fbem-26173 6 76 accuracies accuracy NOUN fbem-26173 6 77 all all PRON fbem-26173 6 78 reach reach VERB fbem-26173 6 79 more more ADJ fbem-26173 6 80 than than ADP fbem-26173 6 81 95 95 NUM fbem-26173 6 82 % % NOUN fbem-26173 6 83 , , PUNCT fbem-26173 6 84 which which PRON fbem-26173 6 85 is be AUX fbem-26173 6 86 higher high ADJ fbem-26173 6 87 than than ADP fbem-26173 6 88 that that PRON fbem-26173 6 89 of of ADP fbem-26173 6 90 the the DET fbem-26173 6 91 single single ADJ fbem-26173 6 92 model model NOUN fbem-26173 6 93 prediction prediction NOUN fbem-26173 6 94 . . PUNCT fbem-26173 7 1 accuracy accuracy NOUN fbem-26173 8 1 and and CCONJ fbem-26173 8 2 passed pass VERB fbem-26173 8 3 the the DET fbem-26173 8 4 validity validity NOUN fbem-26173 8 5 test test NOUN fbem-26173 8 6 of of ADP fbem-26173 8 7 the the DET fbem-26173 8 8 fitted fit VERB fbem-26173 8 9 model model NOUN fbem-26173 8 10 . . PUNCT fbem-26173 9 1 secondly secondly ADV fbem-26173 9 2 , , PUNCT fbem-26173 9 3 based base VERB fbem-26173 9 4 on on ADP fbem-26173 9 5 the the DET fbem-26173 9 6 quantitative quantitative ADJ fbem-26173 9 7 investment investment NOUN fbem-26173 9 8 portfolio portfolio NOUN fbem-26173 9 9 given give VERB fbem-26173 9 10 in in ADP fbem-26173 9 11 the the DET fbem-26173 9 12 above above ADJ fbem-26173 9 13 analysis analysis NOUN fbem-26173 9 14 and and CCONJ fbem-26173 9 15 the the DET fbem-26173 9 16 quantitative quantitative ADJ fbem-26173 9 17 stock stock NOUN fbem-26173 9 18 trading trading NOUN fbem-26173 9 19 strategy strategy NOUN fbem-26173 9 20 of of ADP fbem-26173 9 21 macd macd NOUN fbem-26173 9 22 , , PUNCT fbem-26173 9 23 the the DET fbem-26173 9 24 prediction prediction NOUN fbem-26173 9 25 results result NOUN fbem-26173 9 26 are be AUX fbem-26173 9 27 verified verify VERB fbem-26173 9 28 , , PUNCT fbem-26173 9 29 and and CCONJ fbem-26173 9 30 the the DET fbem-26173 9 31 total total ADJ fbem-26173 9 32 return return NOUN fbem-26173 9 33 of of ADP fbem-26173 9 34 most most ADJ fbem-26173 9 35 stocks stock NOUN fbem-26173 9 36 based base VERB fbem-26173 9 37 on on ADP fbem-26173 9 38 the the DET fbem-26173 9 39 strategy strategy NOUN fbem-26173 9 40 is be AUX fbem-26173 9 41 higher high ADJ fbem-26173 9 42 than than ADP fbem-26173 9 43 40 40 NUM fbem-26173 9 44 % % NOUN fbem-26173 9 45 at at ADP fbem-26173 9 46 the the DET fbem-26173 9 47 highest high ADJ fbem-26173 9 48 , , PUNCT fbem-26173 9 49 and and CCONJ fbem-26173 9 50 the the DET fbem-26173 9 51 loss loss NOUN fbem-26173 9 52 is be AUX fbem-26173 9 53 controlled control VERB fbem-26173 9 54 within within ADP fbem-26173 9 55 10 10 NUM fbem-26173 9 56 % % NOUN fbem-26173 9 57 , , PUNCT fbem-26173 9 58 which which PRON fbem-26173 9 59 indicates indicate VERB fbem-26173 9 60 that that SCONJ fbem-26173 9 61 the the DET fbem-26173 9 62 trading trading NOUN fbem-26173 9 63 strategy strategy NOUN fbem-26173 9 64 in in ADP fbem-26173 9 65 this this DET fbem-26173 9 66 paper paper NOUN fbem-26173 9 67 is be AUX fbem-26173 9 68 effective effective ADJ fbem-26173 9 69 and and CCONJ fbem-26173 9 70 feasible feasible ADJ fbem-26173 9 71 . . PUNCT fbem-26173 10 1 the the DET fbem-26173 10 2 study study NOUN fbem-26173 10 3 concludes conclude VERB fbem-26173 10 4 that that SCONJ fbem-26173 10 5 greater great ADJ fbem-26173 10 6 returns return NOUN fbem-26173 10 7 can can AUX fbem-26173 10 8 be be AUX fbem-26173 10 9 obtained obtain VERB fbem-26173 10 10 by by ADP fbem-26173 10 11 calculating calculate VERB fbem-26173 10 12 the the DET fbem-26173 10 13 total total ADJ fbem-26173 10 14 returns return NOUN fbem-26173 10 15 of of ADP fbem-26173 10 16 stock stock NOUN fbem-26173 10 17 portfolios portfolio NOUN fbem-26173 10 18 based base VERB fbem-26173 10 19 on on ADP fbem-26173 10 20 different different ADJ fbem-26173 10 21 portfolio portfolio NOUN fbem-26173 10 22 approaches approach NOUN fbem-26173 10 23 . . PUNCT fbem-26173 11 1 keywords keyword NOUN fbem-26173 11 2 : : PUNCT fbem-26173 11 3 stock stock NOUN fbem-26173 11 4 price price NOUN fbem-26173 11 5 prediction prediction NOUN fbem-26173 11 6 , , PUNCT fbem-26173 11 7 quantitative quantitative ADJ fbem-26173 11 8 trading trading NOUN fbem-26173 11 9 , , PUNCT fbem-26173 11 10 xgboost xgboost X fbem-26173 11 11 multifactor multifactor NOUN fbem-26173 11 12 stock stock NOUN fbem-26173 11 13 picking picking NOUN fbem-26173 11 14 , , PUNCT fbem-26173 11 15 deep deep ADJ fbem-26173 11 16 learning learning NOUN fbem-26173 11 17 . . PUNCT fbem-26173 12 1 1 1 X fbem-26173 12 2 . . X fbem-26173 12 3 introduction introduction NOUN fbem-26173 12 4 stock stock NOUN fbem-26173 12 5 market market NOUN fbem-26173 12 6 trends trend NOUN fbem-26173 12 7 reveal reveal VERB fbem-26173 12 8 the the DET fbem-26173 12 9 direction direction NOUN fbem-26173 12 10 of of ADP fbem-26173 12 11 national national ADJ fbem-26173 12 12 economic economic ADJ fbem-26173 12 13 development development NOUN fbem-26173 12 14 , , PUNCT fbem-26173 12 15 is be AUX fbem-26173 12 16 an an DET fbem-26173 12 17 important important ADJ fbem-26173 12 18 indicator indicator NOUN fbem-26173 12 19 of of ADP fbem-26173 12 20 economic economic ADJ fbem-26173 12 21 development development NOUN fbem-26173 12 22 potential potential NOUN fbem-26173 12 23 , , PUNCT fbem-26173 12 24 is be AUX fbem-26173 12 25 an an DET fbem-26173 12 26 essential essential ADJ fbem-26173 12 27 and and CCONJ fbem-26173 12 28 important important ADJ fbem-26173 12 29 part part NOUN fbem-26173 12 30 of of ADP fbem-26173 12 31 the the DET fbem-26173 12 32 socialist socialist ADJ fbem-26173 12 33 market market NOUN fbem-26173 12 34 economy economy NOUN fbem-26173 12 35 , , PUNCT fbem-26173 12 36 more more ADJ fbem-26173 12 37 and and CCONJ fbem-26173 12 38 more more ADJ fbem-26173 12 39 scholars scholar NOUN fbem-26173 12 40 in in ADP fbem-26173 12 41 the the DET fbem-26173 12 42 financial financial ADJ fbem-26173 12 43 sector sector NOUN fbem-26173 12 44 began begin VERB fbem-26173 12 45 to to PART fbem-26173 12 46 study study VERB fbem-26173 12 47 it it PRON fbem-26173 12 48 . . PUNCT fbem-26173 13 1 with with ADP fbem-26173 13 2 the the DET fbem-26173 13 3 development development NOUN fbem-26173 13 4 of of ADP fbem-26173 13 5 the the DET fbem-26173 13 6 economy economy NOUN fbem-26173 13 7 , , PUNCT fbem-26173 13 8 people people NOUN fbem-26173 13 9 's 's PART fbem-26173 13 10 income income NOUN fbem-26173 13 11 level level NOUN fbem-26173 13 12 increases increase NOUN fbem-26173 13 13 , , PUNCT fbem-26173 13 14 there there PRON fbem-26173 13 15 are be VERB fbem-26173 13 16 more more ADJ fbem-26173 13 17 deposits deposit NOUN fbem-26173 13 18 for for ADP fbem-26173 13 19 securities security NOUN fbem-26173 13 20 investment investment NOUN fbem-26173 13 21 , , PUNCT fbem-26173 13 22 the the DET fbem-26173 13 23 stock stock NOUN fbem-26173 13 24 market market NOUN fbem-26173 13 25 provides provide VERB fbem-26173 13 26 investors investor NOUN fbem-26173 13 27 with with ADP fbem-26173 13 28 a a DET fbem-26173 13 29 huge huge ADJ fbem-26173 13 30 profit profit NOUN fbem-26173 13 31 space space NOUN fbem-26173 13 32 , , PUNCT fbem-26173 13 33 attracting attract VERB fbem-26173 13 34 more more ADJ fbem-26173 13 35 and and CCONJ fbem-26173 13 36 more more ADJ fbem-26173 13 37 people people NOUN fbem-26173 13 38 's 's PART fbem-26173 13 39 attention attention NOUN fbem-26173 13 40 , , PUNCT fbem-26173 13 41 the the DET fbem-26173 13 42 stock stock NOUN fbem-26173 13 43 price price NOUN fbem-26173 13 44 trend trend NOUN fbem-26173 13 45 is be AUX fbem-26173 13 46 one one NUM fbem-26173 13 47 of of ADP fbem-26173 13 48 the the DET fbem-26173 13 49 most most ADV fbem-26173 13 50 concerned concerned ADJ fbem-26173 13 51 about about ADP fbem-26173 13 52 the the DET fbem-26173 13 53 problem problem NOUN fbem-26173 13 54 of of ADP fbem-26173 13 55 investors investor NOUN fbem-26173 13 56 . . PUNCT fbem-26173 14 1 therefore therefore ADV fbem-26173 14 2 , , PUNCT fbem-26173 14 3 how how SCONJ fbem-26173 14 4 to to PART fbem-26173 14 5 choose choose VERB fbem-26173 14 6 the the DET fbem-26173 14 7 optimal optimal ADJ fbem-26173 14 8 stock stock NOUN fbem-26173 14 9 and and CCONJ fbem-26173 14 10 predict predict VERB fbem-26173 14 11 the the DET fbem-26173 14 12 stock stock NOUN fbem-26173 14 13 return return NOUN fbem-26173 14 14 is be AUX fbem-26173 14 15 extremely extremely ADV fbem-26173 14 16 important important ADJ fbem-26173 14 17 to to ADP fbem-26173 14 18 the the DET fbem-26173 14 19 financial financial ADJ fbem-26173 14 20 academia academia NOUN fbem-26173 14 21 and and CCONJ fbem-26173 14 22 individual individual ADJ fbem-26173 14 23 investors investor NOUN fbem-26173 14 24 . . PUNCT fbem-26173 15 1 therefore therefore ADV fbem-26173 15 2 , , PUNCT fbem-26173 15 3 choosing choose VERB fbem-26173 15 4 the the DET fbem-26173 15 5 appropriate appropriate ADJ fbem-26173 15 6 method method NOUN fbem-26173 15 7 to to PART fbem-26173 15 8 extract extract VERB fbem-26173 15 9 effective effective ADJ fbem-26173 15 10 information information NOUN fbem-26173 15 11 from from ADP fbem-26173 15 12 the the DET fbem-26173 15 13 huge huge ADJ fbem-26173 15 14 amount amount NOUN fbem-26173 15 15 of of ADP fbem-26173 15 16 financial financial ADJ fbem-26173 15 17 data datum NOUN fbem-26173 15 18 to to PART fbem-26173 15 19 forecast forecast VERB fbem-26173 15 20 the the DET fbem-26173 15 21 stock stock NOUN fbem-26173 15 22 market market NOUN fbem-26173 15 23 has have AUX fbem-26173 15 24 become become VERB fbem-26173 15 25 an an DET fbem-26173 15 26 urgent urgent ADJ fbem-26173 15 27 problem problem NOUN fbem-26173 15 28 in in ADP fbem-26173 15 29 china china PROPN fbem-26173 15 30 . . PUNCT fbem-26173 16 1 in in ADP fbem-26173 16 2 recent recent ADJ fbem-26173 16 3 years year NOUN fbem-26173 16 4 , , PUNCT fbem-26173 16 5 research research NOUN fbem-26173 16 6 in in ADP fbem-26173 16 7 the the DET fbem-26173 16 8 field field NOUN fbem-26173 16 9 of of ADP fbem-26173 16 10 stock stock NOUN fbem-26173 16 11 price price NOUN fbem-26173 16 12 trend trend NOUN fbem-26173 16 13 prediction prediction NOUN fbem-26173 16 14 has have AUX fbem-26173 16 15 continued continue VERB fbem-26173 16 16 to to PART fbem-26173 16 17 make make VERB fbem-26173 16 18 new new ADJ fbem-26173 16 19 progress progress NOUN fbem-26173 16 20 , , PUNCT fbem-26173 16 21 and and CCONJ fbem-26173 16 22 a a DET fbem-26173 16 23 variety variety NOUN fbem-26173 16 24 of of ADP fbem-26173 16 25 methods method NOUN fbem-26173 16 26 have have AUX fbem-26173 16 27 emerged emerge VERB fbem-26173 16 28 , , PUNCT fbem-26173 16 29 which which PRON fbem-26173 16 30 can can AUX fbem-26173 16 31 be be AUX fbem-26173 16 32 categorized categorize VERB fbem-26173 16 33 into into ADP fbem-26173 16 34 three three NUM fbem-26173 16 35 main main ADJ fbem-26173 16 36 categories category NOUN fbem-26173 16 37 : : PUNCT fbem-26173 16 38 time time NOUN fbem-26173 16 39 series series PROPN fbem-26173 16 40 prediction prediction NOUN fbem-26173 16 41 methods method NOUN fbem-26173 16 42 , , PUNCT fbem-26173 16 43 traditional traditional ADJ fbem-26173 16 44 machine machine NOUN fbem-26173 16 45 learning learning NOUN fbem-26173 16 46 and and CCONJ fbem-26173 16 47 deep deep ADJ fbem-26173 16 48 learning learning NOUN fbem-26173 16 49 methods method NOUN fbem-26173 16 50 . . PUNCT fbem-26173 17 1 the the DET fbem-26173 17 2 evolution evolution NOUN fbem-26173 17 3 of of ADP fbem-26173 17 4 this this DET fbem-26173 17 5 field field NOUN fbem-26173 17 6 can can AUX fbem-26173 17 7 be be AUX fbem-26173 17 8 traced trace VERB fbem-26173 17 9 back back ADV fbem-26173 17 10 to to ADP fbem-26173 17 11 the the DET fbem-26173 17 12 initial initial ADJ fbem-26173 17 13 stage stage NOUN fbem-26173 17 14 of of ADP fbem-26173 17 15 constructing construct VERB fbem-26173 17 16 models model NOUN fbem-26173 17 17 using use VERB fbem-26173 17 18 traditional traditional ADJ fbem-26173 17 19 linear linear ADJ fbem-26173 17 20 regression regression NOUN fbem-26173 17 21 , , PUNCT fbem-26173 17 22 and and CCONJ fbem-26173 17 23 with with ADP fbem-26173 17 24 the the DET fbem-26173 17 25 boom boom NOUN fbem-26173 17 26 in in ADP fbem-26173 17 27 artificial artificial ADJ fbem-26173 17 28 intelligence intelligence NOUN fbem-26173 17 29 , , PUNCT fbem-26173 17 30 new new ADJ fbem-26173 17 31 forecasting forecasting NOUN fbem-26173 17 32 algorithms algorithm NOUN fbem-26173 17 33 and and CCONJ fbem-26173 17 34 implementation implementation NOUN fbem-26173 17 35 tools tool NOUN fbem-26173 17 36 have have AUX fbem-26173 17 37 emerged emerge VERB fbem-26173 17 38 . . PUNCT fbem-26173 18 1 box box PROPN fbem-26173 18 2 and and CCONJ fbem-26173 18 3 jenkins jenkins PROPN fbem-26173 18 4 created create VERB fbem-26173 18 5 a a DET fbem-26173 18 6 stochastic stochastic ADJ fbem-26173 18 7 time time NOUN fbem-26173 18 8 series series NOUN fbem-26173 18 9 analysis analysis NOUN fbem-26173 18 10 model model NOUN fbem-26173 18 11 autoregressive autoregressive VERB fbem-26173 18 12 sliding slide VERB fbem-26173 18 13 average average ADJ fbem-26173 18 14 model model NOUN fbem-26173 18 15 ( ( PUNCT fbem-26173 18 16 arma arma PROPN fbem-26173 18 17 ) ) PUNCT fbem-26173 18 18 in in ADP fbem-26173 18 19 the the DET fbem-26173 18 20 1970s 1970 NOUN fbem-26173 18 21 [ [ X fbem-26173 18 22 1 1 NUM fbem-26173 18 23 ] ] PUNCT fbem-26173 18 24 . . PUNCT fbem-26173 19 1 ariyo ariyo PROPN fbem-26173 19 2 ( ( PUNCT fbem-26173 19 3 2014 2014 NUM fbem-26173 19 4 ) ) PUNCT fbem-26173 20 1 [ [ X fbem-26173 20 2 2 2 NUM fbem-26173 20 3 ] ] PUNCT fbem-26173 20 4 utilized utilize VERB fbem-26173 20 5 the the DET fbem-26173 20 6 stock stock NOUN fbem-26173 20 7 data datum NOUN fbem-26173 20 8 from from ADP fbem-26173 20 9 the the DET fbem-26173 20 10 new new PROPN fbem-26173 20 11 york york PROPN fbem-26173 20 12 stock stock PROPN fbem-26173 20 13 exchange exchange PROPN fbem-26173 20 14 and and CCONJ fbem-26173 20 15 the the DET fbem-26173 20 16 nigerian nigerian ADJ fbem-26173 20 17 stock stock NOUN fbem-26173 20 18 exchange exchange NOUN fbem-26173 20 19 to to PART fbem-26173 20 20 build build VERB fbem-26173 20 21 an an DET fbem-26173 20 22 arima arima NOUN fbem-26173 20 23 stock stock NOUN fbem-26173 20 24 price price NOUN fbem-26173 20 25 prediction prediction NOUN fbem-26173 20 26 model model NOUN fbem-26173 20 27 , , PUNCT fbem-26173 20 28 and and CCONJ fbem-26173 20 29 experiments experiment NOUN fbem-26173 20 30 have have AUX fbem-26173 20 31 shown show VERB fbem-26173 20 32 that that SCONJ fbem-26173 20 33 the the DET fbem-26173 20 34 arima arima PROPN fbem-26173 20 35 model model NOUN fbem-26173 20 36 has have VERB fbem-26173 20 37 strong strong ADJ fbem-26173 20 38 short short ADJ fbem-26173 20 39 - - PUNCT fbem-26173 20 40 term term NOUN fbem-26173 20 41 prediction prediction NOUN fbem-26173 20 42 ability ability NOUN fbem-26173 20 43 . . PUNCT fbem-26173 21 1 traditional traditional ADJ fbem-26173 21 2 machine machine NOUN fbem-26173 21 3 learning learning NOUN fbem-26173 21 4 methods method NOUN fbem-26173 21 5 use use VERB fbem-26173 21 6 different different ADJ fbem-26173 21 7 feature feature NOUN fbem-26173 21 8 extraction extraction NOUN fbem-26173 21 9 techniques technique NOUN fbem-26173 21 10 to to PART fbem-26173 21 11 select select VERB fbem-26173 21 12 the the DET fbem-26173 21 13 most most ADV fbem-26173 21 14 useful useful ADJ fbem-26173 21 15 feature feature NOUN fbem-26173 21 16 information information NOUN fbem-26173 21 17 to to PART fbem-26173 21 18 improve improve VERB fbem-26173 21 19 stock stock NOUN fbem-26173 21 20 forecasting forecasting NOUN fbem-26173 21 21 performance performance NOUN fbem-26173 21 22 . . PUNCT fbem-26173 22 1 in in ADP fbem-26173 22 2 order order NOUN fbem-26173 22 3 to to PART fbem-26173 22 4 overcome overcome VERB fbem-26173 22 5 the the DET fbem-26173 22 6 situation situation NOUN fbem-26173 22 7 that that SCONJ fbem-26173 22 8 traditional traditional ADJ fbem-26173 22 9 time time NOUN fbem-26173 22 10 series series NOUN fbem-26173 22 11 models model NOUN fbem-26173 22 12 are be AUX fbem-26173 22 13 only only ADV fbem-26173 22 14 limited limit VERB fbem-26173 22 15 to to ADP fbem-26173 22 16 linear linear ADJ fbem-26173 22 17 representation representation NOUN fbem-26173 22 18 , , PUNCT fbem-26173 22 19 wen wen PROPN fbem-26173 22 20 ( ( PUNCT fbem-26173 22 21 2014 2014 NUM fbem-26173 22 22 ) ) PUNCT fbem-26173 23 1 [ [ X fbem-26173 23 2 3 3 X fbem-26173 23 3 ] ] PUNCT fbem-26173 23 4 used use VERB fbem-26173 23 5 singular singular PROPN fbem-26173 23 6 spectrum spectrum NOUN fbem-26173 23 7 analysis analysis NOUN fbem-26173 23 8 ( ( PUNCT fbem-26173 23 9 ssa ssa NOUN fbem-26173 23 10 ) ) PUNCT fbem-26173 23 11 to to PART fbem-26173 23 12 decompose decompose VERB fbem-26173 23 13 stock stock NOUN fbem-26173 23 14 prices price NOUN fbem-26173 23 15 into into ADP fbem-26173 23 16 trends trend NOUN fbem-26173 23 17 , , PUNCT fbem-26173 23 18 market market NOUN fbem-26173 23 19 fluctuations fluctuation NOUN fbem-26173 23 20 and and CCONJ fbem-26173 23 21 noise noise NOUN fbem-26173 23 22 with with ADP fbem-26173 23 23 different different ADJ fbem-26173 23 24 economic economic ADJ fbem-26173 23 25 characteristics characteristic NOUN fbem-26173 23 26 in in ADP fbem-26173 23 27 different different ADJ fbem-26173 23 28 time time NOUN fbem-26173 23 29 periods period NOUN fbem-26173 23 30 and and CCONJ fbem-26173 23 31 input input NOUN fbem-26173 23 32 these these DET fbem-26173 23 33 features feature NOUN fbem-26173 23 34 into into ADP fbem-26173 23 35 support support NOUN fbem-26173 23 36 vector vector NOUN fbem-26173 23 37 machine machine NOUN fbem-26173 23 38 model model NOUN fbem-26173 23 39 for for ADP fbem-26173 23 40 stock stock NOUN fbem-26173 23 41 price price NOUN fbem-26173 23 42 prediction prediction NOUN fbem-26173 23 43 , , PUNCT fbem-26173 23 44 ssa ssa NOUN fbem-26173 23 45 - - PUNCT fbem-26173 23 46 svm svm NOUN fbem-26173 23 47 model model NOUN fbem-26173 23 48 has have VERB fbem-26173 23 49 good good ADJ fbem-26173 23 50 prediction prediction NOUN fbem-26173 23 51 results result NOUN fbem-26173 23 52 . . PUNCT fbem-26173 24 1 salim salim NOUN fbem-26173 24 2 ( ( PUNCT fbem-26173 24 3 2018 2018 NUM fbem-26173 24 4 ) ) PUNCT fbem-26173 25 1 [ [ X fbem-26173 25 2 4 4 X fbem-26173 25 3 ] ] PUNCT fbem-26173 25 4 used use VERB fbem-26173 25 5 a a DET fbem-26173 25 6 model model NOUN fbem-26173 25 7 combining combine VERB fbem-26173 25 8 singular singular NOUN fbem-26173 25 9 spectrum spectrum NOUN fbem-26173 25 10 analysis analysis NOUN fbem-26173 25 11 ( ( PUNCT fbem-26173 25 12 ssa ssa NOUN fbem-26173 25 13 ) ) PUNCT fbem-26173 25 14 , , PUNCT fbem-26173 25 15 support support VERB fbem-26173 25 16 vector vector NOUN fbem-26173 25 17 regression regression NOUN fbem-26173 25 18 ( ( PUNCT fbem-26173 25 19 svr svr PROPN fbem-26173 25 20 ) ) PUNCT fbem-26173 25 21 with with ADP fbem-26173 25 22 particle particle NOUN fbem-26173 25 23 swarm swarm NOUN fbem-26173 25 24 optimization optimization NOUN fbem-26173 25 25 ( ( PUNCT fbem-26173 25 26 pso pso NOUN fbem-26173 25 27 ) ) PUNCT fbem-26173 25 28 , , PUNCT fbem-26173 25 29 ssa ssa NOUN fbem-26173 25 30 decomposed decompose VERB fbem-26173 25 31 the the DET fbem-26173 25 32 stock stock NOUN fbem-26173 25 33 price price NOUN fbem-26173 25 34 time time NOUN fbem-26173 25 35 series series NOUN fbem-26173 25 36 into into ADP fbem-26173 25 37 a a DET fbem-26173 25 38 few few ADJ fbem-26173 25 39 independent independent ADJ fbem-26173 25 40 components component NOUN fbem-26173 25 41 to to PART fbem-26173 25 42 be be AUX fbem-26173 25 43 used use VERB fbem-26173 25 44 as as ADP fbem-26173 25 45 predictors predictor NOUN fbem-26173 25 46 , , PUNCT fbem-26173 25 47 svr svr PROPN fbem-26173 25 48 was be AUX fbem-26173 25 49 applied apply VERB fbem-26173 25 50 to to ADP fbem-26173 25 51 the the DET fbem-26173 25 52 forecasting forecasting NOUN fbem-26173 25 53 task task NOUN fbem-26173 25 54 and and CCONJ fbem-26173 25 55 pso pso NOUN fbem-26173 25 56 was be AUX fbem-26173 25 57 used use VERB fbem-26173 25 58 to to PART fbem-26173 25 59 optimize optimize VERB fbem-26173 25 60 the the DET fbem-26173 25 61 parameters parameter NOUN fbem-26173 25 62 of of ADP fbem-26173 25 63 the the DET fbem-26173 25 64 svr svr PROPN fbem-26173 25 65 , , PUNCT fbem-26173 25 66 ssa ssa NOUN fbem-26173 25 67 - - PUNCT fbem-26173 25 68 psosvr psosvr PROPN fbem-26173 25 69 showed show VERB fbem-26173 25 70 a a DET fbem-26173 25 71 significant significant ADJ fbem-26173 25 72 effect effect NOUN fbem-26173 25 73 in in ADP fbem-26173 25 74 noisy noisy ADJ fbem-26173 25 75 financial financial ADJ fbem-26173 25 76 time time NOUN fbem-26173 25 77 series series NOUN fbem-26173 25 78 analysis analysis NOUN fbem-26173 25 79 and and CCONJ fbem-26173 25 80 forecasting forecasting NOUN fbem-26173 25 81 shows show VERB fbem-26173 25 82 obvious obvious ADJ fbem-26173 25 83 effects effect NOUN fbem-26173 25 84 . . PUNCT fbem-26173 26 1 and and CCONJ fbem-26173 26 2 deep deep ADJ fbem-26173 26 3 learning learning NOUN fbem-26173 26 4 models model NOUN fbem-26173 26 5 can can AUX fbem-26173 26 6 better well ADV fbem-26173 26 7 model model VERB fbem-26173 26 8 nonlinear nonlinear ADJ fbem-26173 26 9 relationships relationship NOUN fbem-26173 26 10 , , PUNCT fbem-26173 26 11 automatically automatically ADV fbem-26173 26 12 extract extract VERB fbem-26173 26 13 features feature NOUN fbem-26173 26 14 , , PUNCT fbem-26173 26 15 process process NOUN fbem-26173 26 16 sequence sequence NOUN fbem-26173 26 17 data datum NOUN fbem-26173 26 18 , , PUNCT fbem-26173 26 19 utilize utilize VERB fbem-26173 26 20 large large ADJ fbem-26173 26 21 - - PUNCT fbem-26173 26 22 scale scale NOUN fbem-26173 26 23 data datum NOUN fbem-26173 26 24 and and CCONJ fbem-26173 26 25 adapt adapt VERB fbem-26173 26 26 to to ADP fbem-26173 26 27 market market NOUN fbem-26173 26 28 changes change NOUN fbem-26173 26 29 , , PUNCT fbem-26173 26 30 thus thus ADV fbem-26173 26 31 improving improve VERB fbem-26173 26 32 the the DET fbem-26173 26 33 forecasting forecasting NOUN fbem-26173 26 34 accuracy accuracy NOUN fbem-26173 26 35 and and CCONJ fbem-26173 26 36 stability stability NOUN fbem-26173 26 37 of of ADP fbem-26173 26 38 stock stock NOUN fbem-26173 26 39 movements movement NOUN fbem-26173 26 40 . . PUNCT fbem-26173 27 1 abdul abdul PROPN fbem-26173 27 2 ( ( PUNCT fbem-26173 27 3 2023 2023 NUM fbem-26173 27 4 ) ) PUNCT fbem-26173 28 1 [ [ X fbem-26173 28 2 5 5 NUM fbem-26173 28 3 ] ] PUNCT fbem-26173 28 4 proposed propose VERB fbem-26173 28 5 an an DET fbem-26173 28 6 optimization optimization NOUN fbem-26173 28 7 method method NOUN fbem-26173 28 8 for for ADP fbem-26173 28 9 stock stock NOUN fbem-26173 28 10 price price NOUN fbem-26173 28 11 forecasting forecasting NOUN fbem-26173 28 12 based base VERB fbem-26173 28 13 on on ADP fbem-26173 28 14 a a DET fbem-26173 28 15 multilayer multilayer ADJ fbem-26173 28 16 sequential sequential ADJ fbem-26173 28 17 longand longand PROPN fbem-26173 28 18 short short ADJ fbem-26173 28 19 - - PUNCT fbem-26173 28 20 term term NOUN fbem-26173 28 21 memory memory NOUN fbem-26173 28 22 ( ( PUNCT fbem-26173 28 23 mls mls PROPN fbem-26173 28 24 - - PUNCT fbem-26173 28 25 lstm lstm ADJ fbem-26173 28 26 ) ) PUNCT fbem-26173 28 27 model model NOUN fbem-26173 28 28 . . PUNCT fbem-26173 29 1 the the DET fbem-26173 29 2 mls mls PROPN fbem-26173 29 3 - - PUNCT fbem-26173 29 4 lstm lstm PROPN fbem-26173 29 5 algorithm algorithm PROPN fbem-26173 29 6 uses use VERB fbem-26173 29 7 normalized normalize VERB fbem-26173 29 8 time time NOUN fbem-26173 29 9 series series PROPN fbem-26173 29 10 data datum NOUN fbem-26173 29 11 divided divide VERB fbem-26173 29 12 into into ADP fbem-26173 29 13 time time NOUN fbem-26173 29 14 steps step NOUN fbem-26173 29 15 to to PART fbem-26173 29 16 determine determine VERB fbem-26173 29 17 the the DET fbem-26173 29 18 relationship relationship NOUN fbem-26173 29 19 between between ADP fbem-26173 29 20 past past ADJ fbem-26173 29 21 and and CCONJ fbem-26173 29 22 future future ADJ fbem-26173 29 23 values value NOUN fbem-26173 29 24 for for ADP fbem-26173 29 25 accurate accurate ADJ fbem-26173 29 26 prediction prediction NOUN fbem-26173 29 27 . . PUNCT fbem-26173 30 1 ren ren NOUN fbem-26173 30 2 et et PROPN fbem-26173 30 3 al al PROPN fbem-26173 30 4 . . PROPN fbem-26173 31 1 ( ( PUNCT fbem-26173 31 2 2023 2023 NUM fbem-26173 31 3 ) ) PUNCT fbem-26173 32 1 [ [ X fbem-26173 32 2 6 6 NUM fbem-26173 32 3 ] ] PUNCT fbem-26173 32 4 proposed propose VERB fbem-26173 32 5 a a DET fbem-26173 32 6 novel novel ADJ fbem-26173 32 7 hybrid hybrid NOUN fbem-26173 32 8 model model NOUN fbem-26173 32 9 for for ADP fbem-26173 32 10 stock stock NOUN fbem-26173 32 11 price price NOUN fbem-26173 32 12 prediction prediction NOUN fbem-26173 32 13 that that PRON fbem-26173 32 14 combines combine VERB fbem-26173 32 15 wavelet wavelet NOUN fbem-26173 32 16 transforms transform VERB fbem-26173 32 17 , , PUNCT fbem-26173 32 18 combining combine VERB fbem-26173 32 19 encoder encoder NOUN fbem-26173 32 20 forests forest NOUN fbem-26173 32 21 with with ADP fbem-26173 32 22 informer informer NOUN fbem-26173 32 23 . . PUNCT fbem-26173 33 1 explored explore VERB fbem-26173 33 2 the the DET fbem-26173 33 3 decomposition decomposition NOUN fbem-26173 33 4 - - PUNCT fbem-26173 33 5 prediction prediction NOUN fbem-26173 33 6 - - PUNCT fbem-26173 33 7 reconstruction reconstruction NOUN fbem-26173 33 8 methodology methodology NOUN fbem-26173 33 9 with with ADP fbem-26173 33 10 machine machine NOUN fbem-26173 33 11 learning learning NOUN fbem-26173 33 12 models model VERB fbem-26173 33 13 the the DET fbem-26173 33 14 impact impact NOUN fbem-26173 33 15 of of ADP fbem-26173 33 16 fusion fusion NOUN fbem-26173 33 17 is be AUX fbem-26173 33 18 aimed aim VERB fbem-26173 33 19 at at ADP fbem-26173 33 20 enhancing enhance VERB fbem-26173 33 21 the the DET fbem-26173 33 22 predictive predictive ADJ fbem-26173 33 23 power power NOUN fbem-26173 33 24 of of ADP fbem-26173 33 25 the the DET fbem-26173 33 26 model model NOUN fbem-26173 33 27 . . PUNCT fbem-26173 34 1 wen wen PROPN fbem-26173 34 2 et et PROPN fbem-26173 34 3 al al PROPN fbem-26173 34 4 . . PROPN fbem-26173 35 1 ( ( PUNCT fbem-26173 35 2 2024 2024 NUM fbem-26173 35 3 ) ) PUNCT fbem-26173 36 1 [ [ X fbem-26173 36 2 7 7 X fbem-26173 36 3 ] ] PUNCT fbem-26173 36 4 developed develop VERB fbem-26173 36 5 a a DET fbem-26173 36 6 hybrid hybrid ADJ fbem-26173 36 7 model model NOUN fbem-26173 36 8 called call VERB fbem-26173 36 9 mvl mvl NOUN fbem-26173 36 10 - - PUNCT fbem-26173 36 11 svm svm PROPN fbem-26173 36 12 for for ADP fbem-26173 36 13 predicting predict VERB fbem-26173 36 14 stock stock NOUN fbem-26173 36 15 price price NOUN fbem-26173 36 16 trends trend NOUN fbem-26173 36 17 by by ADP fbem-26173 36 18 combining combine VERB fbem-26173 36 19 multi multi ADJ fbem-26173 36 20 - - ADJ fbem-26173 36 21 view view ADJ fbem-26173 36 22 learning learning NOUN fbem-26173 36 23 with with ADP fbem-26173 36 24 support support NOUN fbem-26173 36 25 vector vector NOUN fbem-26173 36 26 machines machine NOUN fbem-26173 36 27 ( ( PUNCT fbem-26173 36 28 svm svm PROPN fbem-26173 36 29 ) ) PUNCT fbem-26173 36 30 . . PUNCT fbem-26173 37 1 77 77 NUM fbem-26173 37 2 the the DET fbem-26173 37 3 model model NOUN fbem-26173 37 4 reduces reduce VERB fbem-26173 37 5 the the DET fbem-26173 37 6 loss loss NOUN fbem-26173 37 7 of of ADP fbem-26173 37 8 information information NOUN fbem-26173 37 9 by by ADP fbem-26173 37 10 simply simply ADV fbem-26173 37 11 inputting inputte VERB fbem-26173 37 12 heterogeneous heterogeneous ADJ fbem-26173 37 13 multiview multiview NOUN fbem-26173 37 14 data datum NOUN fbem-26173 37 15 at at ADP fbem-26173 37 16 the the DET fbem-26173 37 17 same same ADJ fbem-26173 37 18 time time NOUN fbem-26173 37 19 . . PUNCT fbem-26173 38 1 the the DET fbem-26173 38 2 model model NOUN fbem-26173 38 3 is be AUX fbem-26173 38 4 used use VERB fbem-26173 38 5 for for ADP fbem-26173 38 6 the the DET fbem-26173 38 7 prediction prediction NOUN fbem-26173 38 8 of of ADP fbem-26173 38 9 stock stock NOUN fbem-26173 38 10 price price NOUN fbem-26173 38 11 trends trend NOUN fbem-26173 38 12 by by ADP fbem-26173 38 13 combining combine VERB fbem-26173 38 14 multi multi ADJ fbem-26173 38 15 - - ADJ fbem-26173 38 16 view view ADJ fbem-26173 38 17 learning learning NOUN fbem-26173 38 18 with with ADP fbem-26173 38 19 support support NOUN fbem-26173 38 20 vector vector NOUN fbem-26173 38 21 machines machine NOUN fbem-26173 38 22 . . PUNCT fbem-26173 39 1 in in ADP fbem-26173 39 2 summary summary NOUN fbem-26173 39 3 , , PUNCT fbem-26173 39 4 the the DET fbem-26173 39 5 stock stock NOUN fbem-26173 39 6 price price NOUN fbem-26173 39 7 prediction prediction NOUN fbem-26173 39 8 algorithm algorithm NOUN fbem-26173 39 9 has have VERB fbem-26173 39 10 high high ADJ fbem-26173 39 11 requirements requirement NOUN fbem-26173 39 12 for for ADP fbem-26173 39 13 prediction prediction NOUN fbem-26173 39 14 accuracy accuracy NOUN fbem-26173 39 15 , , PUNCT fbem-26173 39 16 and and CCONJ fbem-26173 39 17 this this DET fbem-26173 39 18 paper paper NOUN fbem-26173 39 19 improves improve VERB fbem-26173 39 20 the the DET fbem-26173 39 21 technical technical ADJ fbem-26173 39 22 analysis analysis NOUN fbem-26173 39 23 method method NOUN fbem-26173 39 24 based base VERB fbem-26173 39 25 on on ADP fbem-26173 39 26 deep deep ADJ fbem-26173 39 27 learning learn VERB fbem-26173 39 28 lstm lstm ADJ fbem-26173 39 29 algorithm algorithm NOUN fbem-26173 39 30 . . PUNCT fbem-26173 40 1 in in ADP fbem-26173 40 2 this this DET fbem-26173 40 3 paper paper NOUN fbem-26173 40 4 , , PUNCT fbem-26173 40 5 we we PRON fbem-26173 40 6 take take VERB fbem-26173 40 7 csi csi ADJ fbem-26173 40 8 300 300 NUM fbem-26173 40 9 stocks stock NOUN fbem-26173 40 10 for for ADP fbem-26173 40 11 research research NOUN fbem-26173 40 12 , , PUNCT fbem-26173 40 13 use use VERB fbem-26173 40 14 random random ADJ fbem-26173 40 15 forest forest NOUN fbem-26173 40 16 for for ADP fbem-26173 40 17 factor factor NOUN fbem-26173 40 18 selection selection NOUN fbem-26173 40 19 , , PUNCT fbem-26173 40 20 use use VERB fbem-26173 40 21 xgboost xgboost NOUN fbem-26173 40 22 method method NOUN fbem-26173 40 23 to to PART fbem-26173 40 24 select select VERB fbem-26173 40 25 high high ADJ fbem-26173 40 26 - - PUNCT fbem-26173 40 27 quality quality NOUN fbem-26173 40 28 stocks stock NOUN fbem-26173 40 29 from from ADP fbem-26173 40 30 the the DET fbem-26173 40 31 stock stock NOUN fbem-26173 40 32 pool pool NOUN fbem-26173 40 33 , , PUNCT fbem-26173 40 34 and and CCONJ fbem-26173 40 35 then then ADV fbem-26173 40 36 construct construct VERB fbem-26173 40 37 cnn cnn PROPN fbem-26173 40 38 - - PUNCT fbem-26173 40 39 bilstm bilstm NOUN fbem-26173 40 40 - - PUNCT fbem-26173 40 41 attention attention NOUN fbem-26173 40 42 for for ADP fbem-26173 40 43 stock stock NOUN fbem-26173 40 44 price price NOUN fbem-26173 40 45 prediction prediction NOUN fbem-26173 40 46 of of ADP fbem-26173 40 47 the the DET fbem-26173 40 48 selected select VERB fbem-26173 40 49 high high ADJ fbem-26173 40 50 - - PUNCT fbem-26173 40 51 quality quality NOUN fbem-26173 40 52 stocks stock NOUN fbem-26173 40 53 , , PUNCT fbem-26173 40 54 compared compare VERB fbem-26173 40 55 with with ADP fbem-26173 40 56 cnn cnn PROPN fbem-26173 40 57 , , PUNCT fbem-26173 40 58 lstm lstm NOUN fbem-26173 40 59 and and CCONJ fbem-26173 40 60 other other ADJ fbem-26173 40 61 separate separate ADJ fbem-26173 40 62 models model NOUN fbem-26173 40 63 , , PUNCT fbem-26173 40 64 combining combine VERB fbem-26173 40 65 the the DET fbem-26173 40 66 two two NUM fbem-26173 40 67 and and CCONJ fbem-26173 40 68 adding add VERB fbem-26173 40 69 attention attention NOUN fbem-26173 40 70 mechanism mechanism NOUN fbem-26173 40 71 has have VERB fbem-26173 40 72 better well ADJ fbem-26173 40 73 performance performance NOUN fbem-26173 40 74 , , PUNCT fbem-26173 40 75 with with ADP fbem-26173 40 76 better well ADJ fbem-26173 40 77 prediction prediction NOUN fbem-26173 40 78 accuracy accuracy NOUN fbem-26173 40 79 , , PUNCT fbem-26173 40 80 then then ADV fbem-26173 40 81 the the DET fbem-26173 40 82 trading trading NOUN fbem-26173 40 83 strategy strategy NOUN fbem-26173 40 84 is be AUX fbem-26173 40 85 formulated formulate VERB fbem-26173 40 86 by by ADP fbem-26173 40 87 macd macd NOUN fbem-26173 40 88 trading trading NOUN fbem-26173 40 89 strategy strategy NOUN fbem-26173 40 90 , , PUNCT fbem-26173 40 91 and and CCONJ fbem-26173 40 92 finally finally ADV fbem-26173 40 93 backtesting backtesting NOUN fbem-26173 40 94 is be AUX fbem-26173 40 95 performed perform VERB fbem-26173 40 96 based base VERB fbem-26173 40 97 on on ADP fbem-26173 40 98 the the DET fbem-26173 40 99 trading trading NOUN fbem-26173 40 100 strategy strategy NOUN fbem-26173 40 101 to to PART fbem-26173 40 102 check check VERB fbem-26173 40 103 the the DET fbem-26173 40 104 accuracy accuracy NOUN fbem-26173 40 105 and and CCONJ fbem-26173 40 106 effectiveness effectiveness NOUN fbem-26173 40 107 of of ADP fbem-26173 40 108 the the DET fbem-26173 40 109 strategy strategy NOUN fbem-26173 40 110 . . PUNCT fbem-26173 41 1 the the DET fbem-26173 41 2 quantitative quantitative ADJ fbem-26173 41 3 trading trading NOUN fbem-26173 41 4 strategy strategy NOUN fbem-26173 41 5 using use VERB fbem-26173 41 6 this this DET fbem-26173 41 7 deep deep ADJ fbem-26173 41 8 learning learning NOUN fbem-26173 41 9 model model NOUN fbem-26173 41 10 can can AUX fbem-26173 41 11 improve improve VERB fbem-26173 41 12 the the DET fbem-26173 41 13 stability stability NOUN fbem-26173 41 14 and and CCONJ fbem-26173 41 15 accuracy accuracy NOUN fbem-26173 41 16 of of ADP fbem-26173 41 17 stock stock NOUN fbem-26173 41 18 return return NOUN fbem-26173 41 19 prediction prediction NOUN fbem-26173 41 20 , , PUNCT fbem-26173 41 21 which which PRON fbem-26173 41 22 is be AUX fbem-26173 41 23 of of ADP fbem-26173 41 24 some some DET fbem-26173 41 25 significance significance NOUN fbem-26173 41 26 in in ADP fbem-26173 41 27 maintaining maintain VERB fbem-26173 41 28 market market NOUN fbem-26173 41 29 stability stability NOUN fbem-26173 41 30 and and CCONJ fbem-26173 41 31 strengthening strengthen VERB fbem-26173 41 32 the the DET fbem-26173 41 33 adjustment adjustment NOUN fbem-26173 41 34 of of ADP fbem-26173 41 35 national national ADJ fbem-26173 41 36 economic economic ADJ fbem-26173 41 37 policies policy NOUN fbem-26173 41 38 , , PUNCT fbem-26173 41 39 and and CCONJ fbem-26173 41 40 helps help VERB fbem-26173 41 41 investors investor NOUN fbem-26173 41 42 to to PART fbem-26173 41 43 make make VERB fbem-26173 41 44 faster fast ADJ fbem-26173 41 45 decisions decision NOUN fbem-26173 41 46 and and CCONJ fbem-26173 41 47 reduce reduce VERB fbem-26173 41 48 investment investment NOUN fbem-26173 41 49 risks risk NOUN fbem-26173 41 50 when when SCONJ fbem-26173 41 51 making make VERB fbem-26173 41 52 investments investment NOUN fbem-26173 41 53 . . PUNCT fbem-26173 42 1 2 2 X fbem-26173 42 2 . . X fbem-26173 42 3 model model NOUN fbem-26173 42 4 building building PROPN fbem-26173 42 5 and and CCONJ fbem-26173 42 6 trading trading NOUN fbem-26173 42 7 strategies strategy NOUN fbem-26173 42 8 2.1 2.1 NUM fbem-26173 42 9 . . PUNCT fbem-26173 42 10 random random ADJ fbem-26173 42 11 forest forest NOUN fbem-26173 42 12 and and CCONJ fbem-26173 42 13 xgboost xgboost NOUN fbem-26173 42 14 models model NOUN fbem-26173 42 15 random random ADJ fbem-26173 42 16 forest forest NOUN fbem-26173 42 17 is be AUX fbem-26173 42 18 an an DET fbem-26173 42 19 integrated integrate VERB fbem-26173 42 20 learning learning NOUN fbem-26173 42 21 algorithm algorithm NOUN fbem-26173 42 22 using use VERB fbem-26173 42 23 decision decision NOUN fbem-26173 42 24 trees tree NOUN fbem-26173 42 25 as as ADP fbem-26173 42 26 estimators estimator NOUN fbem-26173 42 27 , , PUNCT fbem-26173 42 28 combining combine VERB fbem-26173 42 29 multiple multiple ADJ fbem-26173 42 30 decision decision NOUN fbem-26173 42 31 trees tree NOUN fbem-26173 42 32 together together ADV fbem-26173 42 33 , , PUNCT fbem-26173 42 34 each each DET fbem-26173 42 35 time time NOUN fbem-26173 42 36 the the DET fbem-26173 42 37 dataset dataset NOUN fbem-26173 42 38 is be AUX fbem-26173 42 39 randomly randomly ADV fbem-26173 42 40 have have AUX fbem-26173 42 41 put put VERB fbem-26173 42 42 back back ADV fbem-26173 42 43 selected select VERB fbem-26173 42 44 , , PUNCT fbem-26173 42 45 while while SCONJ fbem-26173 42 46 randomly randomly ADV fbem-26173 42 47 selecting select VERB fbem-26173 42 48 some some PRON fbem-26173 42 49 of of ADP fbem-26173 42 50 the the DET fbem-26173 42 51 features feature NOUN fbem-26173 42 52 as as ADP fbem-26173 42 53 inputs input NOUN fbem-26173 42 54 [ [ X fbem-26173 42 55 8 8 NUM fbem-26173 42 56 ] ] PUNCT fbem-26173 42 57 . . PUNCT fbem-26173 43 1 the the DET fbem-26173 43 2 random random ADJ fbem-26173 43 3 forest forest NOUN fbem-26173 43 4 model model NOUN fbem-26173 43 5 will will AUX fbem-26173 43 6 construct construct VERB fbem-26173 43 7 n n DET fbem-26173 43 8 different different ADJ fbem-26173 43 9 sample sample NOUN fbem-26173 43 10 datasets dataset NOUN fbem-26173 43 11 by by ADP fbem-26173 43 12 randomly randomly ADV fbem-26173 43 13 sampling sample VERB fbem-26173 43 14 in in ADP fbem-26173 43 15 the the DET fbem-26173 43 16 original original ADJ fbem-26173 43 17 stock stock NOUN fbem-26173 43 18 dataset dataset NOUN fbem-26173 43 19 , , PUNCT fbem-26173 43 20 and and CCONJ fbem-26173 43 21 then then ADV fbem-26173 43 22 construct construct VERB fbem-26173 43 23 n n DET fbem-26173 43 24 different different ADJ fbem-26173 43 25 decision decision NOUN fbem-26173 43 26 tree tree NOUN fbem-26173 43 27 models model NOUN fbem-26173 43 28 based base VERB fbem-26173 43 29 on on ADP fbem-26173 43 30 these these DET fbem-26173 43 31 datasets dataset NOUN fbem-26173 43 32 , , PUNCT fbem-26173 43 33 and and CCONJ fbem-26173 43 34 finally finally ADV fbem-26173 43 35 the the DET fbem-26173 43 36 final final ADJ fbem-26173 43 37 result result NOUN fbem-26173 43 38 is be AUX fbem-26173 43 39 obtained obtain VERB fbem-26173 43 40 by by ADP fbem-26173 43 41 averaging average VERB fbem-26173 43 42 these these DET fbem-26173 43 43 decision decision NOUN fbem-26173 43 44 tree tree NOUN fbem-26173 43 45 models model NOUN fbem-26173 43 46 [ [ X fbem-26173 43 47 9 9 NUM fbem-26173 43 48 ] ] PUNCT fbem-26173 43 49 . . PUNCT fbem-26173 44 1 xgboost xgboost X fbem-26173 44 2 iteratively iteratively ADV fbem-26173 44 3 trains train VERB fbem-26173 44 4 multiple multiple ADJ fbem-26173 44 5 decision decision NOUN fbem-26173 44 6 trees tree NOUN fbem-26173 44 7 by by ADP fbem-26173 44 8 gradient gradient ADJ fbem-26173 44 9 boosting boost VERB fbem-26173 44 10 algorithm algorithm NOUN fbem-26173 44 11 and and CCONJ fbem-26173 44 12 combines combine VERB fbem-26173 44 13 them they PRON fbem-26173 44 14 to to PART fbem-26173 44 15 form form VERB fbem-26173 44 16 a a DET fbem-26173 44 17 powerful powerful ADJ fbem-26173 44 18 integrated integrate VERB fbem-26173 44 19 model model NOUN fbem-26173 44 20 [ [ X fbem-26173 44 21 10 10 NUM fbem-26173 44 22 ] ] PUNCT fbem-26173 44 23 . . PUNCT fbem-26173 45 1 it it PRON fbem-26173 45 2 features feature VERB fbem-26173 45 3 regularization regularization NOUN fbem-26173 45 4 , , PUNCT fbem-26173 45 5 feature feature NOUN fbem-26173 45 6 importance importance NOUN fbem-26173 45 7 evaluation evaluation NOUN fbem-26173 45 8 , , PUNCT fbem-26173 45 9 learning learn VERB fbem-26173 45 10 rate rate NOUN fbem-26173 45 11 control control NOUN fbem-26173 45 12 , , PUNCT fbem-26173 45 13 and and CCONJ fbem-26173 45 14 early early ADJ fbem-26173 45 15 stopping stopping NOUN fbem-26173 45 16 strategy strategy NOUN fbem-26173 45 17 , , PUNCT fbem-26173 45 18 which which PRON fbem-26173 45 19 can can AUX fbem-26173 45 20 effectively effectively ADV fbem-26173 45 21 deal deal VERB fbem-26173 45 22 with with ADP fbem-26173 45 23 complex complex ADJ fbem-26173 45 24 nonlinear nonlinear ADJ fbem-26173 45 25 relationships relationship NOUN fbem-26173 45 26 and and CCONJ fbem-26173 45 27 high high ADJ fbem-26173 45 28 - - PUNCT fbem-26173 45 29 dimensional dimensional ADJ fbem-26173 45 30 features feature NOUN fbem-26173 45 31 , , PUNCT fbem-26173 45 32 as as ADV fbem-26173 45 33 well well ADV fbem-26173 45 34 as as ADP fbem-26173 45 35 high high ADJ fbem-26173 45 36 performance performance NOUN fbem-26173 45 37 and and CCONJ fbem-26173 45 38 scalability scalability NOUN fbem-26173 45 39 [ [ X fbem-26173 45 40 11 11 NUM fbem-26173 45 41 ] ] PUNCT fbem-26173 45 42 . . PUNCT fbem-26173 46 1 it it PRON fbem-26173 46 2 is be AUX fbem-26173 46 3 known know VERB fbem-26173 46 4 that that SCONJ fbem-26173 46 5 the the DET fbem-26173 46 6 training training NOUN fbem-26173 46 7 dataset dataset VERB fbem-26173 46 8 𝑇 𝑇 PROPN fbem-26173 46 9 = = PRON fbem-26173 46 10 { { PUNCT fbem-26173 46 11 ( ( PUNCT fbem-26173 46 12 π‘₯1 π‘₯1 NOUN fbem-26173 46 13 , , PUNCT fbem-26173 46 14 𝑦1 𝑦1 NOUN fbem-26173 46 15 ) ) PUNCT fbem-26173 46 16 , , PUNCT fbem-26173 46 17 ( ( PUNCT fbem-26173 46 18 π‘₯2 π‘₯2 NOUN fbem-26173 46 19 , , PUNCT fbem-26173 46 20 𝑦2 𝑦2 PROPN fbem-26173 46 21 ) ) PUNCT fbem-26173 46 22 , , PUNCT fbem-26173 46 23 β‹― β‹― PROPN fbem-26173 46 24 , , PUNCT fbem-26173 46 25 ( ( PUNCT fbem-26173 46 26 π‘₯𝑛 π‘₯𝑛 NOUN fbem-26173 46 27 , , PUNCT fbem-26173 46 28 𝑦𝑛 𝑦𝑛 NOUN fbem-26173 46 29 ) ) PUNCT fbem-26173 46 30 } } PUNCT fbem-26173 46 31 , , PUNCT fbem-26173 46 32 the the DET fbem-26173 46 33 loss loss NOUN fbem-26173 46 34 function function VERB fbem-26173 46 35 𝑙( 𝑙( PROPN fbem-26173 46 36 οΏ½ οΏ½ PROPN fbem-26173 46 37 Μ‚ Μ‚ VERB fbem-26173 46 38 οΏ½ οΏ½ NOUN fbem-26173 46 39 𝑖 𝑖 SYM fbem-26173 46 40 , , PUNCT fbem-26173 46 41 οΏ½ οΏ½ NOUN fbem-26173 46 42 Μ‚ Μ‚ NOUN fbem-26173 46 43 οΏ½ οΏ½ NOUN fbem-26173 46 44 𝑖 𝑖 NUM fbem-26173 46 45 ) ) PUNCT fbem-26173 46 46 , , PUNCT fbem-26173 46 47 the the DET fbem-26173 46 48 regularization regularization NOUN fbem-26173 46 49 term term NOUN fbem-26173 46 50 Ο‰(π‘“π‘˜ Ο‰(π‘“π‘˜ PROPN fbem-26173 46 51 ) ) PUNCT fbem-26173 46 52 , , PUNCT fbem-26173 46 53 then then ADV fbem-26173 46 54 the the DET fbem-26173 46 55 overall overall ADJ fbem-26173 46 56 objective objective ADJ fbem-26173 46 57 function function NOUN fbem-26173 46 58 can can AUX fbem-26173 46 59 be be AUX fbem-26173 46 60 written write VERB fbem-26173 46 61 as as ADP fbem-26173 46 62 : : PUNCT fbem-26173 46 63 𝐿(πœ™ 𝐿(πœ™ NUM fbem-26173 46 64 ) ) PUNCT fbem-26173 46 65 = = SYM fbem-26173 46 66 βˆ‘ βˆ‘ PROPN fbem-26173 46 67 ( ( PUNCT fbem-26173 46 68 𝑙(𝑦𝑖 𝑙(𝑦𝑖 NUM fbem-26173 46 69 , , PUNCT fbem-26173 46 70 οΏ½ οΏ½ PROPN fbem-26173 46 71 Μ‚ Μ‚ NOUN fbem-26173 46 72 οΏ½ οΏ½ NOUN fbem-26173 46 73 𝑖 𝑖 NUM fbem-26173 46 74 ) ) PUNCT fbem-26173 46 75 + + CCONJ fbem-26173 46 76 βˆ‘ βˆ‘ PROPN fbem-26173 46 77 Ο‰(π‘“π‘˜)π‘˜ Ο‰(π‘“π‘˜)π‘˜ ADJ fbem-26173 46 78 ) ) PUNCT fbem-26173 46 79 𝑑 𝑑 PROPN fbem-26173 46 80 ( ( PUNCT fbem-26173 46 81 1 1 NUM fbem-26173 46 82 ) ) PUNCT fbem-26173 46 83 where where SCONJ fbem-26173 46 84 𝐿(πœ™ 𝐿(πœ™ X fbem-26173 46 85 ) ) PUNCT fbem-26173 46 86 is be AUX fbem-26173 46 87 the the DET fbem-26173 46 88 expression expression NOUN fbem-26173 46 89 on on ADP fbem-26173 46 90 linear linear ADJ fbem-26173 46 91 space space NOUN fbem-26173 46 92 ; ; PUNCT fbem-26173 46 93 𝑖 𝑖 X fbem-26173 46 94 is be AUX fbem-26173 46 95 the the DET fbem-26173 46 96 𝑖 𝑖 PROPN fbem-26173 46 97 sample sample NOUN fbem-26173 46 98 , , PUNCT fbem-26173 46 99 π‘˜ π‘˜ PROPN fbem-26173 46 100 is be AUX fbem-26173 46 101 the the DET fbem-26173 46 102 π‘˜ π‘˜ PRON fbem-26173 46 103 tree tree NOUN fbem-26173 46 104 ; ; PUNCT fbem-26173 46 105 Λ†iy Λ†iy NOUN fbem-26173 46 106 is be AUX fbem-26173 46 107 the the DET fbem-26173 46 108 predicted predict VERB fbem-26173 46 109 value value NOUN fbem-26173 46 110 of of ADP fbem-26173 46 111 the the DET fbem-26173 46 112 𝑖 𝑖 PROPN fbem-26173 46 113 sample sample NOUN fbem-26173 46 114 π‘₯𝑖 π‘₯𝑖 X fbem-26173 46 115 , , PUNCT fbem-26173 46 116 οΏ½ οΏ½ PROPN fbem-26173 46 117 Μ‚ Μ‚ NOUN fbem-26173 46 118 οΏ½ οΏ½ NOUN fbem-26173 46 119 𝑖 𝑖 SYM fbem-26173 46 120 = = SYM fbem-26173 46 121 βˆ‘ βˆ‘ PROPN fbem-26173 46 122 π‘“π‘˜(π‘₯𝑖 π‘“π‘˜(π‘₯𝑖 ADJ fbem-26173 46 123 ) ) PUNCT fbem-26173 46 124 𝐾 𝐾 PROPN fbem-26173 46 125 π‘˜=1 π‘˜=1 PROPN fbem-26173 46 126 . . PUNCT fbem-26173 47 1 since since SCONJ fbem-26173 47 2 : : PUNCT fbem-26173 47 3 οΏ½ οΏ½ PROPN fbem-26173 47 4 Μ‚ Μ‚ VERB fbem-26173 47 5 οΏ½ οΏ½ NOUN fbem-26173 47 6 𝑖 𝑖 SYM fbem-26173 47 7 = = SYM fbem-26173 47 8 βˆ‘ βˆ‘ PROPN fbem-26173 47 9 π‘“π‘˜(π‘₯𝑖 π‘“π‘˜(π‘₯𝑖 ADJ fbem-26173 47 10 ) ) PUNCT fbem-26173 47 11 𝐾 𝐾 PROPN fbem-26173 47 12 π‘˜=1 π‘˜=1 NOUN fbem-26173 47 13 = = SYM fbem-26173 47 14 οΏ½ οΏ½ PROPN fbem-26173 47 15 Μ‚ Μ‚ VERB fbem-26173 47 16 οΏ½ οΏ½ PROPN fbem-26173 47 17 𝑖 𝑖 SYM fbem-26173 47 18 π‘‘βˆ’1 π‘‘βˆ’1 PROPN fbem-26173 47 19 + + CCONJ fbem-26173 47 20 𝑓𝑑(π‘₯𝑖 𝑓𝑑(π‘₯𝑖 PROPN fbem-26173 47 21 ) ) PUNCT fbem-26173 47 22 , , PUNCT fbem-26173 47 23 then then ADV fbem-26173 47 24 𝐿(πœ™ 𝐿(πœ™ X fbem-26173 47 25 ) ) PUNCT fbem-26173 47 26 is be AUX fbem-26173 47 27 transformed transform VERB fbem-26173 47 28 into into ADP fbem-26173 47 29 the the DET fbem-26173 47 30 following follow VERB fbem-26173 47 31 form form NOUN fbem-26173 47 32 : : PUNCT fbem-26173 47 33 𝐿𝑑 𝐿𝑑 PROPN fbem-26173 47 34 = = SYM fbem-26173 47 35 βˆ‘ βˆ‘ PUNCT fbem-26173 47 36 ( ( PUNCT fbem-26173 47 37 𝑦𝑖 𝑦𝑖 PROPN fbem-26173 47 38 , , PUNCT fbem-26173 47 39 οΏ½ οΏ½ PROPN fbem-26173 47 40 Μ‚ Μ‚ VERB fbem-26173 47 41 οΏ½ οΏ½ PROPN fbem-26173 47 42 𝑖 𝑖 SYM fbem-26173 47 43 π‘‘βˆ’1 π‘‘βˆ’1 PROPN fbem-26173 47 44 + + CCONJ fbem-26173 47 45 𝑓𝑑(π‘₯𝑖 𝑓𝑑(π‘₯𝑖 PROPN fbem-26173 47 46 ) ) PUNCT fbem-26173 47 47 ) ) PUNCT fbem-26173 48 1 + + CCONJ fbem-26173 48 2 βˆ‘ βˆ‘ PUNCT fbem-26173 48 3 Ο‰(π‘“π‘˜)π‘˜ Ο‰(π‘“π‘˜)π‘˜ PROPN fbem-26173 48 4 𝑛 𝑛 PRON fbem-26173 48 5 𝑖=1 𝑖=1 PROPN fbem-26173 48 6 ( ( PUNCT fbem-26173 48 7 2 2 X fbem-26173 48 8 ) ) PUNCT fbem-26173 48 9 definition definition NOUN fbem-26173 48 10 : : PUNCT fbem-26173 48 11 𝐺𝑗 𝐺𝑗 PROPN fbem-26173 48 12 = = SYM fbem-26173 48 13 βˆ‘ βˆ‘ PUNCT fbem-26173 48 14 π‘”π‘–π‘–βˆˆπΌπ‘— π‘”π‘–π‘–βˆˆπΌπ‘— NOUN fbem-26173 48 15 , , PUNCT fbem-26173 48 16 𝐻𝑗 𝐻𝑗 PROPN fbem-26173 48 17 = = SYM fbem-26173 48 18 βˆ‘ βˆ‘ PUNCT fbem-26173 48 19 β„Žπ‘–π‘–βˆˆπΌπ‘— β„Žπ‘–π‘–βˆˆπΌπ‘— NOUN fbem-26173 48 20 , , PUNCT fbem-26173 48 21 are be AUX fbem-26173 48 22 the the DET fbem-26173 48 23 first first ADJ fbem-26173 48 24 and and CCONJ fbem-26173 48 25 second second ADJ fbem-26173 48 26 order order NOUN fbem-26173 48 27 partial partial ADJ fbem-26173 48 28 derivative derivative ADJ fbem-26173 48 29 cumulative cumulative ADJ fbem-26173 48 30 sums sum NOUN fbem-26173 48 31 , , PUNCT fbem-26173 48 32 respectively respectively ADV fbem-26173 48 33 . . PUNCT fbem-26173 49 1 substituting substitute VERB fbem-26173 49 2 into into ADP fbem-26173 49 3 the the DET fbem-26173 49 4 objective objective ADJ fbem-26173 49 5 function function NOUN fbem-26173 49 6 yields yield NOUN fbem-26173 49 7 : : PUNCT fbem-26173 49 8 𝐿𝑑 𝐿𝑑 PROPN fbem-26173 49 9 = = PUNCT fbem-26173 49 10 βˆ‘ βˆ‘ PUNCT fbem-26173 50 1 [ [ X fbem-26173 50 2 πΊπ‘—π‘Šπ‘— πΊπ‘—π‘Šπ‘— NOUN fbem-26173 50 3 + + CCONJ fbem-26173 50 4 1/2(𝐻𝑗 1/2(𝐻𝑗 NUM fbem-26173 50 5 + + CCONJ fbem-26173 50 6 πœ†)π‘Šπ‘— πœ†)π‘Šπ‘— NUM fbem-26173 50 7 2]𝑇𝑗=1 2]𝑇𝑗=1 NUM fbem-26173 50 8 . . PUNCT fbem-26173 51 1 2.2 2.2 NUM fbem-26173 51 2 . . PUNCT fbem-26173 52 1 convolutional convolutional ADJ fbem-26173 52 2 neural neural ADJ fbem-26173 52 3 network network NOUN fbem-26173 52 4 convolutional convolutional ADJ fbem-26173 52 5 neural neural ADJ fbem-26173 52 6 network network NOUN fbem-26173 52 7 ( ( PUNCT fbem-26173 52 8 cnn cnn PROPN fbem-26173 52 9 ) ) PUNCT fbem-26173 52 10 is be AUX fbem-26173 52 11 a a DET fbem-26173 52 12 deep deep ADJ fbem-26173 52 13 learning learning NOUN fbem-26173 52 14 model model NOUN fbem-26173 52 15 that that PRON fbem-26173 52 16 is be AUX fbem-26173 52 17 widely widely ADV fbem-26173 52 18 used use VERB fbem-26173 52 19 in in ADP fbem-26173 52 20 image image NOUN fbem-26173 52 21 processing processing NOUN fbem-26173 52 22 [ [ X fbem-26173 52 23 12 12 NUM fbem-26173 52 24 ] ] PUNCT fbem-26173 52 25 . . PUNCT fbem-26173 53 1 however however ADV fbem-26173 53 2 , , PUNCT fbem-26173 53 3 it it PRON fbem-26173 53 4 can can AUX fbem-26173 53 5 also also ADV fbem-26173 53 6 efficiently efficiently ADV fbem-26173 53 7 process process VERB fbem-26173 53 8 sequence sequence NOUN fbem-26173 53 9 data datum NOUN fbem-26173 53 10 such such ADJ fbem-26173 53 11 as as ADP fbem-26173 53 12 stock stock NOUN fbem-26173 53 13 price price NOUN fbem-26173 53 14 time time NOUN fbem-26173 53 15 series series NOUN fbem-26173 53 16 [ [ X fbem-26173 53 17 13 13 NUM fbem-26173 53 18 ] ] PUNCT fbem-26173 53 19 . . PUNCT fbem-26173 54 1 in in ADP fbem-26173 54 2 cnn cnn PROPN fbem-26173 54 3 , , PUNCT fbem-26173 54 4 automatic automatic ADJ fbem-26173 54 5 extraction extraction NOUN fbem-26173 54 6 of of ADP fbem-26173 54 7 local local ADJ fbem-26173 54 8 features feature NOUN fbem-26173 54 9 from from ADP fbem-26173 54 10 input input NOUN fbem-26173 54 11 data datum NOUN fbem-26173 54 12 is be AUX fbem-26173 54 13 achieved achieve VERB fbem-26173 54 14 by by ADP fbem-26173 54 15 using use VERB fbem-26173 54 16 multiple multiple ADJ fbem-26173 54 17 convolutional convolutional ADJ fbem-26173 54 18 layers layer NOUN fbem-26173 54 19 with with ADP fbem-26173 54 20 pooling pool VERB fbem-26173 54 21 layers layer NOUN fbem-26173 54 22 . . PUNCT fbem-26173 55 1 in in ADP fbem-26173 55 2 this this DET fbem-26173 55 3 paper paper NOUN fbem-26173 55 4 , , PUNCT fbem-26173 55 5 the the DET fbem-26173 55 6 cnn cnn PROPN fbem-26173 55 7 model model NOUN fbem-26173 55 8 is be AUX fbem-26173 55 9 processed process VERB fbem-26173 55 10 with with ADP fbem-26173 55 11 a a DET fbem-26173 55 12 series series NOUN fbem-26173 55 13 of of ADP fbem-26173 55 14 convolutional convolutional ADJ fbem-26173 55 15 and and CCONJ fbem-26173 55 16 pooling pool VERB fbem-26173 55 17 layers layer NOUN fbem-26173 55 18 to to PART fbem-26173 55 19 capture capture VERB fbem-26173 55 20 the the DET fbem-26173 55 21 patterns pattern NOUN fbem-26173 55 22 and and CCONJ fbem-26173 55 23 trends trend NOUN fbem-26173 55 24 of of ADP fbem-26173 55 25 stock stock NOUN fbem-26173 55 26 prices price NOUN fbem-26173 55 27 in in ADP fbem-26173 55 28 different different ADJ fbem-26173 55 29 time time NOUN fbem-26173 55 30 scales scale NOUN fbem-26173 55 31 . . PUNCT fbem-26173 56 1 input input NOUN fbem-26173 56 2 layer layer NOUN fbem-26173 56 3 : : PUNCT fbem-26173 56 4 the the DET fbem-26173 56 5 input input NOUN fbem-26173 56 6 to to ADP fbem-26173 56 7 the the DET fbem-26173 56 8 model model NOUN fbem-26173 56 9 is be AUX fbem-26173 56 10 a a DET fbem-26173 56 11 stock stock NOUN fbem-26173 56 12 price price NOUN fbem-26173 56 13 time time NOUN fbem-26173 56 14 series series PROPN fbem-26173 56 15 data datum NOUN fbem-26173 56 16 containing contain VERB fbem-26173 56 17 feature feature NOUN fbem-26173 56 18 variables variable NOUN fbem-26173 56 19 , , PUNCT fbem-26173 56 20 where where SCONJ fbem-26173 56 21 each each DET fbem-26173 56 22 time time NOUN fbem-26173 56 23 step step NOUN fbem-26173 56 24 corresponds correspond VERB fbem-26173 56 25 to to ADP fbem-26173 56 26 a a DET fbem-26173 56 27 feature feature NOUN fbem-26173 56 28 vector vector NOUN fbem-26173 56 29 . . PUNCT fbem-26173 57 1 let let VERB fbem-26173 57 2 the the DET fbem-26173 57 3 input input NOUN fbem-26173 57 4 data datum NOUN fbem-26173 57 5 be be VERB fbem-26173 57 6 : : PUNCT fbem-26173 57 7 𝑋 𝑋 PROPN fbem-26173 57 8 = = PUNCT fbem-26173 58 1 [ [ X fbem-26173 58 2 π‘₯1 π‘₯1 NOUN fbem-26173 58 3 , , PUNCT fbem-26173 58 4 π‘₯2 π‘₯2 NOUN fbem-26173 58 5 , , PUNCT fbem-26173 58 6 β‹― β‹― NOUN fbem-26173 58 7 , , PUNCT fbem-26173 58 8 π‘₯𝑇 π‘₯𝑇 PROPN fbem-26173 58 9 ] ] PUNCT fbem-26173 58 10 , , PUNCT fbem-26173 58 11 where where SCONJ fbem-26173 58 12 π‘₯𝑑 π‘₯𝑑 ADV fbem-26173 58 13 represents represent VERB fbem-26173 58 14 the the DET fbem-26173 58 15 feature feature NOUN fbem-26173 58 16 vector vector NOUN fbem-26173 58 17 of of ADP fbem-26173 58 18 time time NOUN fbem-26173 58 19 step𝑑. step𝑑. NOUN fbem-26173 58 20 convolutional convolutional ADJ fbem-26173 58 21 layer layer NOUN fbem-26173 58 22 : : PUNCT fbem-26173 58 23 the the DET fbem-26173 58 24 convolutional convolutional ADJ fbem-26173 58 25 layer layer NOUN fbem-26173 58 26 contains contain VERB fbem-26173 58 27 several several ADJ fbem-26173 58 28 convolutional convolutional ADJ fbem-26173 58 29 kernels kernel NOUN fbem-26173 58 30 ( ( PUNCT fbem-26173 58 31 filters filter NOUN fbem-26173 58 32 ) ) PUNCT fbem-26173 58 33 , , PUNCT fbem-26173 58 34 each each PRON fbem-26173 58 35 of of ADP fbem-26173 58 36 which which PRON fbem-26173 58 37 can can AUX fbem-26173 58 38 be be AUX fbem-26173 58 39 regarded regard VERB fbem-26173 58 40 as as ADP fbem-26173 58 41 a a DET fbem-26173 58 42 feature feature NOUN fbem-26173 58 43 detector detector NOUN fbem-26173 58 44 , , PUNCT fbem-26173 58 45 which which PRON fbem-26173 58 46 utilizes utilize VERB fbem-26173 58 47 convolutional convolutional ADJ fbem-26173 58 48 operations operation NOUN fbem-26173 58 49 to to PART fbem-26173 58 50 extract extract VERB fbem-26173 58 51 local local ADJ fbem-26173 58 52 features feature NOUN fbem-26173 58 53 from from ADP fbem-26173 58 54 the the DET fbem-26173 58 55 input input NOUN fbem-26173 58 56 data datum NOUN fbem-26173 58 57 in in ADP fbem-26173 58 58 order order NOUN fbem-26173 58 59 to to PART fbem-26173 58 60 capture capture VERB fbem-26173 58 61 patterns pattern NOUN fbem-26173 58 62 and and CCONJ fbem-26173 58 63 trends trend NOUN fbem-26173 58 64 in in ADP fbem-26173 58 65 stock stock NOUN fbem-26173 58 66 prices price NOUN fbem-26173 58 67 . . PUNCT fbem-26173 59 1 𝑦𝑖 𝑦𝑖 NOUN fbem-26173 60 1 = = NOUN fbem-26173 60 2 𝑓(βˆ‘ 𝑓(βˆ‘ X fbem-26173 60 3 πœ”π‘—π‘₯𝑖+𝑗 πœ”π‘—π‘₯𝑖+𝑗 NOUN fbem-26173 60 4 + + CCONJ fbem-26173 60 5 𝑏 𝑏 PROPN fbem-26173 60 6 π‘˜βˆ’1 π‘˜βˆ’1 PROPN fbem-26173 60 7 𝑗=0 𝑗=0 PROPN fbem-26173 60 8 ) ) PUNCT fbem-26173 60 9 ( ( PUNCT fbem-26173 60 10 3 3 X fbem-26173 60 11 ) ) PUNCT fbem-26173 60 12 where where SCONJ fbem-26173 60 13 𝑦𝑖 𝑦𝑖 PROPN fbem-26173 60 14 is be AUX fbem-26173 60 15 the the DET fbem-26173 60 16 output output NOUN fbem-26173 60 17 of of ADP fbem-26173 60 18 the the DET fbem-26173 60 19 convolutional convolutional ADJ fbem-26173 60 20 layer layer NOUN fbem-26173 60 21 , , PUNCT fbem-26173 60 22 π‘₯𝑖+𝑗 π‘₯𝑖+𝑗 PROPN fbem-26173 60 23 is be AUX fbem-26173 60 24 the the DET fbem-26173 60 25 𝑖 𝑖 PROPN fbem-26173 60 26 + + NOUN fbem-26173 60 27 𝑗 𝑗 PRON fbem-26173 60 28 element element NOUN fbem-26173 60 29 of of ADP fbem-26173 60 30 the the DET fbem-26173 60 31 input input NOUN fbem-26173 60 32 sequence sequence NOUN fbem-26173 60 33 , , PUNCT fbem-26173 60 34 πœ”π‘— πœ”π‘— PROPN fbem-26173 60 35 is be AUX fbem-26173 60 36 the the DET fbem-26173 60 37 weight weight NOUN fbem-26173 60 38 of of ADP fbem-26173 60 39 the the DET fbem-26173 60 40 convolutional convolutional ADJ fbem-26173 60 41 kernel kernel NOUN fbem-26173 60 42 , , PUNCT fbem-26173 60 43 𝑏 𝑏 PROPN fbem-26173 60 44 is be AUX fbem-26173 60 45 the the DET fbem-26173 60 46 bias bias NOUN fbem-26173 60 47 , , PUNCT fbem-26173 60 48 and and CCONJ fbem-26173 60 49 𝑓(βˆ™ 𝑓(βˆ™ NOUN fbem-26173 60 50 ) ) PUNCT fbem-26173 60 51 is be AUX fbem-26173 60 52 the the DET fbem-26173 60 53 activation activation NOUN fbem-26173 60 54 function function NOUN fbem-26173 60 55 . . PUNCT fbem-26173 61 1 pooling pool VERB fbem-26173 61 2 layer layer NOUN fbem-26173 61 3 : : PUNCT fbem-26173 61 4 used use VERB fbem-26173 61 5 to to PART fbem-26173 61 6 reduce reduce VERB fbem-26173 61 7 the the DET fbem-26173 61 8 dimensionality dimensionality NOUN fbem-26173 61 9 of of ADP fbem-26173 61 10 the the DET fbem-26173 61 11 time time NOUN fbem-26173 61 12 series series NOUN fbem-26173 61 13 and and CCONJ fbem-26173 61 14 extract extract VERB fbem-26173 61 15 higher high ADJ fbem-26173 61 16 - - PUNCT fbem-26173 61 17 level level NOUN fbem-26173 61 18 features feature NOUN fbem-26173 61 19 . . PUNCT fbem-26173 62 1 common common ADJ fbem-26173 62 2 pooling pool VERB fbem-26173 62 3 operations operation NOUN fbem-26173 62 4 include include VERB fbem-26173 62 5 max max PROPN fbem-26173 62 6 pooling pooling NOUN fbem-26173 62 7 and and CCONJ fbem-26173 62 8 average average ADJ fbem-26173 62 9 pooling pooling NOUN fbem-26173 62 10 , , PUNCT fbem-26173 62 11 which which PRON fbem-26173 62 12 extract extract VERB fbem-26173 62 13 local local ADJ fbem-26173 62 14 maxima maxima NOUN fbem-26173 62 15 and and CCONJ fbem-26173 62 16 averages average NOUN fbem-26173 62 17 , , PUNCT fbem-26173 62 18 respectively respectively ADV fbem-26173 62 19 . . PUNCT fbem-26173 63 1 𝑦𝑖 𝑦𝑖 PROPN fbem-26173 63 2 = = SYM fbem-26173 63 3 max max PROPN fbem-26173 63 4 ( ( PUNCT fbem-26173 63 5 π‘₯π‘–βˆ’π‘  π‘₯π‘–βˆ’π‘  NOUN fbem-26173 63 6 , , PUNCT fbem-26173 63 7 π‘₯π‘–βˆ’π‘ +1 π‘₯π‘–βˆ’π‘ +1 PROPN fbem-26173 63 8 , , PUNCT fbem-26173 63 9 β‹― β‹― PROPN fbem-26173 63 10 , , PUNCT fbem-26173 63 11 π‘₯π‘–βˆ’π‘ +π‘ βˆ’1 π‘₯π‘–βˆ’π‘ +π‘ βˆ’1 NOUN fbem-26173 63 12 ) ) PUNCT fbem-26173 63 13 ( ( PUNCT fbem-26173 63 14 4 4 X fbem-26173 63 15 ) ) PUNCT fbem-26173 63 16 where where SCONJ fbem-26173 63 17 𝑦𝑖 𝑦𝑖 PROPN fbem-26173 63 18 is be AUX fbem-26173 63 19 the the DET fbem-26173 63 20 output output NOUN fbem-26173 63 21 of of ADP fbem-26173 63 22 the the DET fbem-26173 63 23 pooling pool VERB fbem-26173 63 24 layer layer NOUN fbem-26173 63 25 , , PUNCT fbem-26173 63 26 π‘₯π‘–βˆ’π‘  π‘₯π‘–βˆ’π‘  NOUN fbem-26173 63 27 to to ADP fbem-26173 63 28 π‘₯π‘–βˆ’π‘ +π‘ βˆ’1 π‘₯π‘–βˆ’π‘ +π‘ βˆ’1 PROPN fbem-26173 63 29 are be AUX fbem-26173 63 30 a a DET fbem-26173 63 31 set set NOUN fbem-26173 63 32 of of ADP fbem-26173 63 33 elements element NOUN fbem-26173 63 34 of of ADP fbem-26173 63 35 the the DET fbem-26173 63 36 input input NOUN fbem-26173 63 37 sequence sequence NOUN fbem-26173 63 38 , , PUNCT fbem-26173 63 39 and and CCONJ fbem-26173 63 40 𝑠 𝑠 PROPN fbem-26173 63 41 is be AUX fbem-26173 63 42 the the DET fbem-26173 63 43 size size NOUN fbem-26173 63 44 of of ADP fbem-26173 63 45 the the DET fbem-26173 63 46 pooling pooling NOUN fbem-26173 63 47 window window NOUN fbem-26173 63 48 . . PUNCT fbem-26173 64 1 2.3 2.3 NUM fbem-26173 64 2 . . PUNCT fbem-26173 65 1 shortand shortand NOUN fbem-26173 65 2 long long ADJ fbem-26173 65 3 - - PUNCT fbem-26173 65 4 term term NOUN fbem-26173 65 5 memory memory NOUN fbem-26173 65 6 neural neural ADJ fbem-26173 65 7 networks network NOUN fbem-26173 65 8 long long ADJ fbem-26173 65 9 short short ADJ fbem-26173 65 10 - - PUNCT fbem-26173 65 11 term term NOUN fbem-26173 65 12 memory memory NOUN fbem-26173 65 13 network network NOUN fbem-26173 65 14 ( ( PUNCT fbem-26173 65 15 lstm lstm PROPN fbem-26173 65 16 ) ) PUNCT fbem-26173 65 17 is be AUX fbem-26173 65 18 a a DET fbem-26173 65 19 recurrent recurrent ADJ fbem-26173 65 20 neural neural ADJ fbem-26173 65 21 network network NOUN fbem-26173 65 22 ( ( PUNCT fbem-26173 65 23 rnn rnn PROPN fbem-26173 65 24 ) ) PUNCT fbem-26173 65 25 variant variant NOUN fbem-26173 65 26 for for ADP fbem-26173 65 27 sequence sequence NOUN fbem-26173 65 28 data datum NOUN fbem-26173 65 29 processing processing NOUN fbem-26173 65 30 [ [ X fbem-26173 65 31 14 14 NUM fbem-26173 65 32 ] ] PUNCT fbem-26173 65 33 . . PUNCT fbem-26173 66 1 compared compare VERB fbem-26173 66 2 to to ADP fbem-26173 66 3 traditional traditional ADJ fbem-26173 66 4 rnns rnn NOUN fbem-26173 66 5 , , PUNCT fbem-26173 66 6 lstm lstm NOUN fbem-26173 66 7 can can AUX fbem-26173 66 8 effectively effectively ADV fbem-26173 66 9 overcome overcome VERB fbem-26173 66 10 the the DET fbem-26173 66 11 problems problem NOUN fbem-26173 66 12 of of ADP fbem-26173 66 13 gradient gradient ADJ fbem-26173 66 14 vanishing vanishing NOUN fbem-26173 66 15 and and CCONJ fbem-26173 66 16 gradient gradient ADJ fbem-26173 66 17 explosion explosion NOUN fbem-26173 66 18 in in ADP fbem-26173 66 19 long long ADJ fbem-26173 66 20 time time NOUN fbem-26173 66 21 sequences sequence NOUN fbem-26173 66 22 through through ADP fbem-26173 66 23 the the DET fbem-26173 66 24 gating gate VERB fbem-26173 66 25 mechanism mechanism NOUN fbem-26173 66 26 to to PART fbem-26173 66 27 better well ADV fbem-26173 66 28 capture capture VERB fbem-26173 66 29 long long ADJ fbem-26173 66 30 term term NOUN fbem-26173 66 31 correlations correlation NOUN fbem-26173 66 32 [ [ X fbem-26173 66 33 15 15 NUM fbem-26173 66 34 ] ] PUNCT fbem-26173 66 35 . . PUNCT fbem-26173 67 1 bilstm bilstm NOUN fbem-26173 67 2 , , PUNCT fbem-26173 67 3 on on ADP fbem-26173 67 4 the the DET fbem-26173 67 5 other other ADJ fbem-26173 67 6 hand hand NOUN fbem-26173 67 7 , , PUNCT fbem-26173 67 8 adds add VERB fbem-26173 67 9 backpropagation backpropagation NOUN fbem-26173 67 10 to to ADP fbem-26173 67 11 lstm lstm NOUN fbem-26173 67 12 and and CCONJ fbem-26173 67 13 is be AUX fbem-26173 67 14 able able ADJ fbem-26173 67 15 to to PART fbem-26173 67 16 utilise utilise VERB fbem-26173 67 17 past past ADP fbem-26173 67 18 and and CCONJ fbem-26173 67 19 future future ADJ fbem-26173 67 20 contextual contextual ADJ fbem-26173 67 21 information information NOUN fbem-26173 67 22 at at ADP fbem-26173 67 23 the the DET fbem-26173 67 24 same same ADJ fbem-26173 67 25 time time NOUN fbem-26173 67 26 , , PUNCT fbem-26173 67 27 which which PRON fbem-26173 67 28 allows allow VERB fbem-26173 67 29 the the DET fbem-26173 67 30 model model NOUN fbem-26173 67 31 to to PART fbem-26173 67 32 better well ADV fbem-26173 67 33 understand understand VERB fbem-26173 67 34 the the DET fbem-26173 67 35 temporal temporal ADJ fbem-26173 67 36 dependencies dependency NOUN fbem-26173 67 37 and and CCONJ fbem-26173 67 38 patterns pattern NOUN fbem-26173 67 39 in in ADP fbem-26173 67 40 sequence sequence NOUN fbem-26173 67 41 data datum NOUN fbem-26173 67 42 [ [ X fbem-26173 67 43 16 16 NUM fbem-26173 67 44 ] ] PUNCT fbem-26173 67 45 . . PUNCT fbem-26173 68 1 the the DET fbem-26173 68 2 core core NOUN fbem-26173 68 3 of of ADP fbem-26173 68 4 the the DET fbem-26173 68 5 lstm lstm PROPN fbem-26173 68 6 network network NOUN fbem-26173 68 7 is be AUX fbem-26173 68 8 the the DET fbem-26173 68 9 memory memory NOUN fbem-26173 68 10 cell cell NOUN fbem-26173 68 11 , , PUNCT fbem-26173 68 12 which which PRON fbem-26173 68 13 controls control VERB fbem-26173 68 14 the the DET fbem-26173 68 15 input input NOUN fbem-26173 68 16 , , PUNCT fbem-26173 68 17 output output NOUN fbem-26173 68 18 and and CCONJ fbem-26173 68 19 forgetting forget VERB fbem-26173 68 20 of of ADP fbem-26173 68 21 information information NOUN fbem-26173 68 22 through through ADP fbem-26173 68 23 a a DET fbem-26173 68 24 series series NOUN fbem-26173 68 25 of of ADP fbem-26173 68 26 gates gate NOUN fbem-26173 68 27 to to PART fbem-26173 68 28 achieve achieve VERB fbem-26173 68 29 effective effective ADJ fbem-26173 68 30 modelling modelling NOUN fbem-26173 68 31 of of ADP fbem-26173 68 32 sequential sequential ADJ fbem-26173 68 33 data datum NOUN fbem-26173 68 34 , , PUNCT fbem-26173 68 35 and and CCONJ fbem-26173 68 36 is be AUX fbem-26173 68 37 mainly mainly ADV fbem-26173 68 38 composed compose VERB fbem-26173 68 39 of of ADP fbem-26173 68 40 four four NUM fbem-26173 68 41 parts part NOUN fbem-26173 68 42 : : PUNCT fbem-26173 68 43 the the DET fbem-26173 68 44 forgetting forget VERB fbem-26173 68 45 gate gate NOUN fbem-26173 68 46 , , PUNCT fbem-26173 68 47 the the DET fbem-26173 68 48 input input NOUN fbem-26173 68 49 gate gate NOUN fbem-26173 68 50 , , PUNCT fbem-26173 68 51 the the DET fbem-26173 68 52 output output NOUN fbem-26173 68 53 gate gate NOUN fbem-26173 68 54 and and CCONJ fbem-26173 68 55 the the DET fbem-26173 68 56 internal internal ADJ fbem-26173 68 57 memory memory NOUN fbem-26173 68 58 cell cell NOUN fbem-26173 68 59 [ [ X fbem-26173 68 60 17 17 NUM fbem-26173 68 61 ] ] PUNCT fbem-26173 68 62 . . PUNCT fbem-26173 69 1 the the DET fbem-26173 69 2 forgetting forget VERB fbem-26173 69 3 gate gate NOUN fbem-26173 69 4 decides decide VERB fbem-26173 69 5 which which DET fbem-26173 69 6 information information NOUN fbem-26173 69 7 in in ADP fbem-26173 69 8 the the DET fbem-26173 69 9 memory memory NOUN fbem-26173 69 10 cell cell NOUN fbem-26173 69 11 should should AUX fbem-26173 69 12 be be AUX fbem-26173 69 13 forgotten forget VERB fbem-26173 69 14 , , PUNCT fbem-26173 69 15 and and CCONJ fbem-26173 69 16 its its PRON fbem-26173 69 17 calculation calculation NOUN fbem-26173 69 18 process process NOUN fbem-26173 69 19 can can AUX fbem-26173 69 20 be be AUX fbem-26173 69 21 expressed express VERB fbem-26173 69 22 as as SCONJ fbem-26173 69 23 follows follow VERB fbem-26173 69 24 : : PUNCT fbem-26173 69 25 𝑓𝑑 𝑓𝑑 PROPN fbem-26173 69 26 = = PUNCT fbem-26173 69 27 𝜎(π‘Šπ‘“π‘₯𝑑 𝜎(π‘Šπ‘“π‘₯𝑑 PROPN fbem-26173 69 28 + + CCONJ fbem-26173 69 29 π‘ˆπ‘“β„Žπ‘‘βˆ’1 π‘ˆπ‘“β„Žπ‘‘βˆ’1 X fbem-26173 69 30 + + CCONJ fbem-26173 69 31 𝑏𝑓 𝑏𝑓 PROPN fbem-26173 69 32 ) ) PUNCT fbem-26173 69 33 ( ( PUNCT fbem-26173 69 34 5 5 X fbem-26173 69 35 ) ) PUNCT fbem-26173 69 36 the the DET fbem-26173 69 37 input input NOUN fbem-26173 69 38 gate gate NOUN fbem-26173 69 39 determines determine NOUN fbem-26173 69 40 which which PRON fbem-26173 69 41 parts part NOUN fbem-26173 69 42 of of ADP fbem-26173 69 43 the the DET fbem-26173 69 44 input input NOUN fbem-26173 69 45 data datum NOUN fbem-26173 69 46 should should AUX fbem-26173 69 47 be be AUX fbem-26173 69 48 updated update VERB fbem-26173 69 49 into into ADP fbem-26173 69 50 the the DET fbem-26173 69 51 memory memory NOUN fbem-26173 69 52 cell cell NOUN fbem-26173 69 53 , , PUNCT fbem-26173 69 54 and and CCONJ fbem-26173 69 55 its its PRON fbem-26173 69 56 computation computation NOUN fbem-26173 69 57 can can AUX fbem-26173 69 58 be be AUX fbem-26173 69 59 expressed express VERB fbem-26173 69 60 as as SCONJ fbem-26173 69 61 follows follow VERB fbem-26173 69 62 : : PUNCT fbem-26173 69 63 78 78 NUM fbem-26173 69 64 𝑖𝑑 𝑖𝑑 NOUN fbem-26173 70 1 = = PUNCT fbem-26173 70 2 𝜎(π‘Šπ‘–π‘₯𝑑 𝜎(π‘Šπ‘–π‘₯𝑑 PROPN fbem-26173 70 3 + + CCONJ fbem-26173 70 4 π‘ˆπ‘–β„Žπ‘‘βˆ’1 π‘ˆπ‘–β„Žπ‘‘βˆ’1 ADJ fbem-26173 70 5 + + CCONJ fbem-26173 70 6 𝑏𝑖 𝑏𝑖 X fbem-26173 70 7 ) ) PUNCT fbem-26173 70 8 ( ( PUNCT fbem-26173 70 9 6 6 X fbem-26173 70 10 ) ) PUNCT fbem-26173 71 1 the the DET fbem-26173 71 2 internal internal ADJ fbem-26173 71 3 memory memory NOUN fbem-26173 71 4 unit unit NOUN fbem-26173 71 5 updates update VERB fbem-26173 71 6 the the DET fbem-26173 71 7 current current ADJ fbem-26173 71 8 memory memory NOUN fbem-26173 71 9 state state NOUN fbem-26173 71 10 based base VERB fbem-26173 71 11 on on ADP fbem-26173 71 12 the the DET fbem-26173 71 13 candidate candidate NOUN fbem-26173 71 14 memory memory NOUN fbem-26173 71 15 state state NOUN fbem-26173 71 16 information information NOUN fbem-26173 71 17 , , PUNCT fbem-26173 71 18 which which PRON fbem-26173 71 19 may may AUX fbem-26173 71 20 be be AUX fbem-26173 71 21 represented represent VERB fbem-26173 71 22 as as ADP fbem-26173 71 23 : : PUNCT fbem-26173 71 24 οΏ½ οΏ½ PROPN fbem-26173 71 25 Μƒ Μƒ PROPN fbem-26173 71 26 οΏ½ οΏ½ NOUN fbem-26173 71 27 𝑑 𝑑 NOUN fbem-26173 71 28 = = SYM fbem-26173 71 29 π‘‘π‘Žπ‘›β„Ž(π‘Šπ‘π‘₯𝑑 π‘‘π‘Žπ‘›β„Ž(π‘Šπ‘π‘₯𝑑 NOUN fbem-26173 71 30 + + NOUN fbem-26173 71 31 π‘ˆπ‘β„Žπ‘‘βˆ’1 π‘ˆπ‘β„Žπ‘‘βˆ’1 NOUN fbem-26173 71 32 + + CCONJ fbem-26173 71 33 𝑏𝑐 𝑏𝑐 X fbem-26173 71 34 ) ) PUNCT fbem-26173 71 35 ( ( PUNCT fbem-26173 71 36 7 7 X fbem-26173 71 37 ) ) PUNCT fbem-26173 71 38 𝐢𝑑 𝐢𝑑 PROPN fbem-26173 71 39 = = SYM fbem-26173 71 40 𝑓𝑑 𝑓𝑑 PROPN fbem-26173 71 41 βˆ— βˆ— X fbem-26173 71 42 πΆπ‘‘βˆ’1 πΆπ‘‘βˆ’1 PROPN fbem-26173 71 43 + + CCONJ fbem-26173 71 44 𝑖𝑑 𝑖𝑑 PROPN fbem-26173 71 45 βˆ— βˆ— PROPN fbem-26173 71 46 οΏ½ οΏ½ PROPN fbem-26173 71 47 Μƒ Μƒ PROPN fbem-26173 71 48 οΏ½ οΏ½ PROPN fbem-26173 71 49 𝑑 𝑑 PROPN fbem-26173 71 50 ( ( PUNCT fbem-26173 71 51 8) 8) NUM fbem-26173 71 52 the the DET fbem-26173 71 53 output output NOUN fbem-26173 71 54 gate gate NOUN fbem-26173 71 55 determines determine NOUN fbem-26173 71 56 which which DET fbem-26173 71 57 information information NOUN fbem-26173 71 58 in in ADP fbem-26173 71 59 the the DET fbem-26173 71 60 memory memory NOUN fbem-26173 71 61 cell cell NOUN fbem-26173 71 62 should should AUX fbem-26173 71 63 be be AUX fbem-26173 71 64 output output NOUN fbem-26173 71 65 to to ADP fbem-26173 71 66 the the DET fbem-26173 71 67 hidden hide VERB fbem-26173 71 68 states state NOUN fbem-26173 71 69 and and CCONJ fbem-26173 71 70 outputs output NOUN fbem-26173 71 71 at at ADP fbem-26173 71 72 the the DET fbem-26173 71 73 next next ADJ fbem-26173 71 74 moment moment NOUN fbem-26173 71 75 and and CCONJ fbem-26173 71 76 can can AUX fbem-26173 71 77 be be AUX fbem-26173 71 78 expressed express VERB fbem-26173 71 79 as as SCONJ fbem-26173 71 80 follows follow VERB fbem-26173 71 81 : : PUNCT fbem-26173 71 82 𝑂𝑑 𝑂𝑑 PROPN fbem-26173 71 83 = = PUNCT fbem-26173 71 84 𝜎(π‘Šπ‘œπ‘₯𝑑 𝜎(π‘Šπ‘œπ‘₯𝑑 PROPN fbem-26173 72 1 + + CCONJ fbem-26173 72 2 π‘ˆπ‘œβ„Žπ‘‘βˆ’1 π‘ˆπ‘œβ„Žπ‘‘βˆ’1 X fbem-26173 73 1 + + CCONJ fbem-26173 73 2 π‘π‘œ π‘π‘œ ADJ fbem-26173 73 3 ) ) PUNCT fbem-26173 73 4 ( ( PUNCT fbem-26173 73 5 9 9 X fbem-26173 73 6 ) ) PUNCT fbem-26173 73 7 β„Žπ‘‘ β„Žπ‘‘ NOUN fbem-26173 73 8 = = PUNCT fbem-26173 74 1 𝑂𝑑 𝑂𝑑 PROPN fbem-26173 74 2 βˆ— βˆ— NOUN fbem-26173 74 3 tanh tanh NOUN fbem-26173 74 4 ( ( PUNCT fbem-26173 74 5 𝐢𝑑 𝐢𝑑 PROPN fbem-26173 74 6 ) ) PUNCT fbem-26173 74 7 ( ( PUNCT fbem-26173 74 8 10 10 NUM fbem-26173 74 9 ) ) PUNCT fbem-26173 74 10 where where SCONJ fbem-26173 74 11 π‘Šπ‘“ π‘Šπ‘“ PROPN fbem-26173 74 12 , , PUNCT fbem-26173 74 13 π‘Šπ‘– π‘Šπ‘– PROPN fbem-26173 74 14 , , PUNCT fbem-26173 74 15 π‘Šπ‘ π‘Šπ‘ PROPN fbem-26173 74 16 , , PUNCT fbem-26173 74 17 π‘Šπ‘œ π‘Šπ‘œ PROPN fbem-26173 74 18 are be AUX fbem-26173 74 19 the the DET fbem-26173 74 20 weight weight NOUN fbem-26173 74 21 matrices matrix NOUN fbem-26173 74 22 of of ADP fbem-26173 74 23 the the DET fbem-26173 74 24 forgetting forget VERB fbem-26173 74 25 gate gate NOUN fbem-26173 74 26 , , PUNCT fbem-26173 74 27 the the DET fbem-26173 74 28 input input NOUN fbem-26173 74 29 gate gate NOUN fbem-26173 74 30 , , PUNCT fbem-26173 74 31 the the DET fbem-26173 74 32 update update NOUN fbem-26173 74 33 unit unit NOUN fbem-26173 74 34 link link NOUN fbem-26173 74 35 and and CCONJ fbem-26173 74 36 the the DET fbem-26173 74 37 output output NOUN fbem-26173 74 38 gate gate NOUN fbem-26173 74 39 , , PUNCT fbem-26173 74 40 𝑏𝑓 𝑏𝑓 PRON fbem-26173 74 41 , , PUNCT fbem-26173 74 42 𝑏𝑖 𝑏𝑖 PROPN fbem-26173 74 43 , , PUNCT fbem-26173 74 44 𝑏𝑐 𝑏𝑐 PROPN fbem-26173 74 45 , , PUNCT fbem-26173 74 46 π‘π‘œ π‘π‘œ PROPN fbem-26173 74 47 are be AUX fbem-26173 74 48 their their PRON fbem-26173 74 49 offsets offset NOUN fbem-26173 74 50 , , PUNCT fbem-26173 74 51 𝑓𝑑 𝑓𝑑 PROPN fbem-26173 74 52 , , PUNCT fbem-26173 74 53 𝑖𝑑 𝑖𝑑 NOUN fbem-26173 74 54 , , PUNCT fbem-26173 74 55 𝑂𝑑 𝑂𝑑 PROPN fbem-26173 74 56 are be AUX fbem-26173 74 57 the the DET fbem-26173 74 58 outputs output NOUN fbem-26173 74 59 of of ADP fbem-26173 74 60 the the DET fbem-26173 74 61 forgetting forget VERB fbem-26173 74 62 gate gate NOUN fbem-26173 74 63 , , PUNCT fbem-26173 74 64 the the DET fbem-26173 74 65 input input NOUN fbem-26173 74 66 gate gate NOUN fbem-26173 74 67 , , PUNCT fbem-26173 74 68 and and CCONJ fbem-26173 74 69 the the DET fbem-26173 74 70 output output NOUN fbem-26173 74 71 gate gate NOUN fbem-26173 74 72 , , PUNCT fbem-26173 74 73 respectively respectively ADV fbem-26173 74 74 , , PUNCT fbem-26173 74 75 𝜎 𝜎 PROPN fbem-26173 74 76 is be AUX fbem-26173 74 77 the the DET fbem-26173 74 78 activation activation NOUN fbem-26173 74 79 function function NOUN fbem-26173 74 80 sigmoid sigmoid NOUN fbem-26173 74 81 , , PUNCT fbem-26173 74 82 β„Žπ‘‘ β„Žπ‘‘ NOUN fbem-26173 74 83 is be AUX fbem-26173 74 84 the the DET fbem-26173 74 85 hidden hidden ADJ fbem-26173 74 86 state state NOUN fbem-26173 74 87 within within ADP fbem-26173 74 88 the the DET fbem-26173 74 89 unit unit NOUN fbem-26173 74 90 , , PUNCT fbem-26173 74 91 and and CCONJ fbem-26173 74 92 π‘ˆ π‘ˆ PROPN fbem-26173 74 93 is be AUX fbem-26173 74 94 the the DET fbem-26173 74 95 corresponding corresponding ADJ fbem-26173 74 96 weight weight NOUN fbem-26173 74 97 matrices matrix NOUN fbem-26173 74 98 of of ADP fbem-26173 74 99 the the DET fbem-26173 74 100 three three NUM fbem-26173 74 101 gates gate NOUN fbem-26173 74 102 . . PUNCT fbem-26173 75 1 the the DET fbem-26173 75 2 bidirectional bidirectional ADJ fbem-26173 75 3 long long ADJ fbem-26173 75 4 short short ADJ fbem-26173 75 5 - - PUNCT fbem-26173 75 6 term term NOUN fbem-26173 75 7 memory memory NOUN fbem-26173 75 8 neural neural ADJ fbem-26173 75 9 network network NOUN fbem-26173 75 10 ( ( PUNCT fbem-26173 75 11 bilstm bilstm NOUN fbem-26173 75 12 ) ) PUNCT fbem-26173 75 13 model model NOUN fbem-26173 75 14 splices splice NOUN fbem-26173 75 15 the the DET fbem-26173 75 16 hidden hide VERB fbem-26173 75 17 states state NOUN fbem-26173 75 18 of of ADP fbem-26173 75 19 the the DET fbem-26173 75 20 forward forward ADJ fbem-26173 75 21 lstm lstm NOUN fbem-26173 75 22 and and CCONJ fbem-26173 75 23 the the DET fbem-26173 75 24 inverse inverse ADJ fbem-26173 75 25 lstm lstm NOUN fbem-26173 75 26 , , PUNCT fbem-26173 75 27 which which PRON fbem-26173 75 28 can can AUX fbem-26173 75 29 refer refer VERB fbem-26173 75 30 to to ADP fbem-26173 75 31 both both CCONJ fbem-26173 75 32 historical historical ADJ fbem-26173 75 33 and and CCONJ fbem-26173 75 34 future future ADJ fbem-26173 75 35 data datum NOUN fbem-26173 75 36 on on ADP fbem-26173 75 37 the the DET fbem-26173 75 38 prediction prediction NOUN fbem-26173 75 39 results result NOUN fbem-26173 75 40 , , PUNCT fbem-26173 75 41 and and CCONJ fbem-26173 75 42 can can AUX fbem-26173 75 43 be be AUX fbem-26173 75 44 expressed express VERB fbem-26173 75 45 as as ADP fbem-26173 75 46 : : PUNCT fbem-26173 75 47 𝐻 𝐻 PROPN fbem-26173 75 48 = = PUNCT fbem-26173 76 1 [ [ X fbem-26173 76 2 β„Žβƒ— β„Žβƒ— NOUN fbem-26173 76 3 1 1 NUM fbem-26173 76 4 , , PUNCT fbem-26173 76 5 β„Žβƒ— β„Žβƒ— NOUN fbem-26173 76 6 2 2 NUM fbem-26173 76 7 , , PUNCT fbem-26173 76 8 β‹― β‹― PROPN fbem-26173 76 9 , , PUNCT fbem-26173 76 10 β„Žβƒ— β„Žβƒ— PROPN fbem-26173 76 11 𝑇 𝑇 PROPN fbem-26173 76 12 , , PUNCT fbem-26173 76 13 β„Žβƒ–βƒ—π‘‡ β„Žβƒ–βƒ—π‘‡ PROPN fbem-26173 76 14 , , PUNCT fbem-26173 76 15 β‹― β‹― PROPN fbem-26173 76 16 , , PUNCT fbem-26173 76 17 β„Žβƒ–βƒ—2 β„Žβƒ–βƒ—2 PROPN fbem-26173 76 18 , , PUNCT fbem-26173 76 19 β„Žβƒ–βƒ—1 β„Žβƒ–βƒ—1 PROPN fbem-26173 76 20 ] ] PUNCT fbem-26173 76 21 ( ( PUNCT fbem-26173 76 22 11 11 NUM fbem-26173 76 23 ) ) PUNCT fbem-26173 76 24 where where SCONJ fbem-26173 76 25 𝐻 𝐻 PROPN fbem-26173 76 26 is be AUX fbem-26173 76 27 the the DET fbem-26173 76 28 output output NOUN fbem-26173 76 29 of of ADP fbem-26173 76 30 bilstm bilstm NOUN fbem-26173 76 31 . . PUNCT fbem-26173 77 1 2.4 2.4 NUM fbem-26173 77 2 . . PUNCT fbem-26173 78 1 attention attention NOUN fbem-26173 78 2 mechanism mechanism NOUN fbem-26173 78 3 attention attention NOUN fbem-26173 78 4 mechanism mechanism NOUN fbem-26173 78 5 is be AUX fbem-26173 78 6 a a DET fbem-26173 78 7 technique technique NOUN fbem-26173 78 8 used use VERB fbem-26173 78 9 to to PART fbem-26173 78 10 enhance enhance VERB fbem-26173 78 11 the the DET fbem-26173 78 12 ability ability NOUN fbem-26173 78 13 of of ADP fbem-26173 78 14 neural neural ADJ fbem-26173 78 15 networks network NOUN fbem-26173 78 16 when when SCONJ fbem-26173 78 17 processing process VERB fbem-26173 78 18 sequential sequential ADJ fbem-26173 78 19 data datum NOUN fbem-26173 78 20 [ [ X fbem-26173 78 21 18 18 NUM fbem-26173 78 22 ] ] PUNCT fbem-26173 78 23 . . PUNCT fbem-26173 79 1 attention attention NOUN fbem-26173 79 2 mechanism mechanism NOUN fbem-26173 79 3 allows allow VERB fbem-26173 79 4 the the DET fbem-26173 79 5 model model NOUN fbem-26173 79 6 to to PART fbem-26173 79 7 assign assign VERB fbem-26173 79 8 different different ADJ fbem-26173 79 9 weights weight NOUN fbem-26173 79 10 and and CCONJ fbem-26173 79 11 attention attention NOUN fbem-26173 79 12 to to ADP fbem-26173 79 13 different different ADJ fbem-26173 79 14 parts part NOUN fbem-26173 79 15 of of ADP fbem-26173 79 16 the the DET fbem-26173 79 17 input input NOUN fbem-26173 79 18 when when SCONJ fbem-26173 79 19 processing process VERB fbem-26173 79 20 sequential sequential ADJ fbem-26173 79 21 data datum NOUN fbem-26173 79 22 . . PUNCT fbem-26173 80 1 in in ADP fbem-26173 80 2 this this DET fbem-26173 80 3 paper paper NOUN fbem-26173 80 4 , , PUNCT fbem-26173 80 5 the the DET fbem-26173 80 6 attention attention NOUN fbem-26173 80 7 mechanism mechanism NOUN fbem-26173 80 8 can can AUX fbem-26173 80 9 be be AUX fbem-26173 80 10 used use VERB fbem-26173 80 11 to to PART fbem-26173 80 12 dynamically dynamically ADV fbem-26173 80 13 select select VERB fbem-26173 80 14 and and CCONJ fbem-26173 80 15 weight weight VERB fbem-26173 80 16 the the DET fbem-26173 80 17 importance importance NOUN fbem-26173 80 18 of of ADP fbem-26173 80 19 different different ADJ fbem-26173 80 20 feature feature NOUN fbem-26173 80 21 variables variable NOUN fbem-26173 80 22 , , PUNCT fbem-26173 80 23 allowing allow VERB fbem-26173 80 24 the the DET fbem-26173 80 25 model model NOUN fbem-26173 80 26 to to PART fbem-26173 80 27 pay pay VERB fbem-26173 80 28 more more ADJ fbem-26173 80 29 attention attention NOUN fbem-26173 80 30 to to ADP fbem-26173 80 31 features feature NOUN fbem-26173 80 32 that that PRON fbem-26173 80 33 have have VERB fbem-26173 80 34 a a DET fbem-26173 80 35 greater great ADJ fbem-26173 80 36 impact impact NOUN fbem-26173 80 37 on on ADP fbem-26173 80 38 stock stock NOUN fbem-26173 80 39 price price NOUN fbem-26173 80 40 forecasting forecasting NOUN fbem-26173 80 41 . . PUNCT fbem-26173 81 1 the the DET fbem-26173 81 2 attention attention NOUN fbem-26173 81 3 weights weight NOUN fbem-26173 81 4 can can AUX fbem-26173 81 5 be be AUX fbem-26173 81 6 obtained obtain VERB fbem-26173 81 7 by by ADP fbem-26173 81 8 calculating calculate VERB fbem-26173 81 9 the the DET fbem-26173 81 10 similarity similarity NOUN fbem-26173 81 11 between between ADP fbem-26173 81 12 the the DET fbem-26173 81 13 input input NOUN fbem-26173 81 14 sequences sequence NOUN fbem-26173 81 15 and and CCONJ fbem-26173 81 16 the the DET fbem-26173 81 17 hidden hidden ADJ fbem-26173 81 18 states state NOUN fbem-26173 81 19 of of ADP fbem-26173 81 20 the the DET fbem-26173 81 21 bilstm bilstm NOUN fbem-26173 81 22 . . PUNCT fbem-26173 82 1 commonly commonly ADV fbem-26173 82 2 used use VERB fbem-26173 82 3 calculation calculation NOUN fbem-26173 82 4 methods method NOUN fbem-26173 82 5 include include VERB fbem-26173 82 6 dot dot NOUN fbem-26173 82 7 product product NOUN fbem-26173 82 8 , , PUNCT fbem-26173 82 9 additive additive NOUN fbem-26173 82 10 and and CCONJ fbem-26173 82 11 multiplicative multiplicative ADJ fbem-26173 82 12 attention attention NOUN fbem-26173 82 13 mechanisms mechanism NOUN fbem-26173 82 14 . . PUNCT fbem-26173 83 1 multiplicative multiplicative ADJ fbem-26173 83 2 attention attention NOUN fbem-26173 83 3 mechanism mechanism NOUN fbem-26173 83 4 : : PUNCT fbem-26173 83 5 𝑒𝑑 𝑒𝑑 NOUN fbem-26173 83 6 = = PUNCT fbem-26173 83 7 π‘ π‘œπ‘“π‘‘π‘šπ‘Žπ‘₯(π‘Šπ‘Ž π‘ π‘œπ‘“π‘‘π‘šπ‘Žπ‘₯(π‘Šπ‘Ž PROPN fbem-26173 83 8 βˆ™ βˆ™ PROPN fbem-26173 83 9 tanh tanh NOUN fbem-26173 83 10 ( ( PUNCT fbem-26173 83 11 π‘Šβ„Ž π‘Šβ„Ž PROPN fbem-26173 83 12 βˆ™ βˆ™ PROPN fbem-26173 83 13 𝐻 𝐻 PROPN fbem-26173 83 14 + + NUM fbem-26173 83 15 π‘β„Ž π‘β„Ž NOUN fbem-26173 83 16 ) ) PUNCT fbem-26173 83 17 ) ) PUNCT fbem-26173 83 18 , , PUNCT fbem-26173 83 19 where where SCONJ fbem-26173 83 20 π‘Šπ‘Ž π‘Šπ‘Ž PROPN fbem-26173 83 21 and and CCONJ fbem-26173 83 22 π‘Šβ„Ž π‘Šβ„Ž PROPN fbem-26173 83 23 are be AUX fbem-26173 83 24 the the DET fbem-26173 83 25 learned learn VERB fbem-26173 83 26 weight weight NOUN fbem-26173 83 27 matrices matrix NOUN fbem-26173 83 28 , , PUNCT fbem-26173 83 29 π‘β„Ž π‘β„Ž NOUN fbem-26173 83 30 is be AUX fbem-26173 83 31 the the DET fbem-26173 83 32 bias bias NOUN fbem-26173 83 33 vector vector NOUN fbem-26173 83 34 , , PUNCT fbem-26173 83 35 and and CCONJ fbem-26173 83 36 𝐻 𝐻 PROPN fbem-26173 83 37 is be AUX fbem-26173 83 38 the the DET fbem-26173 83 39 output output NOUN fbem-26173 83 40 of of ADP fbem-26173 83 41 the the DET fbem-26173 83 42 bilstm bilstm NOUN fbem-26173 83 43 . . PUNCT fbem-26173 84 1 attention attention NOUN fbem-26173 84 2 weights weight NOUN fbem-26173 84 3 are be AUX fbem-26173 84 4 applied apply VERB fbem-26173 84 5 to to ADP fbem-26173 84 6 the the DET fbem-26173 84 7 outputs output NOUN fbem-26173 84 8 of of ADP fbem-26173 84 9 bilstm bilstm NOUN fbem-26173 84 10 , , PUNCT fbem-26173 84 11 which which PRON fbem-26173 84 12 are be AUX fbem-26173 84 13 weighted weight VERB fbem-26173 84 14 and and CCONJ fbem-26173 84 15 summed sum VERB fbem-26173 84 16 to to PART fbem-26173 84 17 obtain obtain VERB fbem-26173 84 18 an an DET fbem-26173 84 19 attentionweighted attentionweighte VERB fbem-26173 84 20 representation representation NOUN fbem-26173 84 21 : : PUNCT fbem-26173 84 22 𝐴 𝐴 PROPN fbem-26173 84 23 = = PUNCT fbem-26173 84 24 βˆ‘ βˆ‘ PUNCT fbem-26173 84 25 𝑒𝑑 𝑒𝑑 ADP fbem-26173 84 26 βˆ™ βˆ™ PROPN fbem-26173 84 27 𝑇 𝑇 PROPN fbem-26173 84 28 𝑑=1 𝑑=1 PROPN fbem-26173 84 29 𝐻𝑑 𝐻𝑑 NOUN fbem-26173 84 30 ( ( PUNCT fbem-26173 84 31 12 12 NUM fbem-26173 84 32 ) ) PUNCT fbem-26173 84 33 finally finally ADV fbem-26173 84 34 , , PUNCT fbem-26173 84 35 the the DET fbem-26173 84 36 attention attention NOUN fbem-26173 84 37 - - PUNCT fbem-26173 84 38 weighted weight VERB fbem-26173 84 39 representation representation NOUN fbem-26173 84 40 a a PRON fbem-26173 84 41 can can AUX fbem-26173 84 42 be be AUX fbem-26173 84 43 fed feed VERB fbem-26173 84 44 into into ADP fbem-26173 84 45 a a DET fbem-26173 84 46 subsequent subsequent ADJ fbem-26173 84 47 layer layer NOUN fbem-26173 84 48 ( ( PUNCT fbem-26173 84 49 e.g. e.g. ADV fbem-26173 84 50 , , PUNCT fbem-26173 84 51 a a DET fbem-26173 84 52 fully fully ADV fbem-26173 84 53 connected connect VERB fbem-26173 84 54 layer layer NOUN fbem-26173 84 55 ) ) PUNCT fbem-26173 84 56 for for ADP fbem-26173 84 57 further further ADJ fbem-26173 84 58 processing processing NOUN fbem-26173 84 59 and and CCONJ fbem-26173 84 60 prediction prediction NOUN fbem-26173 84 61 . . PUNCT fbem-26173 85 1 2.5 2.5 NUM fbem-26173 85 2 . . PUNCT fbem-26173 86 1 cnn cnn PROPN fbem-26173 86 2 - - PUNCT fbem-26173 86 3 bilstm bilstm NOUN fbem-26173 86 4 - - PUNCT fbem-26173 86 5 attention attention NOUN fbem-26173 86 6 model model NOUN fbem-26173 86 7 in in ADP fbem-26173 86 8 this this DET fbem-26173 86 9 paper paper NOUN fbem-26173 86 10 , , PUNCT fbem-26173 86 11 three three NUM fbem-26173 86 12 models model NOUN fbem-26173 86 13 , , PUNCT fbem-26173 86 14 cnn cnn PROPN fbem-26173 86 15 , , PUNCT fbem-26173 86 16 bilstm bilstm NOUN fbem-26173 86 17 and and CCONJ fbem-26173 86 18 attention attention NOUN fbem-26173 86 19 , , PUNCT fbem-26173 86 20 are be AUX fbem-26173 86 21 combined combine VERB fbem-26173 86 22 to to PART fbem-26173 86 23 form form VERB fbem-26173 86 24 a a DET fbem-26173 86 25 stock stock NOUN fbem-26173 86 26 prediction prediction NOUN fbem-26173 86 27 framework framework NOUN fbem-26173 86 28 cnnbilstm cnnbilstm NOUN fbem-26173 86 29 - - PUNCT fbem-26173 86 30 attention attention NOUN fbem-26173 86 31 . . PUNCT fbem-26173 87 1 the the DET fbem-26173 87 2 network network NOUN fbem-26173 87 3 structure structure NOUN fbem-26173 87 4 of of ADP fbem-26173 87 5 the the DET fbem-26173 87 6 stock stock NOUN fbem-26173 87 7 price price NOUN fbem-26173 87 8 prediction prediction NOUN fbem-26173 87 9 model model PROPN fbem-26173 87 10 cnn cnn PROPN fbem-26173 87 11 - - PUNCT fbem-26173 87 12 bilstm bilstm NOUN fbem-26173 87 13 - - PUNCT fbem-26173 87 14 attention attention NOUN fbem-26173 87 15 is be AUX fbem-26173 87 16 shown show VERB fbem-26173 87 17 in in ADP fbem-26173 87 18 fig fig NOUN fbem-26173 87 19 . . PUNCT fbem-26173 88 1 1 1 X fbem-26173 88 2 . . PUNCT fbem-26173 88 3 firstly firstly ADV fbem-26173 88 4 , , PUNCT fbem-26173 88 5 the the DET fbem-26173 88 6 model model NOUN fbem-26173 88 7 extracted extract VERB fbem-26173 88 8 five five NUM fbem-26173 88 9 stock stock NOUN fbem-26173 88 10 features feature NOUN fbem-26173 88 11 in in ADP fbem-26173 88 12 historical historical ADJ fbem-26173 88 13 time time NOUN fbem-26173 88 14 through through ADP fbem-26173 88 15 cnn cnn PROPN fbem-26173 88 16 , , PUNCT fbem-26173 88 17 secondly secondly ADV fbem-26173 88 18 , , PUNCT fbem-26173 88 19 bilstm bilstm NOUN fbem-26173 88 20 trained train VERB fbem-26173 88 21 and and CCONJ fbem-26173 88 22 predicted predict VERB fbem-26173 88 23 daily daily ADJ fbem-26173 88 24 individual individual ADJ fbem-26173 88 25 stock stock NOUN fbem-26173 88 26 data datum NOUN fbem-26173 88 27 through through ADP fbem-26173 88 28 two two NUM fbem-26173 88 29 lstm lstm ADJ fbem-26173 88 30 layers layer NOUN fbem-26173 88 31 , , PUNCT fbem-26173 88 32 forward forward ADV fbem-26173 88 33 and and CCONJ fbem-26173 88 34 backward backward ADV fbem-26173 88 35 , , PUNCT fbem-26173 88 36 which which PRON fbem-26173 88 37 captured capture VERB fbem-26173 88 38 the the DET fbem-26173 88 39 forward forward NOUN fbem-26173 88 40 and and CCONJ fbem-26173 88 41 reverse reverse VERB fbem-26173 88 42 information information NOUN fbem-26173 88 43 of of ADP fbem-26173 88 44 the the DET fbem-26173 88 45 sequence sequence NOUN fbem-26173 88 46 , , PUNCT fbem-26173 88 47 respectively respectively ADV fbem-26173 88 48 , , PUNCT fbem-26173 88 49 and and CCONJ fbem-26173 88 50 based base VERB fbem-26173 88 51 on on ADP fbem-26173 88 52 this this PRON fbem-26173 88 53 , , PUNCT fbem-26173 88 54 the the DET fbem-26173 88 55 attention attention NOUN fbem-26173 88 56 mechanism mechanism NOUN fbem-26173 88 57 was be AUX fbem-26173 88 58 introduced introduce VERB fbem-26173 88 59 , , PUNCT fbem-26173 88 60 which which PRON fbem-26173 88 61 gave give VERB fbem-26173 88 62 a a DET fbem-26173 88 63 different different ADJ fbem-26173 88 64 weight weight NOUN fbem-26173 88 65 to to ADP fbem-26173 88 66 the the DET fbem-26173 88 67 features feature NOUN fbem-26173 88 68 of of ADP fbem-26173 88 69 each each DET fbem-26173 88 70 stock stock NOUN fbem-26173 88 71 , , PUNCT fbem-26173 88 72 to to PART fbem-26173 88 73 improve improve VERB fbem-26173 88 74 the the DET fbem-26173 88 75 neural neural ADJ fbem-26173 88 76 network network NOUN fbem-26173 88 77 's 's PART fbem-26173 88 78 learning learning NOUN fbem-26173 88 79 ability ability NOUN fbem-26173 88 80 of of ADP fbem-26173 88 81 the the DET fbem-26173 88 82 neural neural ADJ fbem-26173 88 83 network network NOUN fbem-26173 88 84 to to PART fbem-26173 88 85 achieve achieve VERB fbem-26173 88 86 effective effective ADJ fbem-26173 88 87 mining mining NOUN fbem-26173 88 88 of of ADP fbem-26173 88 89 individual individual ADJ fbem-26173 88 90 stock stock NOUN fbem-26173 88 91 features feature NOUN fbem-26173 88 92 . . PUNCT fbem-26173 89 1 figure figure VERB fbem-26173 89 2 1 1 NUM fbem-26173 89 3 . . PUNCT fbem-26173 90 1 network network NOUN fbem-26173 90 2 structure structure NOUN fbem-26173 90 3 of of ADP fbem-26173 90 4 stock stock NOUN fbem-26173 90 5 price price NOUN fbem-26173 90 6 prediction prediction NOUN fbem-26173 90 7 model model NOUN fbem-26173 90 8 cnnbilstm cnnbilstm PROPN fbem-26173 90 9 - - PUNCT fbem-26173 90 10 attention attention NOUN fbem-26173 90 11 2.6 2.6 NUM fbem-26173 90 12 . . PUNCT fbem-26173 91 1 quantitative quantitative ADJ fbem-26173 91 2 trading trading NOUN fbem-26173 91 3 strategies strategy NOUN fbem-26173 91 4 this this DET fbem-26173 91 5 article article NOUN fbem-26173 91 6 is be AUX fbem-26173 91 7 a a DET fbem-26173 91 8 quantitative quantitative ADJ fbem-26173 91 9 trading trading NOUN fbem-26173 91 10 strategy strategy NOUN fbem-26173 91 11 based base VERB fbem-26173 91 12 on on ADP fbem-26173 91 13 the the DET fbem-26173 91 14 macd macd NOUN fbem-26173 91 15 indicator indicator NOUN fbem-26173 91 16 . . PUNCT fbem-26173 92 1 macd macd NOUN fbem-26173 92 2 ( ( PUNCT fbem-26173 92 3 moving move VERB fbem-26173 92 4 average average ADJ fbem-26173 92 5 convergence convergence NOUN fbem-26173 92 6 divergence divergence NOUN fbem-26173 92 7 ) ) PUNCT fbem-26173 92 8 is be AUX fbem-26173 92 9 an an DET fbem-26173 92 10 indicator indicator NOUN fbem-26173 92 11 commonly commonly ADV fbem-26173 92 12 used use VERB fbem-26173 92 13 in in ADP fbem-26173 92 14 technical technical ADJ fbem-26173 92 15 analysis analysis NOUN fbem-26173 92 16 to to PART fbem-26173 92 17 identify identify VERB fbem-26173 92 18 the the DET fbem-26173 92 19 strength strength NOUN fbem-26173 92 20 , , PUNCT fbem-26173 92 21 direction direction NOUN fbem-26173 92 22 and and CCONJ fbem-26173 92 23 possible possible ADJ fbem-26173 92 24 reversal reversal NOUN fbem-26173 92 25 points point NOUN fbem-26173 92 26 of of ADP fbem-26173 92 27 a a DET fbem-26173 92 28 trend trend NOUN fbem-26173 92 29 [ [ X fbem-26173 92 30 19 19 NUM fbem-26173 92 31 ] ] PUNCT fbem-26173 92 32 . . PUNCT fbem-26173 93 1 macd macd NOUN fbem-26173 93 2 consists consist VERB fbem-26173 93 3 of of ADP fbem-26173 93 4 two two NUM fbem-26173 93 5 moving move VERB fbem-26173 93 6 averages average NOUN fbem-26173 93 7 : : PUNCT fbem-26173 93 8 a a DET fbem-26173 93 9 fast fast ADJ fbem-26173 93 10 line line NOUN fbem-26173 93 11 ( ( PUNCT fbem-26173 93 12 macd macd NOUN fbem-26173 93 13 line line NOUN fbem-26173 93 14 ) ) PUNCT fbem-26173 93 15 and and CCONJ fbem-26173 93 16 a a DET fbem-26173 93 17 slow slow ADJ fbem-26173 93 18 line line NOUN fbem-26173 93 19 ( ( PUNCT fbem-26173 93 20 signal signal ADJ fbem-26173 93 21 line line NOUN fbem-26173 93 22 ) ) PUNCT fbem-26173 93 23 , , PUNCT fbem-26173 93 24 and and CCONJ fbem-26173 93 25 a a DET fbem-26173 93 26 bar bar NOUN fbem-26173 93 27 chart chart NOUN fbem-26173 93 28 that that PRON fbem-26173 93 29 indicates indicate VERB fbem-26173 93 30 the the DET fbem-26173 93 31 difference difference NOUN fbem-26173 93 32 between between ADP fbem-26173 93 33 the the DET fbem-26173 93 34 two two NUM fbem-26173 93 35 . . PUNCT fbem-26173 94 1 when when SCONJ fbem-26173 94 2 the the DET fbem-26173 94 3 macd macd NOUN fbem-26173 94 4 line line NOUN fbem-26173 94 5 crosses cross VERB fbem-26173 94 6 the the DET fbem-26173 94 7 signal signal ADJ fbem-26173 94 8 line line NOUN fbem-26173 94 9 from from ADP fbem-26173 94 10 below below ADV fbem-26173 94 11 , , PUNCT fbem-26173 94 12 creating create VERB fbem-26173 94 13 a a DET fbem-26173 94 14 β€˜ ' PUNCT fbem-26173 94 15 golden golden ADJ fbem-26173 94 16 cross cross NOUN fbem-26173 94 17 ’ ' PUNCT fbem-26173 94 18 , , PUNCT fbem-26173 94 19 it it PRON fbem-26173 94 20 indicates indicate VERB fbem-26173 94 21 the the DET fbem-26173 94 22 possible possible ADJ fbem-26173 94 23 start start NOUN fbem-26173 94 24 of of ADP fbem-26173 94 25 an an DET fbem-26173 94 26 uptrend uptrend NOUN fbem-26173 94 27 , , PUNCT fbem-26173 94 28 so so ADV fbem-26173 94 29 consider consider VERB fbem-26173 94 30 buying buying NOUN fbem-26173 94 31 ; ; PUNCT fbem-26173 94 32 when when SCONJ fbem-26173 94 33 the the DET fbem-26173 94 34 macd macd NOUN fbem-26173 94 35 line line NOUN fbem-26173 94 36 crosses cross VERB fbem-26173 94 37 the the DET fbem-26173 94 38 signal signal ADJ fbem-26173 94 39 line line NOUN fbem-26173 94 40 from from ADP fbem-26173 94 41 above above ADV fbem-26173 94 42 , , PUNCT fbem-26173 94 43 creating create VERB fbem-26173 94 44 a a DET fbem-26173 94 45 β€˜ ' PUNCT fbem-26173 94 46 dead dead ADJ fbem-26173 94 47 cross cross NOUN fbem-26173 94 48 ’ ' PUNCT fbem-26173 94 49 , , PUNCT fbem-26173 94 50 it it PRON fbem-26173 94 51 indicates indicate VERB fbem-26173 94 52 the the DET fbem-26173 94 53 possible possible ADJ fbem-26173 94 54 start start NOUN fbem-26173 94 55 of of ADP fbem-26173 94 56 a a DET fbem-26173 94 57 downtrend downtrend NOUN fbem-26173 94 58 , , PUNCT fbem-26173 94 59 so so ADV fbem-26173 94 60 consider consider VERB fbem-26173 94 61 selling sell VERB fbem-26173 94 62 . . PUNCT fbem-26173 95 1 3 3 X fbem-26173 95 2 . . X fbem-26173 95 3 empirical empirical ADJ fbem-26173 95 4 analysis analysis NOUN fbem-26173 95 5 3.1 3.1 NUM fbem-26173 95 6 . . PUNCT fbem-26173 95 7 data datum NOUN fbem-26173 95 8 sources source NOUN fbem-26173 95 9 and and CCONJ fbem-26173 95 10 selection selection NOUN fbem-26173 95 11 of of ADP fbem-26173 95 12 indicators indicator NOUN fbem-26173 95 13 this this DET fbem-26173 95 14 paper paper NOUN fbem-26173 95 15 collects collect VERB fbem-26173 95 16 the the DET fbem-26173 95 17 daily daily ADJ fbem-26173 95 18 frequency frequency NOUN fbem-26173 95 19 data datum NOUN fbem-26173 95 20 of of ADP fbem-26173 95 21 csi csi ADJ fbem-26173 95 22 300 300 NUM fbem-26173 95 23 constituents constituent NOUN fbem-26173 95 24 from from ADP fbem-26173 95 25 january january PROPN fbem-26173 95 26 2020 2020 NUM fbem-26173 95 27 to to ADP fbem-26173 95 28 december december PROPN fbem-26173 95 29 2022 2022 NUM fbem-26173 95 30 , , PUNCT fbem-26173 95 31 which which PRON fbem-26173 95 32 includes include VERB fbem-26173 95 33 the the DET fbem-26173 95 34 stock stock NOUN fbem-26173 95 35 market market NOUN fbem-26173 95 36 indicators indicator NOUN fbem-26173 95 37 and and CCONJ fbem-26173 95 38 fundamental fundamental ADJ fbem-26173 95 39 indicators indicator NOUN fbem-26173 95 40 , , PUNCT fbem-26173 95 41 which which PRON fbem-26173 95 42 are be AUX fbem-26173 95 43 gain gain NOUN fbem-26173 95 44 , , PUNCT fbem-26173 95 45 amplitude amplitude NOUN fbem-26173 95 46 , , PUNCT fbem-26173 95 47 yesterday yesterday NOUN fbem-26173 95 48 's 's PART fbem-26173 95 49 closing closing NOUN fbem-26173 95 50 price price NOUN fbem-26173 95 51 , , PUNCT fbem-26173 95 52 opening opening NOUN fbem-26173 95 53 price price NOUN fbem-26173 95 54 , , PUNCT fbem-26173 95 55 closing closing NOUN fbem-26173 95 56 price price NOUN fbem-26173 95 57 , , PUNCT fbem-26173 95 58 high high ADJ fbem-26173 95 59 price price NOUN fbem-26173 95 60 , , PUNCT fbem-26173 95 61 low low ADJ fbem-26173 95 62 price price NOUN fbem-26173 95 63 , , PUNCT fbem-26173 95 64 average average ADJ fbem-26173 95 65 price price NOUN fbem-26173 95 66 , , PUNCT fbem-26173 95 67 number number NOUN fbem-26173 95 68 of of ADP fbem-26173 95 69 shares share NOUN fbem-26173 95 70 traded trade VERB fbem-26173 95 71 , , PUNCT fbem-26173 95 72 turnover turnover NOUN fbem-26173 95 73 , , PUNCT fbem-26173 95 74 turnover turnover NOUN fbem-26173 95 75 , , PUNCT fbem-26173 95 76 and and CCONJ fbem-26173 95 77 the the DET fbem-26173 95 78 total total ADJ fbem-26173 95 79 market market NOUN fbem-26173 95 80 capitalisation capitalisation NOUN fbem-26173 95 81 , , PUNCT fbem-26173 95 82 market market NOUN fbem-26173 95 83 capitalisation capitalisation NOUN fbem-26173 95 84 in in ADP fbem-26173 95 85 circulation circulation NOUN fbem-26173 95 86 , , PUNCT fbem-26173 95 87 total total ADJ fbem-26173 95 88 share share NOUN fbem-26173 95 89 capital capital NOUN fbem-26173 95 90 , , PUNCT fbem-26173 95 91 outstanding outstanding ADJ fbem-26173 95 92 share share NOUN fbem-26173 95 93 capital capital NOUN fbem-26173 95 94 , , PUNCT fbem-26173 95 95 price price NOUN fbem-26173 95 96 - - PUNCT fbem-26173 95 97 earnings earning NOUN fbem-26173 95 98 ratio ratio NOUN fbem-26173 95 99 , , PUNCT fbem-26173 95 100 and and CCONJ fbem-26173 95 101 price price NOUN fbem-26173 95 102 - - PUNCT fbem-26173 95 103 to to ADP fbem-26173 95 104 - - PUNCT fbem-26173 95 105 capitalisation capitalisation NOUN fbem-26173 95 106 ratio ratio NOUN fbem-26173 95 107 , , PUNCT fbem-26173 95 108 etc etc X fbem-26173 95 109 . . X fbem-26173 96 1 this this DET fbem-26173 96 2 paper paper NOUN fbem-26173 96 3 gives give VERB fbem-26173 96 4 17 17 NUM fbem-26173 96 5 indicators indicator NOUN fbem-26173 96 6 that that PRON fbem-26173 96 7 can can AUX fbem-26173 96 8 evaluate evaluate VERB fbem-26173 96 9 both both PRON fbem-26173 96 10 the the DET fbem-26173 96 11 value value NOUN fbem-26173 96 12 and and CCONJ fbem-26173 96 13 the the DET fbem-26173 96 14 risk risk NOUN fbem-26173 96 15 of of ADP fbem-26173 96 16 a a DET fbem-26173 96 17 stock stock NOUN fbem-26173 96 18 , , PUNCT fbem-26173 96 19 which which PRON fbem-26173 96 20 are be AUX fbem-26173 96 21 mainly mainly ADV fbem-26173 96 22 used use VERB fbem-26173 96 23 to to PART fbem-26173 96 24 reflect reflect VERB fbem-26173 96 25 the the DET fbem-26173 96 26 change change NOUN fbem-26173 96 27 of of ADP fbem-26173 96 28 the the DET fbem-26173 96 29 price price NOUN fbem-26173 96 30 and and CCONJ fbem-26173 96 31 the the DET fbem-26173 96 32 market market NOUN fbem-26173 96 33 activity activity NOUN fbem-26173 96 34 , , PUNCT fbem-26173 96 35 as as ADV fbem-26173 96 36 well well ADV fbem-26173 96 37 as as ADP fbem-26173 96 38 to to PART fbem-26173 96 39 assess assess VERB fbem-26173 96 40 the the DET fbem-26173 96 41 investment investment NOUN fbem-26173 96 42 value value NOUN fbem-26173 96 43 of of ADP fbem-26173 96 44 a a DET fbem-26173 96 45 stock stock NOUN fbem-26173 96 46 . . PUNCT fbem-26173 97 1 this this DET fbem-26173 97 2 paper paper NOUN fbem-26173 97 3 gives give VERB fbem-26173 97 4 data datum NOUN fbem-26173 97 5 of of ADP fbem-26173 97 6 17 17 NUM fbem-26173 97 7 indicators indicator NOUN fbem-26173 97 8 capable capable ADJ fbem-26173 97 9 of of ADP fbem-26173 97 10 evaluating evaluate VERB fbem-26173 97 11 both both CCONJ fbem-26173 97 12 the the DET fbem-26173 97 13 value value NOUN fbem-26173 97 14 and and CCONJ fbem-26173 97 15 risk risk NOUN fbem-26173 97 16 of of ADP fbem-26173 97 17 a a DET fbem-26173 97 18 stock stock NOUN fbem-26173 97 19 , , PUNCT fbem-26173 97 20 which which PRON fbem-26173 97 21 are be AUX fbem-26173 97 22 mainly mainly ADV fbem-26173 97 23 used use VERB fbem-26173 97 24 to to PART fbem-26173 97 25 reflect reflect VERB fbem-26173 97 26 the the DET fbem-26173 97 27 changes change NOUN fbem-26173 97 28 in in ADP fbem-26173 97 29 stock stock NOUN fbem-26173 97 30 prices price NOUN fbem-26173 97 31 and and CCONJ fbem-26173 97 32 the the DET fbem-26173 97 33 activity activity NOUN fbem-26173 97 34 of of ADP fbem-26173 97 35 the the DET fbem-26173 97 36 market market NOUN fbem-26173 97 37 , , PUNCT fbem-26173 97 38 as as ADV fbem-26173 97 39 well well ADV fbem-26173 97 40 as as ADP fbem-26173 97 41 to to PART fbem-26173 97 42 assess assess VERB fbem-26173 97 43 the the DET fbem-26173 97 44 investment investment NOUN fbem-26173 97 45 value value NOUN fbem-26173 97 46 of of ADP fbem-26173 97 47 a a DET fbem-26173 97 48 stock stock NOUN fbem-26173 97 49 . . PUNCT fbem-26173 98 1 yang yang PROPN fbem-26173 98 2 et et PROPN fbem-26173 98 3 al al PROPN fbem-26173 98 4 . . PROPN fbem-26173 99 1 integrated integrate VERB fbem-26173 99 2 three three NUM fbem-26173 99 3 types type NOUN fbem-26173 99 4 of of ADP fbem-26173 99 5 factors factor NOUN fbem-26173 99 6 , , PUNCT fbem-26173 99 7 which which PRON fbem-26173 99 8 are be AUX fbem-26173 99 9 the the DET fbem-26173 99 10 price price NOUN fbem-26173 99 11 factor factor NOUN fbem-26173 99 12 , , PUNCT fbem-26173 99 13 the the DET fbem-26173 99 14 hard hard ADV fbem-26173 99 15 - - PUNCT fbem-26173 99 16 coded code VERB fbem-26173 99 17 factor factor NOUN fbem-26173 99 18 , , PUNCT fbem-26173 99 19 and and CCONJ fbem-26173 99 20 the the DET fbem-26173 99 21 factor factor NOUN fbem-26173 99 22 based base VERB fbem-26173 99 23 on on ADP fbem-26173 99 24 the the DET fbem-26173 99 25 rolling rolling ADJ fbem-26173 99 26 arithmetic arithmetic NOUN fbem-26173 100 1 [ [ X fbem-26173 100 2 20 20 NUM fbem-26173 100 3 ] ] PUNCT fbem-26173 100 4 . . PUNCT fbem-26173 101 1 in in ADP fbem-26173 101 2 addition addition NOUN fbem-26173 101 3 to to ADP fbem-26173 101 4 these these DET fbem-26173 101 5 three three NUM fbem-26173 101 6 types type NOUN fbem-26173 101 7 of of ADP fbem-26173 101 8 factors factor NOUN fbem-26173 101 9 , , PUNCT fbem-26173 101 10 based base VERB fbem-26173 101 11 on on ADP fbem-26173 101 12 the the DET fbem-26173 101 13 original original ADJ fbem-26173 101 14 data datum NOUN fbem-26173 101 15 , , PUNCT fbem-26173 101 16 this this DET fbem-26173 101 17 paper paper NOUN fbem-26173 101 18 constructs construct VERB fbem-26173 101 19 the the DET fbem-26173 101 20 sentiment sentiment NOUN fbem-26173 101 21 indicator indicator NOUN fbem-26173 101 22 arbr arbr PROPN fbem-26173 101 23 , , PUNCT fbem-26173 101 24 which which PRON fbem-26173 101 25 measures measure VERB fbem-26173 101 26 the the DET fbem-26173 101 27 power power NOUN fbem-26173 101 28 of of ADP fbem-26173 101 29 buyers buyer NOUN fbem-26173 101 30 and and CCONJ fbem-26173 101 31 sellers seller NOUN fbem-26173 101 32 in in ADP fbem-26173 101 33 the the DET fbem-26173 101 34 market market NOUN fbem-26173 101 35 , , PUNCT fbem-26173 101 36 and and CCONJ fbem-26173 101 37 some some DET fbem-26173 101 38 commonly commonly ADV fbem-26173 101 39 used use VERB fbem-26173 101 40 technical technical ADJ fbem-26173 101 41 indicators indicator NOUN fbem-26173 101 42 , , PUNCT fbem-26173 101 43 such such ADJ fbem-26173 101 44 as as ADP fbem-26173 101 45 moving move VERB fbem-26173 101 46 averages average NOUN fbem-26173 101 47 and and CCONJ fbem-26173 101 48 deviation deviation NOUN fbem-26173 101 49 ratios ratio NOUN fbem-26173 101 50 . . PUNCT fbem-26173 102 1 the the DET fbem-26173 102 2 types type NOUN fbem-26173 102 3 of of ADP fbem-26173 102 4 the the DET fbem-26173 102 5 above above ADJ fbem-26173 102 6 factors factor NOUN fbem-26173 102 7 are be AUX fbem-26173 102 8 shown show VERB fbem-26173 102 9 in in ADP fbem-26173 102 10 table table NOUN fbem-26173 102 11 1 1 NUM fbem-26173 102 12 . . X fbem-26173 102 13 79 79 NUM fbem-26173 102 14 table table NOUN fbem-26173 102 15 1 1 NUM fbem-26173 102 16 . . PUNCT fbem-26173 102 17 factor factor NOUN fbem-26173 102 18 description description NOUN fbem-26173 102 19 table table NOUN fbem-26173 102 20 style style NOUN fbem-26173 102 21 name name NOUN fbem-26173 102 22 formulas formula NOUN fbem-26173 102 23 hidden hide VERB fbem-26173 102 24 meaning meaning NOUN fbem-26173 102 25 price price NOUN fbem-26173 102 26 factor factor NOUN fbem-26173 102 27 open0 open0 NOUN fbem-26173 102 28 open open ADJ fbem-26173 102 29 / / SYM fbem-26173 102 30 close close ADJ fbem-26173 102 31 percentage percentage NOUN fbem-26173 102 32 of of ADP fbem-26173 102 33 opening opening NOUN fbem-26173 102 34 price price NOUN fbem-26173 102 35 relative relative ADJ fbem-26173 102 36 to to ADP fbem-26173 102 37 closing closing NOUN fbem-26173 102 38 price price NOUN fbem-26173 102 39 high0 high0 NOUN fbem-26173 102 40 high high ADJ fbem-26173 102 41 / / SYM fbem-26173 102 42 close close ADJ fbem-26173 102 43 highest high ADJ fbem-26173 102 44 price price NOUN fbem-26173 102 45 as as ADP fbem-26173 102 46 a a DET fbem-26173 102 47 percentage percentage NOUN fbem-26173 102 48 of of ADP fbem-26173 102 49 closing closing NOUN fbem-26173 102 50 price price NOUN fbem-26173 102 51 hard hard ADJ fbem-26173 102 52 coding code VERB fbem-26173 102 53 factor factor NOUN fbem-26173 102 54 kmid kmid NOUN fbem-26173 102 55 ( ( PUNCT fbem-26173 102 56 close close VERB fbem-26173 102 57 - - PUNCT fbem-26173 102 58 open)/open open)/open NOUN fbem-26173 102 59 percentage percentage NOUN fbem-26173 102 60 increase increase NOUN fbem-26173 102 61 ( ( PUNCT fbem-26173 102 62 decrease decrease NOUN fbem-26173 102 63 ) ) PUNCT fbem-26173 102 64 of of ADP fbem-26173 102 65 closing closing NOUN fbem-26173 102 66 price price NOUN fbem-26173 102 67 over over ADP fbem-26173 102 68 opening opening NOUN fbem-26173 102 69 price price NOUN fbem-26173 102 70 ksft ksft NOUN fbem-26173 102 71 ( ( PUNCT fbem-26173 102 72 2*close 2*close NUM fbem-26173 102 73 - - PUNCT fbem-26173 102 74 high high ADJ fbem-26173 102 75 - - PUNCT fbem-26173 102 76 low)/open low)/open NOUN fbem-26173 102 77 percentage percentage NOUN fbem-26173 102 78 of of ADP fbem-26173 102 79 the the DET fbem-26173 102 80 difference difference NOUN fbem-26173 102 81 between between ADP fbem-26173 102 82 the the DET fbem-26173 102 83 closing closing NOUN fbem-26173 102 84 price price NOUN fbem-26173 102 85 and and CCONJ fbem-26173 102 86 the the DET fbem-26173 102 87 high high ADJ fbem-26173 102 88 and and CCONJ fbem-26173 102 89 low low ADJ fbem-26173 102 90 prices price NOUN fbem-26173 102 91 relative relative ADJ fbem-26173 102 92 to to ADP fbem-26173 102 93 the the DET fbem-26173 102 94 opening opening NOUN fbem-26173 102 95 price price NOUN fbem-26173 102 96 factorization factorization NOUN fbem-26173 102 97 based base VERB fbem-26173 102 98 on on ADP fbem-26173 102 99 rolling roll VERB fbem-26173 102 100 operators operator NOUN fbem-26173 102 101 roc5 roc5 VERB fbem-26173 102 102 ref ref NOUN fbem-26173 102 103 ( ( PUNCT fbem-26173 102 104 close close VERB fbem-26173 102 105 , , PUNCT fbem-26173 102 106 5)/close 5)/close NUM fbem-26173 102 107 ratio ratio NOUN fbem-26173 102 108 of of ADP fbem-26173 102 109 the the DET fbem-26173 102 110 closing closing NOUN fbem-26173 102 111 price price NOUN fbem-26173 102 112 on on ADP fbem-26173 102 113 day day NOUN fbem-26173 102 114 t-5 t-5 NOUN fbem-26173 102 115 to to ADP fbem-26173 102 116 the the DET fbem-26173 102 117 closing closing NOUN fbem-26173 102 118 price price NOUN fbem-26173 102 119 on on ADP fbem-26173 102 120 that that DET fbem-26173 102 121 day day NOUN fbem-26173 102 122 roc10 roc10 ADJ fbem-26173 102 123 ref ref NOUN fbem-26173 102 124 ( ( PUNCT fbem-26173 102 125 close close ADJ fbem-26173 102 126 , , PUNCT fbem-26173 102 127 10)/close 10)/close NUM fbem-26173 102 128 ratio ratio NOUN fbem-26173 102 129 of of ADP fbem-26173 102 130 closing closing NOUN fbem-26173 102 131 price price NOUN fbem-26173 102 132 on on ADP fbem-26173 102 133 day day NOUN fbem-26173 102 134 t-10 t-10 PUNCT fbem-26173 102 135 to to ADP fbem-26173 102 136 closing close VERB fbem-26173 102 137 price price NOUN fbem-26173 102 138 on on ADP fbem-26173 102 139 that that DET fbem-26173 102 140 day day NOUN fbem-26173 102 141 technical technical ADJ fbem-26173 102 142 factor factor NOUN fbem-26173 102 143 5d_ma 5d_ma NUM fbem-26173 102 144 sum sum NOUN fbem-26173 102 145 of of ADP fbem-26173 102 146 closing closing NOUN fbem-26173 102 147 prices price NOUN fbem-26173 102 148 of of ADP fbem-26173 102 149 the the DET fbem-26173 102 150 previous previous ADJ fbem-26173 102 151 5 5 NUM fbem-26173 102 152 days/5 days/5 NOUN fbem-26173 102 153 average average NOUN fbem-26173 102 154 of of ADP fbem-26173 102 155 the the DET fbem-26173 102 156 closing close VERB fbem-26173 102 157 prices price NOUN fbem-26173 102 158 of of ADP fbem-26173 102 159 the the DET fbem-26173 102 160 previous previous ADJ fbem-26173 102 161 5 5 NUM fbem-26173 102 162 days day NOUN fbem-26173 102 163 bias bias NOUN fbem-26173 102 164 ( ( PUNCT fbem-26173 102 165 closing closing NOUN fbem-26173 102 166 price price NOUN fbem-26173 102 167 5d_ma)/5d_ma 5d_ma)/5d_ma NUM fbem-26173 102 168 percentage percentage NOUN fbem-26173 102 169 of of ADP fbem-26173 102 170 difference difference NOUN fbem-26173 102 171 between between ADP fbem-26173 102 172 closing closing NOUN fbem-26173 102 173 price price NOUN fbem-26173 102 174 and and CCONJ fbem-26173 102 175 5 5 NUM fbem-26173 102 176 - - PUNCT fbem-26173 102 177 day day NOUN fbem-26173 102 178 moving move VERB fbem-26173 102 179 average average ADJ fbem-26173 102 180 emotional emotional ADJ fbem-26173 102 181 factor factor NOUN fbem-26173 102 182 arbr arbr NOUN fbem-26173 102 183 ( ( PUNCT fbem-26173 102 184 high high ADJ fbem-26173 102 185 - - PUNCT fbem-26173 102 186 open)/(open open)/(open ADJ fbem-26173 102 187 - - PUNCT fbem-26173 102 188 low)-(highlastclose)/(lastclose low)-(highlastclose)/(lastclose ADJ fbem-26173 102 189 - - PUNCT fbem-26173 102 190 low low ADJ fbem-26173 102 191 ) ) PUNCT fbem-26173 102 192 difference difference NOUN fbem-26173 102 193 between between ADP fbem-26173 102 194 popularity popularity NOUN fbem-26173 102 195 indicator indicator NOUN fbem-26173 102 196 ar ar PROPN fbem-26173 102 197 and and CCONJ fbem-26173 102 198 willingness willingness PROPN fbem-26173 102 199 indicator indicator NOUN fbem-26173 102 200 br br NOUN fbem-26173 102 201 including include VERB fbem-26173 102 202 the the DET fbem-26173 102 203 original original ADJ fbem-26173 102 204 indicators indicator NOUN fbem-26173 102 205 , , PUNCT fbem-26173 102 206 there there PRON fbem-26173 102 207 are be VERB fbem-26173 102 208 25 25 NUM fbem-26173 102 209 factors factor NOUN fbem-26173 102 210 in in ADP fbem-26173 102 211 this this DET fbem-26173 102 212 paper paper NOUN fbem-26173 102 213 , , PUNCT fbem-26173 102 214 and and CCONJ fbem-26173 102 215 it it PRON fbem-26173 102 216 is be AUX fbem-26173 102 217 found find VERB fbem-26173 102 218 by by ADP fbem-26173 102 219 correlation correlation NOUN fbem-26173 102 220 analysis analysis NOUN fbem-26173 102 221 that that SCONJ fbem-26173 102 222 many many ADJ fbem-26173 102 223 feature feature NOUN fbem-26173 102 224 factors factor NOUN fbem-26173 102 225 are be AUX fbem-26173 102 226 highly highly ADV fbem-26173 102 227 correlated correlate VERB fbem-26173 102 228 , , PUNCT fbem-26173 102 229 i.e. i.e. X fbem-26173 102 230 , , PUNCT fbem-26173 102 231 generating generate VERB fbem-26173 102 232 the the DET fbem-26173 102 233 problem problem NOUN fbem-26173 102 234 of of ADP fbem-26173 102 235 multiple multiple ADJ fbem-26173 102 236 covariance covariance NOUN fbem-26173 102 237 , , PUNCT fbem-26173 102 238 which which PRON fbem-26173 102 239 will will AUX fbem-26173 102 240 make make VERB fbem-26173 102 241 the the DET fbem-26173 102 242 model model NOUN fbem-26173 102 243 unstable unstable ADJ fbem-26173 102 244 and and CCONJ fbem-26173 102 245 the the DET fbem-26173 102 246 reliability reliability NOUN fbem-26173 102 247 of of ADP fbem-26173 102 248 the the DET fbem-26173 102 249 prediction prediction NOUN fbem-26173 102 250 results result NOUN fbem-26173 102 251 reduced reduce VERB fbem-26173 102 252 . . PUNCT fbem-26173 103 1 so so ADV fbem-26173 103 2 , , PUNCT fbem-26173 103 3 in in ADP fbem-26173 103 4 this this DET fbem-26173 103 5 paper paper NOUN fbem-26173 103 6 , , PUNCT fbem-26173 103 7 after after ADP fbem-26173 103 8 feature feature NOUN fbem-26173 103 9 extraction extraction NOUN fbem-26173 103 10 , , PUNCT fbem-26173 103 11 the the DET fbem-26173 103 12 number number NOUN fbem-26173 103 13 of of ADP fbem-26173 103 14 redundant redundant ADJ fbem-26173 103 15 features feature NOUN fbem-26173 103 16 is be AUX fbem-26173 103 17 reduced reduce VERB fbem-26173 103 18 by by ADP fbem-26173 103 19 feature feature NOUN fbem-26173 103 20 selection selection NOUN fbem-26173 103 21 , , PUNCT fbem-26173 103 22 and and CCONJ fbem-26173 103 23 the the DET fbem-26173 103 24 correlation correlation NOUN fbem-26173 103 25 coefficient coefficient NOUN fbem-26173 103 26 of of ADP fbem-26173 103 27 0.9 0.9 NUM fbem-26173 103 28 is be AUX fbem-26173 103 29 used use VERB fbem-26173 103 30 as as ADP fbem-26173 103 31 the the DET fbem-26173 103 32 threshold threshold NOUN fbem-26173 103 33 to to PART fbem-26173 103 34 remove remove VERB fbem-26173 103 35 some some DET fbem-26173 103 36 features feature NOUN fbem-26173 103 37 . . PUNCT fbem-26173 104 1 after after ADP fbem-26173 104 2 screening screen VERB fbem-26173 104 3 , , PUNCT fbem-26173 104 4 a a DET fbem-26173 104 5 total total NOUN fbem-26173 104 6 of of ADP fbem-26173 104 7 8 8 NUM fbem-26173 104 8 factors factor NOUN fbem-26173 104 9 are be AUX fbem-26173 104 10 removed remove VERB fbem-26173 104 11 as as SCONJ fbem-26173 104 12 yesterday yesterday NOUN fbem-26173 104 13 's 's PART fbem-26173 104 14 closing closing NOUN fbem-26173 104 15 price price NOUN fbem-26173 104 16 , , PUNCT fbem-26173 104 17 closing closing NOUN fbem-26173 104 18 price price NOUN fbem-26173 104 19 , , PUNCT fbem-26173 104 20 high high ADJ fbem-26173 104 21 price price NOUN fbem-26173 104 22 , , PUNCT fbem-26173 104 23 low low ADJ fbem-26173 104 24 price price NOUN fbem-26173 104 25 , , PUNCT fbem-26173 104 26 average average ADJ fbem-26173 104 27 price price NOUN fbem-26173 104 28 , , PUNCT fbem-26173 104 29 total total ADJ fbem-26173 104 30 equity equity NOUN fbem-26173 104 31 , , PUNCT fbem-26173 104 32 kmid kmid PROPN fbem-26173 104 33 , , PUNCT fbem-26173 104 34 5d_ma 5d_ma NUM fbem-26173 104 35 , , PUNCT fbem-26173 104 36 and and CCONJ fbem-26173 104 37 the the DET fbem-26173 104 38 remaining remain VERB fbem-26173 104 39 17 17 NUM fbem-26173 104 40 factors factor NOUN fbem-26173 104 41 are be AUX fbem-26173 104 42 used use VERB fbem-26173 104 43 as as ADP fbem-26173 104 44 candidate candidate NOUN fbem-26173 104 45 factors factor NOUN fbem-26173 104 46 . . PUNCT fbem-26173 105 1 3.2 3.2 NUM fbem-26173 105 2 . . PUNCT fbem-26173 105 3 data datum NOUN fbem-26173 105 4 preprocessing preprocesse VERB fbem-26173 105 5 3.2.1 3.2.1 NUM fbem-26173 105 6 . . PUNCT fbem-26173 106 1 data datum NOUN fbem-26173 106 2 normalisation normalisation NOUN fbem-26173 106 3 data datum NOUN fbem-26173 106 4 dimensionlessness dimensionlessness NOUN fbem-26173 106 5 can can AUX fbem-26173 106 6 eliminate eliminate VERB fbem-26173 106 7 the the DET fbem-26173 106 8 incomparability incomparability NOUN fbem-26173 106 9 between between ADP fbem-26173 106 10 feature feature NOUN fbem-26173 106 11 variables variable NOUN fbem-26173 106 12 , , PUNCT fbem-26173 106 13 which which PRON fbem-26173 106 14 usually usually ADV fbem-26173 106 15 includes include VERB fbem-26173 106 16 z z NOUN fbem-26173 106 17 - - PUNCT fbem-26173 106 18 score score NOUN fbem-26173 106 19 normalization normalization NOUN fbem-26173 106 20 and and CCONJ fbem-26173 106 21 min min PROPN fbem-26173 106 22 - - ADJ fbem-26173 106 23 max max PROPN fbem-26173 106 24 normalization normalization NOUN fbem-26173 106 25 , , PUNCT fbem-26173 106 26 where where SCONJ fbem-26173 106 27 normalization normalization NOUN fbem-26173 106 28 processing processing NOUN fbem-26173 106 29 is be AUX fbem-26173 106 30 also also ADV fbem-26173 106 31 a a DET fbem-26173 106 32 frequently frequently ADV fbem-26173 106 33 used use VERB fbem-26173 106 34 method method NOUN fbem-26173 106 35 in in ADP fbem-26173 106 36 neural neural ADJ fbem-26173 106 37 networks network NOUN fbem-26173 106 38 to to PART fbem-26173 106 39 improve improve VERB fbem-26173 106 40 the the DET fbem-26173 106 41 model model NOUN fbem-26173 106 42 prediction prediction NOUN fbem-26173 106 43 performance performance NOUN fbem-26173 106 44 . . PUNCT fbem-26173 107 1 from from ADP fbem-26173 107 2 the the DET fbem-26173 107 3 data data NOUN fbem-26173 107 4 set set VERB fbem-26173 107 5 can can AUX fbem-26173 107 6 be be AUX fbem-26173 107 7 seen see VERB fbem-26173 107 8 that that SCONJ fbem-26173 107 9 the the DET fbem-26173 107 10 number number NOUN fbem-26173 107 11 of of ADP fbem-26173 107 12 traded trade VERB fbem-26173 107 13 shares share NOUN fbem-26173 107 14 and and CCONJ fbem-26173 107 15 other other ADJ fbem-26173 107 16 variables variable NOUN fbem-26173 107 17 are be AUX fbem-26173 107 18 of of ADP fbem-26173 107 19 large large ADJ fbem-26173 107 20 magnitude magnitude NOUN fbem-26173 107 21 , , PUNCT fbem-26173 107 22 so so CCONJ fbem-26173 107 23 before before ADP fbem-26173 107 24 the the DET fbem-26173 107 25 data data NOUN fbem-26173 107 26 analysis analysis NOUN fbem-26173 107 27 , , PUNCT fbem-26173 107 28 this this DET fbem-26173 107 29 paper paper NOUN fbem-26173 107 30 normalizes normalize VERB fbem-26173 107 31 the the DET fbem-26173 107 32 sample sample NOUN fbem-26173 107 33 set set NOUN fbem-26173 107 34 , , PUNCT fbem-26173 107 35 and and CCONJ fbem-26173 107 36 under under ADP fbem-26173 107 37 the the DET fbem-26173 107 38 premise premise NOUN fbem-26173 107 39 of of ADP fbem-26173 107 40 keeping keep VERB fbem-26173 107 41 the the DET fbem-26173 107 42 data datum NOUN fbem-26173 107 43 distribution distribution NOUN fbem-26173 107 44 unchanged unchanged ADJ fbem-26173 107 45 , , PUNCT fbem-26173 107 46 the the DET fbem-26173 107 47 range range NOUN fbem-26173 107 48 is be AUX fbem-26173 107 49 scaled scale VERB fbem-26173 107 50 to to ADP fbem-26173 107 51 [ [ X fbem-26173 107 52 0 0 NUM fbem-26173 107 53 , , PUNCT fbem-26173 107 54 1 1 NUM fbem-26173 107 55 ] ] PUNCT fbem-26173 107 56 , , PUNCT fbem-26173 107 57 which which PRON fbem-26173 107 58 is be AUX fbem-26173 107 59 expressed express VERB fbem-26173 107 60 by by ADP fbem-26173 107 61 the the DET fbem-26173 107 62 formula formula NOUN fbem-26173 107 63 : : PUNCT fbem-26173 107 64 π‘₯β€² π‘₯β€² PUNCT fbem-26173 108 1 = = X fbem-26173 108 2 π‘₯βˆ’min π‘₯βˆ’min NOUN fbem-26173 108 3 ( ( PUNCT fbem-26173 108 4 π‘₯ π‘₯ NOUN fbem-26173 108 5 ) ) PUNCT fbem-26173 108 6 max(π‘₯)βˆ’min max(π‘₯)βˆ’min NOUN fbem-26173 108 7 ( ( PUNCT fbem-26173 108 8 π‘₯ π‘₯ NOUN fbem-26173 108 9 ) ) PUNCT fbem-26173 108 10 ( ( PUNCT fbem-26173 108 11 13 13 NUM fbem-26173 108 12 ) ) PUNCT fbem-26173 108 13 3.2.2 3.2.2 NUM fbem-26173 108 14 . . PUNCT fbem-26173 108 15 random random ADJ fbem-26173 108 16 forest forest NOUN fbem-26173 108 17 selection selection NOUN fbem-26173 108 18 factor factor NOUN fbem-26173 108 19 results result VERB fbem-26173 108 20 random random ADJ fbem-26173 108 21 forest forest NOUN fbem-26173 108 22 is be AUX fbem-26173 108 23 a a DET fbem-26173 108 24 machine machine NOUN fbem-26173 108 25 learning learn VERB fbem-26173 108 26 algorithm algorithm NOUN fbem-26173 108 27 based base VERB fbem-26173 108 28 on on ADP fbem-26173 108 29 decision decision NOUN fbem-26173 108 30 trees tree NOUN fbem-26173 108 31 , , PUNCT fbem-26173 108 32 which which PRON fbem-26173 108 33 can can AUX fbem-26173 108 34 select select VERB fbem-26173 108 35 the the DET fbem-26173 108 36 most most ADV fbem-26173 108 37 important important ADJ fbem-26173 108 38 feature feature NOUN fbem-26173 108 39 variables variable NOUN fbem-26173 108 40 for for ADP fbem-26173 108 41 regression regression NOUN fbem-26173 108 42 prediction prediction NOUN fbem-26173 108 43 among among ADP fbem-26173 108 44 a a DET fbem-26173 108 45 large large ADJ fbem-26173 108 46 number number NOUN fbem-26173 108 47 of of ADP fbem-26173 108 48 features feature NOUN fbem-26173 108 49 , , PUNCT fbem-26173 108 50 and and CCONJ fbem-26173 108 51 rank rank VERB fbem-26173 108 52 the the DET fbem-26173 108 53 importance importance NOUN fbem-26173 108 54 of of ADP fbem-26173 108 55 each each DET fbem-26173 108 56 feature feature NOUN fbem-26173 108 57 , , PUNCT fbem-26173 108 58 so so SCONJ fbem-26173 108 59 as as SCONJ fbem-26173 108 60 to to PART fbem-26173 108 61 achieve achieve VERB fbem-26173 108 62 the the DET fbem-26173 108 63 effect effect NOUN fbem-26173 108 64 of of ADP fbem-26173 108 65 dimensionality dimensionality NOUN fbem-26173 108 66 reduction reduction NOUN fbem-26173 108 67 and and CCONJ fbem-26173 108 68 eliminate eliminate VERB fbem-26173 108 69 redundant redundant ADJ fbem-26173 108 70 features feature NOUN fbem-26173 108 71 , , PUNCT fbem-26173 108 72 which which PRON fbem-26173 108 73 helps help VERB fbem-26173 108 74 to to PART fbem-26173 108 75 improve improve VERB fbem-26173 108 76 the the DET fbem-26173 108 77 performance performance NOUN fbem-26173 108 78 of of ADP fbem-26173 108 79 the the DET fbem-26173 108 80 learning learning NOUN fbem-26173 108 81 algorithm algorithm NOUN fbem-26173 108 82 . . PUNCT fbem-26173 109 1 figure figure NOUN fbem-26173 109 2 2 2 NUM fbem-26173 109 3 is be AUX fbem-26173 109 4 based base VERB fbem-26173 109 5 on on ADP fbem-26173 109 6 the the DET fbem-26173 109 7 importance importance NOUN fbem-26173 109 8 score score NOUN fbem-26173 109 9 of of ADP fbem-26173 109 10 each each DET fbem-26173 109 11 factor factor NOUN fbem-26173 109 12 of of ADP fbem-26173 109 13 the the DET fbem-26173 109 14 random random ADJ fbem-26173 109 15 forest forest NOUN fbem-26173 109 16 , , PUNCT fbem-26173 109 17 the the PRON fbem-26173 109 18 larger large ADJ fbem-26173 109 19 the the DET fbem-26173 109 20 score score NOUN fbem-26173 109 21 value value NOUN fbem-26173 109 22 , , PUNCT fbem-26173 109 23 the the PRON fbem-26173 109 24 higher high ADJ fbem-26173 109 25 the the DET fbem-26173 109 26 degree degree NOUN fbem-26173 109 27 of of ADP fbem-26173 109 28 importance importance NOUN fbem-26173 109 29 of of ADP fbem-26173 109 30 the the DET fbem-26173 109 31 influence influence NOUN fbem-26173 109 32 factor factor NOUN fbem-26173 109 33 , , PUNCT fbem-26173 109 34 it it PRON fbem-26173 109 35 can can AUX fbem-26173 109 36 be be AUX fbem-26173 109 37 seen see VERB fbem-26173 109 38 that that SCONJ fbem-26173 109 39 the the DET fbem-26173 109 40 closing closing NOUN fbem-26173 109 41 price price NOUN fbem-26173 109 42 and and CCONJ fbem-26173 109 43 yesterday yesterday NOUN fbem-26173 109 44 's 's PART fbem-26173 109 45 closing closing NOUN fbem-26173 109 46 price price NOUN fbem-26173 109 47 has have VERB fbem-26173 109 48 the the DET fbem-26173 109 49 greatest great ADJ fbem-26173 109 50 impact impact NOUN fbem-26173 109 51 on on ADP fbem-26173 109 52 the the DET fbem-26173 109 53 stock stock NOUN fbem-26173 109 54 gain gain NOUN fbem-26173 109 55 . . PUNCT fbem-26173 110 1 in in ADP fbem-26173 110 2 this this DET fbem-26173 110 3 paper paper NOUN fbem-26173 110 4 , , PUNCT fbem-26173 110 5 we we PRON fbem-26173 110 6 take take VERB fbem-26173 110 7 the the DET fbem-26173 110 8 indexes index NOUN fbem-26173 110 9 ranked rank VERB fbem-26173 110 10 in in ADP fbem-26173 110 11 the the DET fbem-26173 110 12 top top ADJ fbem-26173 110 13 10 10 NUM fbem-26173 110 14 as as ADP fbem-26173 110 15 the the DET fbem-26173 110 16 influence influence NOUN fbem-26173 110 17 factors factor NOUN fbem-26173 110 18 of of ADP fbem-26173 110 19 the the DET fbem-26173 110 20 rate rate NOUN fbem-26173 110 21 of of ADP fbem-26173 110 22 increase increase NOUN fbem-26173 110 23 , , PUNCT fbem-26173 110 24 which which PRON fbem-26173 110 25 are be AUX fbem-26173 110 26 used use VERB fbem-26173 110 27 in in ADP fbem-26173 110 28 the the DET fbem-26173 110 29 construction construction NOUN fbem-26173 110 30 of of ADP fbem-26173 110 31 the the DET fbem-26173 110 32 next next ADJ fbem-26173 110 33 prediction prediction NOUN fbem-26173 110 34 model model NOUN fbem-26173 110 35 . . PUNCT fbem-26173 111 1 figure figure NOUN fbem-26173 111 2 2 2 NUM fbem-26173 111 3 . . PUNCT fbem-26173 111 4 factor factor NOUN fbem-26173 111 5 importance importance NOUN fbem-26173 111 6 score score NOUN fbem-26173 111 7 3.2.3 3.2.3 NUM fbem-26173 111 8 . . PUNCT fbem-26173 112 1 xgboost xgboost PROPN fbem-26173 113 1 stock stock NOUN fbem-26173 113 2 picks pick VERB fbem-26173 113 3 results result NOUN fbem-26173 113 4 in in ADP fbem-26173 113 5 the the DET fbem-26173 113 6 above above ADJ fbem-26173 113 7 10 10 NUM fbem-26173 113 8 influencing influence VERB fbem-26173 113 9 factors factor NOUN fbem-26173 113 10 of of ADP fbem-26173 113 11 the the DET fbem-26173 113 12 increase increase NOUN fbem-26173 113 13 were be AUX fbem-26173 113 14 determined determine VERB fbem-26173 113 15 by by ADP fbem-26173 113 16 random random ADJ fbem-26173 113 17 forest forest NOUN fbem-26173 113 18 regression regression NOUN fbem-26173 113 19 , , PUNCT fbem-26173 113 20 excluding exclude VERB fbem-26173 113 21 variables variable NOUN fbem-26173 113 22 such such ADJ fbem-26173 113 23 as as ADP fbem-26173 113 24 market market NOUN fbem-26173 113 25 capitalisation capitalisation NOUN fbem-26173 113 26 , , PUNCT fbem-26173 113 27 outstanding outstanding ADJ fbem-26173 113 28 share share NOUN fbem-26173 113 29 capital capital NOUN fbem-26173 113 30 , , PUNCT fbem-26173 113 31 opening opening NOUN fbem-26173 113 32 price price NOUN fbem-26173 113 33 , , PUNCT fbem-26173 113 34 total total ADJ fbem-26173 113 35 market market NOUN fbem-26173 113 36 capitalisation capitalisation NOUN fbem-26173 113 37 , , PUNCT fbem-26173 113 38 p p X fbem-26173 113 39 / / SYM fbem-26173 113 40 b b NOUN fbem-26173 113 41 ratio ratio NOUN fbem-26173 113 42 , , PUNCT fbem-26173 113 43 number number NOUN fbem-26173 113 44 of of ADP fbem-26173 113 45 shares share NOUN fbem-26173 113 46 traded trade VERB fbem-26173 113 47 , , PUNCT fbem-26173 113 48 p p X fbem-26173 113 49 / / SYM fbem-26173 113 50 e e NOUN fbem-26173 113 51 ratio ratio NOUN fbem-26173 113 52 , , PUNCT fbem-26173 113 53 etc etc X fbem-26173 113 54 . . X fbem-26173 114 1 xgboost xgboost PROPN fbem-26173 114 2 is be AUX fbem-26173 114 3 the the DET fbem-26173 114 4 integrated integrated ADJ fbem-26173 114 5 model model NOUN fbem-26173 114 6 with with ADP fbem-26173 114 7 high high ADJ fbem-26173 114 8 predictive predictive ADJ fbem-26173 114 9 accuracy accuracy NOUN fbem-26173 114 10 and and CCONJ fbem-26173 114 11 operational operational ADJ fbem-26173 114 12 efficiency efficiency NOUN fbem-26173 114 13 , , PUNCT fbem-26173 114 14 which which PRON fbem-26173 114 15 can can AUX fbem-26173 114 16 make make VERB fbem-26173 114 17 full full ADJ fbem-26173 114 18 use use NOUN fbem-26173 114 19 of of ADP fbem-26173 114 20 a a DET fbem-26173 114 21 large large ADJ fbem-26173 114 22 amount amount NOUN fbem-26173 114 23 of of ADP fbem-26173 114 24 data datum NOUN fbem-26173 114 25 in in ADP fbem-26173 114 26 the the DET fbem-26173 114 27 financial financial ADJ fbem-26173 114 28 field field NOUN fbem-26173 114 29 , , PUNCT fbem-26173 114 30 and and CCONJ fbem-26173 114 31 at at ADP fbem-26173 114 32 the the DET fbem-26173 114 33 same same ADJ fbem-26173 114 34 time time NOUN fbem-26173 114 35 , , PUNCT fbem-26173 114 36 has have VERB fbem-26173 114 37 very very ADV fbem-26173 114 38 good good ADJ fbem-26173 114 39 stability stability NOUN fbem-26173 114 40 and and CCONJ fbem-26173 114 41 explanatory explanatory ADJ fbem-26173 114 42 power power NOUN fbem-26173 114 43 , , PUNCT fbem-26173 114 44 which which PRON fbem-26173 114 45 is be AUX fbem-26173 114 46 more more ADV fbem-26173 114 47 and and CCONJ fbem-26173 114 48 more more ADV fbem-26173 114 49 frequently frequently ADV fbem-26173 114 50 applied apply VERB fbem-26173 114 51 in in ADP fbem-26173 114 52 the the DET fbem-26173 114 53 modern modern ADJ fbem-26173 114 54 multi multi ADJ fbem-26173 114 55 - - ADJ fbem-26173 114 56 factor factor NOUN fbem-26173 114 57 stock stock NOUN fbem-26173 114 58 selection selection NOUN fbem-26173 114 59 models model NOUN fbem-26173 114 60 . . PUNCT fbem-26173 115 1 therefore therefore ADV fbem-26173 115 2 , , PUNCT fbem-26173 115 3 in in ADP fbem-26173 115 4 this this DET fbem-26173 115 5 paper paper NOUN fbem-26173 115 6 , , PUNCT fbem-26173 115 7 we we PRON fbem-26173 115 8 will will AUX fbem-26173 115 9 divide divide VERB fbem-26173 115 10 the the DET fbem-26173 115 11 training training NOUN fbem-26173 115 12 set set NOUN fbem-26173 115 13 and and CCONJ fbem-26173 115 14 test test NOUN fbem-26173 115 15 set set VERB fbem-26173 115 16 according accord VERB fbem-26173 115 17 to to ADP fbem-26173 115 18 the the DET fbem-26173 115 19 ratio ratio NOUN fbem-26173 115 20 of of ADP fbem-26173 115 21 8:2 8:2 NUM fbem-26173 115 22 , , PUNCT fbem-26173 115 23 and and CCONJ fbem-26173 115 24 construct construct VERB fbem-26173 115 25 the the DET fbem-26173 115 26 multifactor multifactor NOUN fbem-26173 115 27 regression regression NOUN fbem-26173 115 28 model model NOUN fbem-26173 115 29 through through ADP fbem-26173 115 30 xgboost xgboost ADJ fbem-26173 115 31 algorithm algorithm NOUN fbem-26173 115 32 and and CCONJ fbem-26173 115 33 influence influence NOUN fbem-26173 115 34 variables variable NOUN fbem-26173 115 35 , , PUNCT fbem-26173 115 36 and and CCONJ fbem-26173 115 37 the the DET fbem-26173 115 38 parameters parameter NOUN fbem-26173 115 39 are be AUX fbem-26173 115 40 set set VERB fbem-26173 115 41 as as SCONJ fbem-26173 115 42 follows follow VERB fbem-26173 115 43 : : PUNCT fbem-26173 115 44 the the DET fbem-26173 115 45 number number NOUN fbem-26173 115 46 of of ADP fbem-26173 115 47 trees tree NOUN fbem-26173 115 48 n_estimators n_estimator NOUN fbem-26173 115 49 is be AUX fbem-26173 115 50 taken take VERB fbem-26173 115 51 to to PART fbem-26173 115 52 be be AUX fbem-26173 115 53 200 200 NUM fbem-26173 115 54 , , PUNCT fbem-26173 115 55 the the DET fbem-26173 115 56 learning learn VERB fbem-26173 115 57 rate rate NOUN fbem-26173 115 58 learning_rate learning_rate PROPN fbem-26173 115 59 is be AUX fbem-26173 115 60 taken take VERB fbem-26173 115 61 to to PART fbem-26173 115 62 be be AUX fbem-26173 115 63 0.01 0.01 NUM fbem-26173 115 64 , , PUNCT fbem-26173 115 65 the the DET fbem-26173 115 66 base base NOUN fbem-26173 115 67 classifiers classifier NOUN fbem-26173 115 68 booster booster NOUN fbem-26173 115 69 is be AUX fbem-26173 115 70 the the DET fbem-26173 115 71 decision decision NOUN fbem-26173 115 72 tree tree NOUN fbem-26173 115 73 gbtree gbtree NOUN fbem-26173 115 74 , , PUNCT fbem-26173 115 75 the the DET fbem-26173 115 76 objective objective ADJ fbem-26173 115 77 function function NOUN fbem-26173 115 78 objective objective NOUN fbem-26173 115 79 is be AUX fbem-26173 115 80 the the DET fbem-26173 115 81 squared square VERB fbem-26173 115 82 error error NOUN fbem-26173 115 83 reg reg NOUN fbem-26173 115 84 : : PUNCT fbem-26173 115 85 squarederror squarederror NOUN fbem-26173 115 86 . . PUNCT fbem-26173 116 1 from from ADP fbem-26173 116 2 the the DET fbem-26173 116 3 analysis analysis NOUN fbem-26173 116 4 of of ADP fbem-26173 116 5 the the DET fbem-26173 116 6 evaluation evaluation NOUN fbem-26173 116 7 metrics metric NOUN fbem-26173 116 8 in in ADP fbem-26173 116 9 table table NOUN fbem-26173 116 10 2 2 NUM fbem-26173 116 11 , , PUNCT fbem-26173 116 12 it it PRON fbem-26173 116 13 can can AUX fbem-26173 116 14 be be AUX fbem-26173 116 15 seen see VERB fbem-26173 116 16 that that SCONJ fbem-26173 116 17 the the DET fbem-26173 116 18 mae mae PROPN fbem-26173 116 19 , , PUNCT fbem-26173 116 20 mse mse PROPN fbem-26173 116 21 , , PUNCT fbem-26173 116 22 and and CCONJ fbem-26173 116 23 rmse rmse NOUN fbem-26173 116 24 are be AUX fbem-26173 116 25 small small ADJ fbem-26173 116 26 , , PUNCT fbem-26173 116 27 and and CCONJ fbem-26173 116 28 the the DET fbem-26173 116 29 r r NOUN fbem-26173 116 30 - - PUNCT fbem-26173 116 31 square square NOUN fbem-26173 116 32 of of ADP fbem-26173 116 33 both both CCONJ fbem-26173 116 34 the the DET fbem-26173 116 35 training training NOUN fbem-26173 116 36 set set NOUN fbem-26173 116 37 and and CCONJ fbem-26173 116 38 the the DET fbem-26173 116 39 test test NOUN fbem-26173 116 40 set set NOUN fbem-26173 116 41 reaches reach VERB fbem-26173 116 42 more more ADJ fbem-26173 116 43 than than ADP fbem-26173 116 44 0.95 0.95 NUM fbem-26173 116 45 , , PUNCT fbem-26173 116 46 all all PRON fbem-26173 116 47 of of ADP fbem-26173 116 48 which which PRON fbem-26173 116 49 indicate indicate VERB fbem-26173 116 50 that that SCONJ fbem-26173 116 51 xgboost xgboost PROPN fbem-26173 116 52 fits fit VERB fbem-26173 116 53 the the DET fbem-26173 116 54 rise rise NOUN fbem-26173 116 55 better well ADV fbem-26173 116 56 . . PUNCT fbem-26173 117 1 according accord VERB fbem-26173 117 2 to to ADP fbem-26173 117 3 the the DET fbem-26173 117 4 stock stock NOUN fbem-26173 117 5 rise rise NOUN fbem-26173 117 6 trend trend NOUN fbem-26173 117 7 of of ADP fbem-26173 117 8 part part NOUN fbem-26173 117 9 of of ADP fbem-26173 117 10 the the DET fbem-26173 117 11 test test NOUN fbem-26173 117 12 set3 set3 NOUN fbem-26173 117 13 it it PRON fbem-26173 117 14 can can AUX fbem-26173 117 15 be be AUX fbem-26173 117 16 seen see VERB fbem-26173 117 17 that that SCONJ fbem-26173 117 18 the the DET fbem-26173 117 19 local local ADJ fbem-26173 117 20 prediction prediction NOUN fbem-26173 117 21 accuracy accuracy NOUN fbem-26173 117 22 of of ADP fbem-26173 117 23 the the DET fbem-26173 117 24 model model NOUN fbem-26173 117 25 decreases decrease VERB fbem-26173 117 26 when when SCONJ fbem-26173 117 27 the the DET fbem-26173 117 28 stock stock NOUN fbem-26173 117 29 rise rise NOUN fbem-26173 117 30 fluctuates fluctuate VERB fbem-26173 117 31 a a DET fbem-26173 117 32 lot lot NOUN fbem-26173 117 33 , , PUNCT fbem-26173 117 34 but but CCONJ fbem-26173 117 35 from from ADP fbem-26173 117 36 the the DET fbem-26173 117 37 overall overall ADJ fbem-26173 117 38 trend trend NOUN fbem-26173 117 39 , , PUNCT fbem-26173 117 40 the the DET fbem-26173 117 41 predicted predict VERB fbem-26173 117 42 value value NOUN fbem-26173 117 43 is be AUX fbem-26173 117 44 roughly roughly ADV fbem-26173 117 45 the the DET fbem-26173 117 46 same same ADJ fbem-26173 117 47 as as ADP fbem-26173 117 48 the the DET fbem-26173 117 49 real real ADJ fbem-26173 117 50 value value NOUN fbem-26173 117 51 . . PUNCT fbem-26173 118 1 80 80 NUM fbem-26173 118 2 table table NOUN fbem-26173 118 3 2 2 NUM fbem-26173 118 4 . . PUNCT fbem-26173 118 5 xgboost xgboost PROPN fbem-26173 119 1 model model PROPN fbem-26173 119 2 prediction prediction NOUN fbem-26173 119 3 results result VERB fbem-26173 119 4 evaluation evaluation NOUN fbem-26173 119 5 indicators indicator NOUN fbem-26173 119 6 mae mae PROPN fbem-26173 119 7 mse mse PROPN fbem-26173 119 8 rmse rmse PROPN fbem-26173 120 1 r2_score r2_score PROPN fbem-26173 120 2 _ _ PUNCT fbem-26173 120 3 train train NOUN fbem-26173 120 4 r2_scor r2_scor NOUN fbem-26173 120 5 e_test e_test ADV fbem-26173 120 6 projected project VERB fbem-26173 120 7 results result NOUN fbem-26173 120 8 0.273 0.273 NUM fbem-26173 120 9 0 0 NUM fbem-26173 120 10 0.326 0.326 NUM fbem-26173 120 11 3 3 NUM fbem-26173 120 12 0.571 0.571 NUM fbem-26173 120 13 3 3 NUM fbem-26173 120 14 0.9801 0.9801 NUM fbem-26173 120 15 0.9604 0.9604 NUM fbem-26173 120 16 in in ADP fbem-26173 120 17 this this DET fbem-26173 120 18 paper paper NOUN fbem-26173 120 19 , , PUNCT fbem-26173 120 20 the the DET fbem-26173 120 21 xgboost xgboost PROPN fbem-26173 120 22 regression regression NOUN fbem-26173 120 23 model model NOUN fbem-26173 120 24 is be AUX fbem-26173 120 25 constructed construct VERB fbem-26173 120 26 based base VERB fbem-26173 120 27 on on ADP fbem-26173 120 28 the the DET fbem-26173 120 29 influencing influence VERB fbem-26173 120 30 factors factor NOUN fbem-26173 120 31 after after ADP fbem-26173 120 32 filtering filter VERB fbem-26173 120 33 out out ADP fbem-26173 120 34 the the DET fbem-26173 120 35 redundant redundant ADJ fbem-26173 120 36 features feature NOUN fbem-26173 120 37 , , PUNCT fbem-26173 120 38 that that ADV fbem-26173 120 39 is is ADV fbem-26173 120 40 , , PUNCT fbem-26173 120 41 the the DET fbem-26173 120 42 xgboost xgboost PROPN fbem-26173 120 43 model model NOUN fbem-26173 120 44 will will AUX fbem-26173 120 45 take take VERB fbem-26173 120 46 into into ADP fbem-26173 120 47 account account NOUN fbem-26173 120 48 the the DET fbem-26173 120 49 weights weight NOUN fbem-26173 120 50 of of ADP fbem-26173 120 51 the the DET fbem-26173 120 52 influencing influence VERB fbem-26173 120 53 factors factor NOUN fbem-26173 120 54 in in ADP fbem-26173 120 55 the the DET fbem-26173 120 56 prediction prediction NOUN fbem-26173 120 57 , , PUNCT fbem-26173 120 58 and and CCONJ fbem-26173 120 59 get get VERB fbem-26173 120 60 the the DET fbem-26173 120 61 final final ADJ fbem-26173 120 62 prediction prediction NOUN fbem-26173 120 63 results result NOUN fbem-26173 120 64 through through ADP fbem-26173 120 65 the the DET fbem-26173 120 66 weighted weight VERB fbem-26173 120 67 summation summation NOUN fbem-26173 120 68 of of ADP fbem-26173 120 69 each each DET fbem-26173 120 70 factor factor NOUN fbem-26173 120 71 . . PUNCT fbem-26173 121 1 therefore therefore ADV fbem-26173 121 2 , , PUNCT fbem-26173 121 3 compared compare VERB fbem-26173 121 4 with with ADP fbem-26173 121 5 the the DET fbem-26173 121 6 real real ADJ fbem-26173 121 7 value value NOUN fbem-26173 121 8 , , PUNCT fbem-26173 121 9 the the DET fbem-26173 121 10 predicted predict VERB fbem-26173 121 11 value value NOUN fbem-26173 121 12 of of ADP fbem-26173 121 13 stock stock NOUN fbem-26173 121 14 gain gain NOUN fbem-26173 121 15 takes take VERB fbem-26173 121 16 into into ADP fbem-26173 121 17 account account NOUN fbem-26173 121 18 the the DET fbem-26173 121 19 market market NOUN fbem-26173 121 20 performance performance NOUN fbem-26173 121 21 and and CCONJ fbem-26173 121 22 investment investment NOUN fbem-26173 121 23 risk risk NOUN fbem-26173 121 24 of of ADP fbem-26173 121 25 the the DET fbem-26173 121 26 stock stock NOUN fbem-26173 121 27 , , PUNCT fbem-26173 121 28 and and CCONJ fbem-26173 121 29 is be AUX fbem-26173 121 30 more more ADV fbem-26173 121 31 representative representative NOUN fbem-26173 121 32 of of ADP fbem-26173 121 33 the the DET fbem-26173 121 34 daily daily ADJ fbem-26173 121 35 return return NOUN fbem-26173 121 36 of of ADP fbem-26173 121 37 each each DET fbem-26173 121 38 stock stock NOUN fbem-26173 121 39 under under ADP fbem-26173 121 40 the the DET fbem-26173 121 41 influence influence NOUN fbem-26173 121 42 of of ADP fbem-26173 121 43 important important ADJ fbem-26173 121 44 factors factor NOUN fbem-26173 121 45 . . PUNCT fbem-26173 122 1 in in ADP fbem-26173 122 2 this this DET fbem-26173 122 3 paper paper NOUN fbem-26173 122 4 , , PUNCT fbem-26173 122 5 the the DET fbem-26173 122 6 geometric geometric ADJ fbem-26173 122 7 mean mean NOUN fbem-26173 122 8 return return NOUN fbem-26173 122 9 of of ADP fbem-26173 122 10 each each DET fbem-26173 122 11 stock stock NOUN fbem-26173 122 12 in in ADP fbem-26173 122 13 the the DET fbem-26173 122 14 range range NOUN fbem-26173 122 15 of of ADP fbem-26173 122 16 julydecember julydecember PROPN fbem-26173 122 17 2022 2022 NUM fbem-26173 122 18 is be AUX fbem-26173 122 19 calculated calculate VERB fbem-26173 122 20 , , PUNCT fbem-26173 122 21 and and CCONJ fbem-26173 122 22 then then ADV fbem-26173 122 23 the the DET fbem-26173 122 24 four four NUM fbem-26173 122 25 stocks stock NOUN fbem-26173 122 26 with with ADP fbem-26173 122 27 the the DET fbem-26173 122 28 highest high ADJ fbem-26173 122 29 ranking ranking NOUN fbem-26173 122 30 are be AUX fbem-26173 122 31 selected select VERB fbem-26173 122 32 . . PUNCT fbem-26173 123 1 the the DET fbem-26173 123 2 results result NOUN fbem-26173 123 3 are be AUX fbem-26173 123 4 shown show VERB fbem-26173 123 5 in in ADP fbem-26173 123 6 table table NOUN fbem-26173 123 7 3 3 NUM fbem-26173 123 8 , , PUNCT fbem-26173 123 9 and and CCONJ fbem-26173 123 10 four four NUM fbem-26173 123 11 stocks stock NOUN fbem-26173 123 12 , , PUNCT fbem-26173 123 13 china china PROPN fbem-26173 123 14 unicom unicom PROPN fbem-26173 123 15 , , PUNCT fbem-26173 123 16 kingsoft kingsoft PROPN fbem-26173 123 17 office office PROPN fbem-26173 123 18 , , PUNCT fbem-26173 123 19 china china PROPN fbem-26173 123 20 life life NOUN fbem-26173 123 21 , , PUNCT fbem-26173 123 22 and and CCONJ fbem-26173 123 23 china china PROPN fbem-26173 123 24 mobile mobile NOUN fbem-26173 123 25 , , PUNCT fbem-26173 123 26 are be AUX fbem-26173 123 27 finally finally ADV fbem-26173 123 28 selected select VERB fbem-26173 123 29 for for ADP fbem-26173 123 30 the the DET fbem-26173 123 31 portfolio portfolio NOUN fbem-26173 123 32 . . PUNCT fbem-26173 124 1 figure figure VERB fbem-26173 124 2 3 3 NUM fbem-26173 124 3 . . PUNCT fbem-26173 124 4 partial partial ADJ fbem-26173 124 5 test test NOUN fbem-26173 124 6 set set VERB fbem-26173 124 7 fitting fitting ADJ fbem-26173 124 8 plots plot NOUN fbem-26173 124 9 table table NOUN fbem-26173 124 10 3 3 NUM fbem-26173 124 11 . . PUNCT fbem-26173 125 1 ranking rank VERB fbem-26173 125 2 of of ADP fbem-26173 125 3 the the DET fbem-26173 125 4 average average ADJ fbem-26173 125 5 return return NOUN fbem-26173 125 6 of of ADP fbem-26173 125 7 each each DET fbem-26173 125 8 stock stock NOUN fbem-26173 125 9 stock stock NOUN fbem-26173 125 10 code code PROPN fbem-26173 125 11 60005 60005 NUM fbem-26173 125 12 0 0 NUM fbem-26173 125 13 688111 688111 NUM fbem-26173 125 14 60162 60162 NUM fbem-26173 125 15 8 8 NUM fbem-26173 125 16 600941 600941 NUM fbem-26173 125 17 601808 601808 NUM fbem-26173 125 18 601021 601021 NUM fbem-26173 125 19 600570 600570 NUM fbem-26173 125 20 600036 600036 NUM fbem-26173 125 21 601728 601728 NUM fbem-26173 125 22 600150 600150 NUM fbem-26173 126 1 stock stock NOUN fbem-26173 126 2 name name NOUN fbem-26173 126 3 china china PROPN fbem-26173 126 4 unico unico PROPN fbem-26173 126 5 m m PROPN fbem-26173 126 6 kingso kingso PROPN fbem-26173 126 7 ft ft PROPN fbem-26173 126 8 office office PROPN fbem-26173 126 9 china china PROPN fbem-26173 126 10 life life PROPN fbem-26173 126 11 china china PROPN fbem-26173 126 12 mobile mobile PROPN fbem-26173 126 13 china china PROPN fbem-26173 126 14 national national PROPN fbem-26173 126 15 offshore offshore ADJ fbem-26173 126 16 oil oil NOUN fbem-26173 126 17 service service NOUN fbem-26173 126 18 spring spring NOUN fbem-26173 126 19 airline airline NOUN fbem-26173 126 20 s s PART fbem-26173 126 21 hang hang PROPN fbem-26173 126 22 seng seng PROPN fbem-26173 126 23 electronics electronics PROPN fbem-26173 126 24 china china PROPN fbem-26173 126 25 merchants merchants PROPN fbem-26173 126 26 bank bank PROPN fbem-26173 126 27 china china PROPN fbem-26173 126 28 teleco teleco PROPN fbem-26173 126 29 m m PROPN fbem-26173 126 30 china china PROPN fbem-26173 126 31 shipbuild shipbuild PROPN fbem-26173 126 32 ing ing ADJ fbem-26173 126 33 average average ADJ fbem-26173 126 34 rate rate NOUN fbem-26173 126 35 of of ADP fbem-26173 126 36 return return NOUN fbem-26173 126 37 /% /% PUNCT fbem-26173 127 1 1.3113 1.3113 NUM fbem-26173 127 2 1.2602 1.2602 NUM fbem-26173 127 3 1.1819 1.1819 NUM fbem-26173 127 4 1.1737 1.1737 NUM fbem-26173 127 5 1.1625 1.1625 NUM fbem-26173 127 6 1.1612 1.1612 NUM fbem-26173 127 7 1.1539 1.1539 NUM fbem-26173 127 8 1.1474 1.1474 NUM fbem-26173 127 9 1.1447 1.1447 NUM fbem-26173 127 10 1.1417 1.1417 NUM fbem-26173 127 11 3.3 3.3 NUM fbem-26173 127 12 . . PUNCT fbem-26173 128 1 analysis analysis NOUN fbem-26173 128 2 of of ADP fbem-26173 128 3 results result NOUN fbem-26173 128 4 3.3.1 3.3.1 NUM fbem-26173 128 5 . . PUNCT fbem-26173 129 1 cnn cnn PROPN fbem-26173 129 2 - - PUNCT fbem-26173 129 3 bilstm bilstm NOUN fbem-26173 129 4 - - PUNCT fbem-26173 129 5 attention attention NOUN fbem-26173 129 6 stock stock NOUN fbem-26173 129 7 price price NOUN fbem-26173 129 8 prediction prediction NOUN fbem-26173 129 9 based base VERB fbem-26173 129 10 on on ADP fbem-26173 129 11 the the DET fbem-26173 129 12 correlation correlation NOUN fbem-26173 129 13 analysis analysis NOUN fbem-26173 129 14 , , PUNCT fbem-26173 129 15 this this DET fbem-26173 129 16 paper paper NOUN fbem-26173 129 17 selects select VERB fbem-26173 129 18 the the DET fbem-26173 129 19 top top ADJ fbem-26173 129 20 five five NUM fbem-26173 129 21 factors factor NOUN fbem-26173 129 22 with with ADP fbem-26173 129 23 the the DET fbem-26173 129 24 highest high ADJ fbem-26173 129 25 correlation correlation NOUN fbem-26173 129 26 with with ADP fbem-26173 129 27 the the DET fbem-26173 129 28 closing closing NOUN fbem-26173 129 29 price price NOUN fbem-26173 129 30 : : PUNCT fbem-26173 129 31 yesterday yesterday NOUN fbem-26173 129 32 's 's PART fbem-26173 129 33 closing closing NOUN fbem-26173 129 34 price price NOUN fbem-26173 129 35 , , PUNCT fbem-26173 129 36 opening opening NOUN fbem-26173 129 37 price price NOUN fbem-26173 129 38 , , PUNCT fbem-26173 129 39 high high ADJ fbem-26173 129 40 price price NOUN fbem-26173 129 41 , , PUNCT fbem-26173 129 42 low low ADJ fbem-26173 129 43 price price NOUN fbem-26173 129 44 and and CCONJ fbem-26173 129 45 average average ADJ fbem-26173 129 46 price price NOUN fbem-26173 129 47 for for ADP fbem-26173 129 48 the the DET fbem-26173 129 49 next next ADJ fbem-26173 129 50 model model NOUN fbem-26173 129 51 construction construction NOUN fbem-26173 129 52 . . PUNCT fbem-26173 130 1 after after ADP fbem-26173 130 2 the the DET fbem-26173 130 3 stock stock NOUN fbem-26173 130 4 selection selection NOUN fbem-26173 130 5 model model NOUN fbem-26173 130 6 selection selection NOUN fbem-26173 130 7 in in ADP fbem-26173 130 8 the the DET fbem-26173 130 9 previous previous ADJ fbem-26173 130 10 chapter chapter NOUN fbem-26173 130 11 , , PUNCT fbem-26173 130 12 this this DET fbem-26173 130 13 paper paper NOUN fbem-26173 130 14 will will AUX fbem-26173 130 15 688111 688111 NUM fbem-26173 130 16 kingsoft kingsoft PROPN fbem-26173 130 17 office office PROPN fbem-26173 130 18 , , PUNCT fbem-26173 130 19 600050 600050 NUM fbem-26173 130 20 china china PROPN fbem-26173 130 21 unicom unicom PROPN fbem-26173 130 22 , , PUNCT fbem-26173 130 23 601628 601628 NUM fbem-26173 130 24 china china PROPN fbem-26173 130 25 life life NOUN fbem-26173 130 26 and and CCONJ fbem-26173 130 27 600941 600941 NUM fbem-26173 130 28 china china PROPN fbem-26173 130 29 mobile mobile ADJ fbem-26173 130 30 total total ADJ fbem-26173 130 31 four four NUM fbem-26173 130 32 stocks stock NOUN fbem-26173 130 33 of of ADP fbem-26173 130 34 daily daily ADJ fbem-26173 130 35 frequency frequency NOUN fbem-26173 130 36 data datum NOUN fbem-26173 130 37 set set VERB fbem-26173 130 38 for for ADP fbem-26173 130 39 experiment experiment NOUN fbem-26173 130 40 . . PUNCT fbem-26173 131 1 in in ADP fbem-26173 131 2 the the DET fbem-26173 131 3 process process NOUN fbem-26173 131 4 of of ADP fbem-26173 131 5 stock stock NOUN fbem-26173 131 6 price price NOUN fbem-26173 131 7 prediction prediction NOUN fbem-26173 131 8 , , PUNCT fbem-26173 131 9 in in ADP fbem-26173 131 10 order order NOUN fbem-26173 131 11 to to PART fbem-26173 131 12 improve improve VERB fbem-26173 131 13 the the DET fbem-26173 131 14 learning learning NOUN fbem-26173 131 15 ability ability NOUN fbem-26173 131 16 of of ADP fbem-26173 131 17 the the DET fbem-26173 131 18 model model NOUN fbem-26173 131 19 and and CCONJ fbem-26173 131 20 the the DET fbem-26173 131 21 prediction prediction NOUN fbem-26173 131 22 ability ability NOUN fbem-26173 131 23 of of ADP fbem-26173 131 24 the the DET fbem-26173 131 25 stock stock NOUN fbem-26173 131 26 price price NOUN fbem-26173 131 27 change change NOUN fbem-26173 131 28 , , PUNCT fbem-26173 131 29 this this DET fbem-26173 131 30 paper paper NOUN fbem-26173 131 31 divided divide VERB fbem-26173 131 32 the the DET fbem-26173 131 33 data datum NOUN fbem-26173 131 34 set set VERB fbem-26173 131 35 into into ADP fbem-26173 131 36 two two NUM fbem-26173 131 37 parts part NOUN fbem-26173 131 38 , , PUNCT fbem-26173 131 39 in in ADP fbem-26173 131 40 which which PRON fbem-26173 131 41 the the DET fbem-26173 131 42 first first ADJ fbem-26173 131 43 70 70 NUM fbem-26173 131 44 % % NOUN fbem-26173 131 45 of of ADP fbem-26173 131 46 the the DET fbem-26173 131 47 time time NOUN fbem-26173 131 48 series series NOUN fbem-26173 131 49 was be AUX fbem-26173 131 50 used use VERB fbem-26173 131 51 as as ADP fbem-26173 131 52 a a DET fbem-26173 131 53 training training NOUN fbem-26173 131 54 set set VERB fbem-26173 131 55 to to PART fbem-26173 131 56 train train VERB fbem-26173 131 57 the the DET fbem-26173 131 58 network network NOUN fbem-26173 131 59 model model NOUN fbem-26173 131 60 , , PUNCT fbem-26173 131 61 and and CCONJ fbem-26173 131 62 the the DET fbem-26173 131 63 remaining remain VERB fbem-26173 131 64 30 30 NUM fbem-26173 131 65 % % NOUN fbem-26173 131 66 of of ADP fbem-26173 131 67 the the DET fbem-26173 131 68 data datum NOUN fbem-26173 131 69 set set VERB fbem-26173 131 70 was be AUX fbem-26173 131 71 used use VERB fbem-26173 131 72 as as ADP fbem-26173 131 73 a a DET fbem-26173 131 74 test test NOUN fbem-26173 131 75 set set NOUN fbem-26173 131 76 , , PUNCT fbem-26173 131 77 and and CCONJ fbem-26173 131 78 the the DET fbem-26173 131 79 feature feature NOUN fbem-26173 131 80 data datum NOUN fbem-26173 131 81 of of ADP fbem-26173 131 82 the the DET fbem-26173 131 83 test test NOUN fbem-26173 131 84 set set NOUN fbem-26173 131 85 was be AUX fbem-26173 131 86 used use VERB fbem-26173 131 87 to to PART fbem-26173 131 88 predict predict VERB fbem-26173 131 89 the the DET fbem-26173 131 90 four four NUM fbem-26173 131 91 stock stock NOUN fbem-26173 131 92 prices price NOUN fbem-26173 131 93 . . PUNCT fbem-26173 132 1 based base VERB fbem-26173 132 2 on on ADP fbem-26173 132 3 this this PRON fbem-26173 132 4 , , PUNCT fbem-26173 132 5 the the DET fbem-26173 132 6 prediction prediction NOUN fbem-26173 132 7 results result VERB fbem-26173 132 8 in in ADP fbem-26173 132 9 the the DET fbem-26173 132 10 test test NOUN fbem-26173 132 11 set set NOUN fbem-26173 132 12 were be AUX fbem-26173 132 13 compared compare VERB fbem-26173 132 14 with with ADP fbem-26173 132 15 the the DET fbem-26173 132 16 real real ADJ fbem-26173 132 17 results result NOUN fbem-26173 132 18 and and CCONJ fbem-26173 132 19 the the DET fbem-26173 132 20 fitting fitting ADJ fbem-26173 132 21 effect effect NOUN fbem-26173 132 22 was be AUX fbem-26173 132 23 analyzed analyze VERB fbem-26173 132 24 . . PUNCT fbem-26173 133 1 in in ADP fbem-26173 133 2 this this DET fbem-26173 133 3 paper paper NOUN fbem-26173 133 4 , , PUNCT fbem-26173 133 5 model model NOUN fbem-26173 133 6 construction construction NOUN fbem-26173 133 7 and and CCONJ fbem-26173 133 8 experiments experiment NOUN fbem-26173 133 9 are be AUX fbem-26173 133 10 carried carry VERB fbem-26173 133 11 out out ADP fbem-26173 133 12 . . PUNCT fbem-26173 134 1 the the DET fbem-26173 134 2 relu relu NOUN fbem-26173 134 3 activation activation NOUN fbem-26173 134 4 function function NOUN fbem-26173 134 5 is be AUX fbem-26173 134 6 used use VERB fbem-26173 134 7 for for ADP fbem-26173 134 8 the the DET fbem-26173 134 9 onedimensional onedimensional ADJ fbem-26173 134 10 cnn cnn PROPN fbem-26173 134 11 model model NOUN fbem-26173 134 12 , , PUNCT fbem-26173 134 13 the the DET fbem-26173 134 14 tanh tanh NOUN fbem-26173 134 15 activation activation NOUN fbem-26173 134 16 function function NOUN fbem-26173 134 17 is be AUX fbem-26173 134 18 used use VERB fbem-26173 134 19 for for ADP fbem-26173 134 20 the the DET fbem-26173 134 21 bidirectional bidirectional ADJ fbem-26173 134 22 lstm lstm PROPN fbem-26173 134 23 model model NOUN fbem-26173 134 24 , , PUNCT fbem-26173 134 25 and and CCONJ fbem-26173 134 26 the the DET fbem-26173 134 27 sigmoid sigmoid NOUN fbem-26173 134 28 activation activation NOUN fbem-26173 134 29 function function NOUN fbem-26173 134 30 is be AUX fbem-26173 134 31 used use VERB fbem-26173 134 32 for for ADP fbem-26173 134 33 the the DET fbem-26173 134 34 attention attention NOUN fbem-26173 134 35 model model NOUN fbem-26173 134 36 . . PUNCT fbem-26173 135 1 in in ADP fbem-26173 135 2 this this DET fbem-26173 135 3 model model NOUN fbem-26173 135 4 , , PUNCT fbem-26173 135 5 the the DET fbem-26173 135 6 mean mean ADJ fbem-26173 135 7 square square ADJ fbem-26173 135 8 error error NOUN fbem-26173 135 9 is be AUX fbem-26173 135 10 used use VERB fbem-26173 135 11 as as ADP fbem-26173 135 12 the the DET fbem-26173 135 13 loss loss NOUN fbem-26173 135 14 function function NOUN fbem-26173 135 15 , , PUNCT fbem-26173 135 16 and and CCONJ fbem-26173 135 17 the the DET fbem-26173 135 18 optimiser optimiser NOUN fbem-26173 135 19 is be AUX fbem-26173 135 20 specified specify VERB fbem-26173 135 21 as as ADP fbem-26173 135 22 adam adam PROPN fbem-26173 135 23 , , PUNCT fbem-26173 135 24 and and CCONJ fbem-26173 135 25 the the DET fbem-26173 135 26 evaluation evaluation NOUN fbem-26173 135 27 metric metric NOUN fbem-26173 135 28 is be AUX fbem-26173 135 29 specified specify VERB fbem-26173 135 30 as as ADP fbem-26173 135 31 the the DET fbem-26173 135 32 mean mean ADJ fbem-26173 135 33 absolute absolute ADJ fbem-26173 135 34 error error NOUN fbem-26173 135 35 . . PUNCT fbem-26173 136 1 the the DET fbem-26173 136 2 parameters parameter NOUN fbem-26173 136 3 are be AUX fbem-26173 136 4 set set VERB fbem-26173 136 5 as as SCONJ fbem-26173 136 6 follows follow VERB fbem-26173 136 7 : : PUNCT fbem-26173 136 8 window window NOUN fbem-26173 136 9 , , PUNCT fbem-26173 136 10 kernel_size kernel_size PROPN fbem-26173 136 11 and and CCONJ fbem-26173 136 12 pool_size pool_size NOUN fbem-26173 136 13 is be AUX fbem-26173 136 14 all all PRON fbem-26173 136 15 set set VERB fbem-26173 136 16 to to ADP fbem-26173 136 17 1 1 NUM fbem-26173 136 18 , , PUNCT fbem-26173 136 19 lstm_units lstm_unit NOUN fbem-26173 136 20 and and CCONJ fbem-26173 136 21 filters filter NOUN fbem-26173 136 22 are be AUX fbem-26173 136 23 set set VERB fbem-26173 136 24 to to ADP fbem-26173 136 25 128 128 NUM fbem-26173 136 26 , , PUNCT fbem-26173 136 27 dropout dropout NOUN fbem-26173 136 28 is be AUX fbem-26173 136 29 set set VERB fbem-26173 136 30 to to ADP fbem-26173 136 31 0.01 0.01 NUM fbem-26173 136 32 , , PUNCT fbem-26173 136 33 epochs epoch NOUN fbem-26173 136 34 are be AUX fbem-26173 136 35 set set VERB fbem-26173 136 36 to to ADP fbem-26173 136 37 200 200 NUM fbem-26173 136 38 , , PUNCT fbem-26173 136 39 and and CCONJ fbem-26173 136 40 the the DET fbem-26173 136 41 number number NOUN fbem-26173 136 42 of of ADP fbem-26173 136 43 training training NOUN fbem-26173 136 44 rounds round NOUN fbem-26173 136 45 is be AUX fbem-26173 136 46 set set VERB fbem-26173 136 47 to to ADP fbem-26173 136 48 200 200 NUM fbem-26173 136 49 . . PUNCT fbem-26173 137 1 epochs epoch NOUN fbem-26173 137 2 ) ) PUNCT fbem-26173 137 3 is be AUX fbem-26173 137 4 200 200 NUM fbem-26173 137 5 , , PUNCT fbem-26173 137 6 the the DET fbem-26173 137 7 number number NOUN fbem-26173 137 8 of of ADP fbem-26173 137 9 feature feature NOUN fbem-26173 137 10 dimensions dimension NOUN fbem-26173 137 11 ( ( PUNCT fbem-26173 137 12 input_dim input_dim PROPN fbem-26173 137 13 ) ) PUNCT fbem-26173 137 14 is be AUX fbem-26173 137 15 5 5 NUM fbem-26173 137 16 , , PUNCT fbem-26173 137 17 and and CCONJ fbem-26173 137 18 the the DET fbem-26173 137 19 number number NOUN fbem-26173 137 20 of of ADP fbem-26173 137 21 samples sample NOUN fbem-26173 137 22 per per ADP fbem-26173 137 23 training training NOUN fbem-26173 137 24 round round NOUN fbem-26173 137 25 ( ( PUNCT fbem-26173 137 26 batch_size batch_size NOUN fbem-26173 137 27 ) ) PUNCT fbem-26173 137 28 is be AUX fbem-26173 137 29 256 256 NUM fbem-26173 137 30 . . PUNCT fbem-26173 138 1 for for ADP fbem-26173 138 2 this this DET fbem-26173 138 3 stock stock NOUN fbem-26173 138 4 price price NOUN fbem-26173 138 5 prediction prediction NOUN fbem-26173 138 6 framework framework NOUN fbem-26173 138 7 , , PUNCT fbem-26173 138 8 the the DET fbem-26173 138 9 daily daily ADJ fbem-26173 138 10 dataset dataset NOUN fbem-26173 138 11 of of ADP fbem-26173 138 12 each each DET fbem-26173 138 13 stock stock NOUN fbem-26173 138 14 is be AUX fbem-26173 138 15 pre pre ADJ fbem-26173 138 16 - - VERB fbem-26173 138 17 processed processed ADJ fbem-26173 138 18 and and CCONJ fbem-26173 138 19 fed feed VERB fbem-26173 138 20 into into ADP fbem-26173 138 21 the the DET fbem-26173 138 22 model model NOUN fbem-26173 138 23 for for ADP fbem-26173 138 24 learning learning NOUN fbem-26173 138 25 and and CCONJ fbem-26173 138 26 training training NOUN fbem-26173 138 27 respectively respectively ADV fbem-26173 138 28 , , PUNCT fbem-26173 138 29 and and CCONJ fbem-26173 138 30 through through ADP fbem-26173 138 31 continuous continuous ADJ fbem-26173 138 32 experimentation experimentation NOUN fbem-26173 138 33 and and CCONJ fbem-26173 138 34 observation observation NOUN fbem-26173 138 35 of of ADP fbem-26173 138 36 the the DET fbem-26173 138 37 results result NOUN fbem-26173 138 38 , , PUNCT fbem-26173 138 39 the the DET fbem-26173 138 40 knowledge knowledge NOUN fbem-26173 138 41 and and CCONJ fbem-26173 138 42 experience experience NOUN fbem-26173 138 43 are be AUX fbem-26173 138 44 used use VERB fbem-26173 138 45 for for ADP fbem-26173 138 46 constant constant ADJ fbem-26173 138 47 parameter parameter NOUN fbem-26173 138 48 adjustment adjustment NOUN fbem-26173 138 49 and and CCONJ fbem-26173 138 50 optimisation optimisation NOUN fbem-26173 138 51 training training NOUN fbem-26173 138 52 , , PUNCT fbem-26173 138 53 finally finally ADV fbem-26173 138 54 forming form VERB fbem-26173 138 55 a a DET fbem-26173 138 56 set set NOUN fbem-26173 138 57 of of ADP fbem-26173 138 58 parameter parameter NOUN fbem-26173 138 59 values value NOUN fbem-26173 138 60 that that PRON fbem-26173 138 61 are be AUX fbem-26173 138 62 better well ADJ fbem-26173 138 63 for for ADP fbem-26173 138 64 the the DET fbem-26173 138 65 prediction prediction NOUN fbem-26173 138 66 of of ADP fbem-26173 138 67 each each DET fbem-26173 138 68 stock stock NOUN fbem-26173 138 69 , , PUNCT fbem-26173 138 70 and and CCONJ fbem-26173 138 71 then then ADV fbem-26173 138 72 inverse inverse ADJ fbem-26173 138 73 normalisation normalisation NOUN fbem-26173 138 74 is be AUX fbem-26173 138 75 performed perform VERB fbem-26173 138 76 on on ADP fbem-26173 138 77 the the DET fbem-26173 138 78 output output NOUN fbem-26173 138 79 stock stock NOUN fbem-26173 138 80 price price NOUN fbem-26173 138 81 prediction prediction NOUN fbem-26173 138 82 values value NOUN fbem-26173 138 83 to to PART fbem-26173 138 84 obtain obtain VERB fbem-26173 138 85 the the DET fbem-26173 138 86 final final ADJ fbem-26173 138 87 prediction prediction NOUN fbem-26173 138 88 value value NOUN fbem-26173 138 89 of of ADP fbem-26173 138 90 the the DET fbem-26173 138 91 model model NOUN fbem-26173 138 92 . . PUNCT fbem-26173 139 1 the the DET fbem-26173 139 2 results result NOUN fbem-26173 139 3 of of ADP fbem-26173 139 4 the the DET fbem-26173 139 5 model model NOUN fbem-26173 139 6 are be AUX fbem-26173 139 7 shown show VERB fbem-26173 139 8 in in ADP fbem-26173 139 9 table table NOUN fbem-26173 139 10 4 4 NUM fbem-26173 139 11 . . PUNCT fbem-26173 139 12 table table NOUN fbem-26173 139 13 4 4 NUM fbem-26173 139 14 . . PUNCT fbem-26173 139 15 model model NOUN fbem-26173 139 16 prediction prediction NOUN fbem-26173 139 17 results result VERB fbem-26173 139 18 stock stock NOUN fbem-26173 139 19 688111 688111 NUM fbem-26173 139 20 kingsoft kingsoft PROPN fbem-26173 139 21 office office PROPN fbem-26173 139 22 600050 600050 NUM fbem-26173 140 1 china china PROPN fbem-26173 140 2 unicom unicom PROPN fbem-26173 140 3 601628 601628 NUM fbem-26173 140 4 china china PROPN fbem-26173 140 5 life life PROPN fbem-26173 140 6 600941 600941 NUM fbem-26173 140 7 china china PROPN fbem-26173 140 8 mobile mobile PROPN fbem-26173 140 9 mse mse PROPN fbem-26173 140 10 3.4486 3.4486 NUM fbem-26173 140 11 0.0023 0.0023 NUM fbem-26173 140 12 0.0415 0.0415 NUM fbem-26173 140 13 0.5458 0.5458 NUM fbem-26173 140 14 mape mape NOUN fbem-26173 140 15 0.69 0.69 NUM fbem-26173 140 16 % % NOUN fbem-26173 140 17 0.99 0.99 NUM fbem-26173 140 18 % % NOUN fbem-26173 140 19 0.55 0.55 NUM fbem-26173 140 20 % % NOUN fbem-26173 140 21 0.58 0.58 NUM fbem-26173 140 22 % % NOUN fbem-26173 141 1 r2score r2score NOUN fbem-26173 141 2 0.9968 0.9968 NUM fbem-26173 141 3 0.9788 0.9788 NUM fbem-26173 141 4 0.9952 0.9952 NUM fbem-26173 141 5 0.9479 0.9479 NUM fbem-26173 141 6 81 81 NUM fbem-26173 141 7 from from ADP fbem-26173 141 8 the the DET fbem-26173 141 9 results result NOUN fbem-26173 141 10 of of ADP fbem-26173 141 11 each each DET fbem-26173 141 12 evaluation evaluation NOUN fbem-26173 141 13 index index NOUN fbem-26173 141 14 in in ADP fbem-26173 141 15 table table NOUN fbem-26173 141 16 4 4 NUM fbem-26173 141 17 , , PUNCT fbem-26173 141 18 it it PRON fbem-26173 141 19 can can AUX fbem-26173 141 20 be be AUX fbem-26173 141 21 found find VERB fbem-26173 141 22 that that SCONJ fbem-26173 141 23 the the DET fbem-26173 141 24 mape mape NOUN fbem-26173 141 25 of of ADP fbem-26173 141 26 the the DET fbem-26173 141 27 four four NUM fbem-26173 141 28 stocks stock NOUN fbem-26173 141 29 is be AUX fbem-26173 141 30 less less ADJ fbem-26173 141 31 than than ADP fbem-26173 141 32 1 1 NUM fbem-26173 141 33 % % NOUN fbem-26173 141 34 , , PUNCT fbem-26173 141 35 which which PRON fbem-26173 141 36 indicates indicate VERB fbem-26173 141 37 that that SCONJ fbem-26173 141 38 the the DET fbem-26173 141 39 prediction prediction NOUN fbem-26173 141 40 accuracy accuracy NOUN fbem-26173 141 41 of of ADP fbem-26173 141 42 the the DET fbem-26173 141 43 stocks stock NOUN fbem-26173 141 44 is be AUX fbem-26173 141 45 more more ADJ fbem-26173 141 46 than than ADP fbem-26173 141 47 99 99 NUM fbem-26173 141 48 % % NOUN fbem-26173 141 49 . . PUNCT fbem-26173 142 1 except except SCONJ fbem-26173 142 2 for for ADP fbem-26173 142 3 kingsoft kingsoft PROPN fbem-26173 142 4 office office NOUN fbem-26173 142 5 the the DET fbem-26173 142 6 mse mse NOUN fbem-26173 142 7 of of ADP fbem-26173 142 8 the the DET fbem-26173 142 9 other other ADJ fbem-26173 142 10 3 3 NUM fbem-26173 142 11 stocks stock NOUN fbem-26173 142 12 is be AUX fbem-26173 142 13 not not PART fbem-26173 142 14 more more ADJ fbem-26173 142 15 than than ADP fbem-26173 142 16 0.6 0.6 NUM fbem-26173 142 17 . . PUNCT fbem-26173 143 1 for for ADP fbem-26173 143 2 the the DET fbem-26173 143 3 fit fit NOUN fbem-26173 143 4 of of ADP fbem-26173 143 5 this this DET fbem-26173 143 6 regression regression NOUN fbem-26173 143 7 model model NOUN fbem-26173 143 8 , , PUNCT fbem-26173 143 9 it it PRON fbem-26173 143 10 can can AUX fbem-26173 143 11 be be AUX fbem-26173 143 12 seen see VERB fbem-26173 143 13 that that SCONJ fbem-26173 143 14 the the DET fbem-26173 143 15 coefficients coefficient NOUN fbem-26173 143 16 of of ADP fbem-26173 143 17 determination determination NOUN fbem-26173 143 18 of of ADP fbem-26173 143 19 the the DET fbem-26173 143 20 4 4 NUM fbem-26173 143 21 stocks stock NOUN fbem-26173 143 22 are be AUX fbem-26173 143 23 greater great ADJ fbem-26173 143 24 than than ADP fbem-26173 143 25 0.94 0.94 NUM fbem-26173 143 26 , , PUNCT fbem-26173 143 27 which which PRON fbem-26173 143 28 indicates indicate VERB fbem-26173 143 29 that that SCONJ fbem-26173 143 30 the the DET fbem-26173 143 31 predictive predictive ADJ fbem-26173 143 32 performance performance NOUN fbem-26173 143 33 of of ADP fbem-26173 143 34 the the DET fbem-26173 143 35 model model NOUN fbem-26173 143 36 is be AUX fbem-26173 143 37 better well ADJ fbem-26173 143 38 . . PUNCT fbem-26173 144 1 the the DET fbem-26173 144 2 experimental experimental ADJ fbem-26173 144 3 results result NOUN fbem-26173 144 4 show show VERB fbem-26173 144 5 that that SCONJ fbem-26173 144 6 in in ADP fbem-26173 144 7 the the DET fbem-26173 144 8 stock stock NOUN fbem-26173 144 9 dataset dataset NOUN fbem-26173 144 10 tested test VERB fbem-26173 144 11 in in ADP fbem-26173 144 12 this this DET fbem-26173 144 13 paper paper NOUN fbem-26173 144 14 , , PUNCT fbem-26173 144 15 the the DET fbem-26173 144 16 stock stock NOUN fbem-26173 144 17 price price NOUN fbem-26173 144 18 prediction prediction NOUN fbem-26173 144 19 model model NOUN fbem-26173 144 20 used use VERB fbem-26173 144 21 , , PUNCT fbem-26173 144 22 cnn cnn PROPN fbem-26173 144 23 - - PUNCT fbem-26173 144 24 bilstm bilstm NOUN fbem-26173 144 25 - - PUNCT fbem-26173 144 26 attention attention NOUN fbem-26173 144 27 model model NOUN fbem-26173 144 28 , , PUNCT fbem-26173 144 29 has have VERB fbem-26173 144 30 better well ADJ fbem-26173 144 31 generalisation generalisation NOUN fbem-26173 144 32 ability ability NOUN fbem-26173 144 33 and and CCONJ fbem-26173 144 34 robustness robustness NOUN fbem-26173 144 35 , , PUNCT fbem-26173 144 36 which which PRON fbem-26173 144 37 is be AUX fbem-26173 144 38 beneficial beneficial ADJ fbem-26173 144 39 for for ADP fbem-26173 144 40 the the DET fbem-26173 144 41 subsequent subsequent ADJ fbem-26173 144 42 quantitative quantitative ADJ fbem-26173 144 43 trading trading NOUN fbem-26173 144 44 strategies strategy NOUN fbem-26173 144 45 and and CCONJ fbem-26173 144 46 backtesting backteste VERB fbem-26173 144 47 experiments experiment NOUN fbem-26173 144 48 . . PUNCT fbem-26173 145 1 in in ADP fbem-26173 145 2 order order NOUN fbem-26173 145 3 to to PART fbem-26173 145 4 more more ADV fbem-26173 145 5 intuitively intuitively ADV fbem-26173 145 6 observe observe VERB fbem-26173 145 7 the the DET fbem-26173 145 8 prediction prediction NOUN fbem-26173 145 9 effect effect NOUN fbem-26173 145 10 of of ADP fbem-26173 145 11 the the DET fbem-26173 145 12 model model NOUN fbem-26173 145 13 on on ADP fbem-26173 145 14 the the DET fbem-26173 145 15 test test NOUN fbem-26173 145 16 set set NOUN fbem-26173 145 17 , , PUNCT fbem-26173 145 18 the the DET fbem-26173 145 19 fitting fitting ADJ fbem-26173 145 20 plots plot NOUN fbem-26173 145 21 of of ADP fbem-26173 145 22 4 4 NUM fbem-26173 145 23 stocks stock NOUN fbem-26173 145 24 , , PUNCT fbem-26173 145 25 688111 688111 NUM fbem-26173 145 26 kingsoft kingsoft PROPN fbem-26173 145 27 office office PROPN fbem-26173 145 28 , , PUNCT fbem-26173 145 29 600050 600050 NUM fbem-26173 145 30 china china PROPN fbem-26173 145 31 unicom unicom PROPN fbem-26173 145 32 , , PUNCT fbem-26173 145 33 601628 601628 NUM fbem-26173 145 34 china china PROPN fbem-26173 145 35 life life NOUN fbem-26173 145 36 , , PUNCT fbem-26173 145 37 and and CCONJ fbem-26173 145 38 600941 600941 NUM fbem-26173 145 39 china china PROPN fbem-26173 145 40 mobile mobile NOUN fbem-26173 145 41 , , PUNCT fbem-26173 145 42 are be AUX fbem-26173 145 43 next next ADV fbem-26173 145 44 plotted plot VERB fbem-26173 145 45 on on ADP fbem-26173 145 46 the the DET fbem-26173 145 47 test test NOUN fbem-26173 145 48 set set NOUN fbem-26173 145 49 , , PUNCT fbem-26173 145 50 which which PRON fbem-26173 145 51 are be AUX fbem-26173 145 52 sequentially sequentially ADV fbem-26173 145 53 shown show VERB fbem-26173 145 54 in in ADP fbem-26173 145 55 figures figure NOUN fbem-26173 145 56 4 4 NUM fbem-26173 145 57 - - PUNCT fbem-26173 145 58 figure figure NOUN fbem-26173 145 59 7 7 NUM fbem-26173 145 60 , , PUNCT fbem-26173 145 61 respectively respectively ADV fbem-26173 145 62 . . PUNCT fbem-26173 146 1 figure figure NOUN fbem-26173 146 2 4 4 NUM fbem-26173 146 3 . . X fbem-26173 146 4 688111 688111 NUM fbem-26173 146 5 kingsoft kingsoft PROPN fbem-26173 146 6 office office NOUN fbem-26173 146 7 test test NOUN fbem-26173 146 8 set set VERB fbem-26173 146 9 fitting fitting ADJ fbem-26173 146 10 chart chart NOUN fbem-26173 146 11 figure figure NOUN fbem-26173 146 12 5 5 NUM fbem-26173 146 13 . . NUM fbem-26173 146 14 600050 600050 NUM fbem-26173 147 1 china china PROPN fbem-26173 147 2 unicom unicom PROPN fbem-26173 147 3 test test PROPN fbem-26173 147 4 set set VERB fbem-26173 147 5 fit fit ADJ fbem-26173 147 6 chart chart NOUN fbem-26173 147 7 figure figure NOUN fbem-26173 147 8 6 6 NUM fbem-26173 147 9 . . SYM fbem-26173 147 10 601628 601628 NUM fbem-26173 148 1 china china PROPN fbem-26173 148 2 life life NOUN fbem-26173 148 3 test test NOUN fbem-26173 148 4 set set VERB fbem-26173 148 5 fit fit ADJ fbem-26173 148 6 chart chart NOUN fbem-26173 148 7 figure figure NOUN fbem-26173 148 8 7 7 NUM fbem-26173 148 9 . . NUM fbem-26173 148 10 600941 600941 NUM fbem-26173 148 11 china china PROPN fbem-26173 148 12 mobile mobile ADJ fbem-26173 148 13 test test NOUN fbem-26173 148 14 set set VERB fbem-26173 148 15 fit fit ADJ fbem-26173 148 16 chart chart NOUN fbem-26173 148 17 from from ADP fbem-26173 148 18 the the DET fbem-26173 148 19 prediction prediction NOUN fbem-26173 148 20 graphs graph NOUN fbem-26173 148 21 of of ADP fbem-26173 148 22 these these DET fbem-26173 148 23 4 4 NUM fbem-26173 148 24 stocks stock NOUN fbem-26173 148 25 on on ADP fbem-26173 148 26 the the DET fbem-26173 148 27 test test NOUN fbem-26173 148 28 set set NOUN fbem-26173 148 29 , , PUNCT fbem-26173 148 30 the the DET fbem-26173 148 31 model model NOUN fbem-26173 148 32 can can AUX fbem-26173 148 33 roughly roughly ADV fbem-26173 148 34 predict predict VERB fbem-26173 148 35 the the DET fbem-26173 148 36 trend trend NOUN fbem-26173 148 37 of of ADP fbem-26173 148 38 the the DET fbem-26173 148 39 stock stock NOUN fbem-26173 148 40 price price NOUN fbem-26173 148 41 , , PUNCT fbem-26173 148 42 for for ADP fbem-26173 148 43 the the DET fbem-26173 148 44 local local ADJ fbem-26173 148 45 peaks peak NOUN fbem-26173 148 46 with with ADP fbem-26173 148 47 the the DET fbem-26173 148 48 accuracy accuracy NOUN fbem-26173 148 49 is be AUX fbem-26173 148 50 lacking lack VERB fbem-26173 148 51 , , PUNCT fbem-26173 148 52 but but CCONJ fbem-26173 148 53 in in ADP fbem-26173 148 54 general general ADJ fbem-26173 148 55 , , PUNCT fbem-26173 148 56 the the DET fbem-26173 148 57 generalisation generalisation NOUN fbem-26173 148 58 ability ability NOUN fbem-26173 148 59 and and CCONJ fbem-26173 148 60 stability stability NOUN fbem-26173 148 61 of of ADP fbem-26173 148 62 the the DET fbem-26173 148 63 model model NOUN fbem-26173 148 64 are be AUX fbem-26173 148 65 good good ADJ fbem-26173 148 66 . . PUNCT fbem-26173 149 1 moreover moreover ADV fbem-26173 149 2 , , PUNCT fbem-26173 149 3 when when SCONJ fbem-26173 149 4 the the DET fbem-26173 149 5 model model NOUN fbem-26173 149 6 reaches reach VERB fbem-26173 149 7 the the DET fbem-26173 149 8 highest high ADJ fbem-26173 149 9 point point NOUN fbem-26173 149 10 or or CCONJ fbem-26173 149 11 the the DET fbem-26173 149 12 lowest low ADJ fbem-26173 149 13 point point NOUN fbem-26173 149 14 , , PUNCT fbem-26173 149 15 its its PRON fbem-26173 149 16 prediction prediction NOUN fbem-26173 149 17 curve curve NOUN fbem-26173 149 18 has have VERB fbem-26173 149 19 almost almost ADV fbem-26173 149 20 no no PRON fbem-26173 149 21 offset offset NOUN fbem-26173 149 22 compared compare VERB fbem-26173 149 23 with with ADP fbem-26173 149 24 the the DET fbem-26173 149 25 real real ADJ fbem-26173 149 26 curve curve NOUN fbem-26173 149 27 , , PUNCT fbem-26173 149 28 which which PRON fbem-26173 149 29 indicates indicate VERB fbem-26173 149 30 that that SCONJ fbem-26173 149 31 the the DET fbem-26173 149 32 model model NOUN fbem-26173 149 33 can can AUX fbem-26173 149 34 eliminate eliminate VERB fbem-26173 149 35 the the DET fbem-26173 149 36 phase phase NOUN fbem-26173 149 37 difference difference NOUN fbem-26173 149 38 without without ADP fbem-26173 149 39 time time NOUN fbem-26173 149 40 delay delay NOUN fbem-26173 149 41 phenomenon phenomenon NOUN fbem-26173 149 42 . . PUNCT fbem-26173 150 1 and and CCONJ fbem-26173 150 2 it it PRON fbem-26173 150 3 can can AUX fbem-26173 150 4 be be AUX fbem-26173 150 5 seen see VERB fbem-26173 150 6 from from ADP fbem-26173 150 7 the the DET fbem-26173 150 8 results result NOUN fbem-26173 150 9 of of ADP fbem-26173 150 10 the the DET fbem-26173 150 11 above above ADJ fbem-26173 150 12 evaluation evaluation NOUN fbem-26173 150 13 indexes index NOUN fbem-26173 150 14 that that PRON fbem-26173 150 15 the the DET fbem-26173 150 16 overall overall ADJ fbem-26173 150 17 performance performance NOUN fbem-26173 150 18 of of ADP fbem-26173 150 19 the the DET fbem-26173 150 20 model model NOUN fbem-26173 150 21 is be AUX fbem-26173 150 22 better well ADJ fbem-26173 150 23 and and CCONJ fbem-26173 150 24 the the DET fbem-26173 150 25 error error NOUN fbem-26173 150 26 of of ADP fbem-26173 150 27 prediction prediction NOUN fbem-26173 150 28 is be AUX fbem-26173 150 29 smaller small ADJ fbem-26173 150 30 . . PUNCT fbem-26173 151 1 3.3.2 3.3.2 X fbem-26173 151 2 . . PUNCT fbem-26173 151 3 quantitative quantitative ADJ fbem-26173 151 4 stock stock NOUN fbem-26173 151 5 trading trading NOUN fbem-26173 151 6 strategies strategy NOUN fbem-26173 151 7 and and CCONJ fbem-26173 151 8 backtesting backteste VERB fbem-26173 151 9 in in ADP fbem-26173 151 10 this this DET fbem-26173 151 11 paper paper NOUN fbem-26173 151 12 , , PUNCT fbem-26173 151 13 the the DET fbem-26173 151 14 macd macd NOUN fbem-26173 151 15 trading trading NOUN fbem-26173 151 16 strategy strategy NOUN fbem-26173 151 17 is be AUX fbem-26173 151 18 applied apply VERB fbem-26173 151 19 to to ADP fbem-26173 151 20 4 4 NUM fbem-26173 151 21 stocks stock NOUN fbem-26173 151 22 , , PUNCT fbem-26173 151 23 688111 688111 NUM fbem-26173 151 24 kingsoft kingsoft PROPN fbem-26173 151 25 office office PROPN fbem-26173 151 26 , , PUNCT fbem-26173 151 27 600050 600050 NUM fbem-26173 151 28 china china PROPN fbem-26173 151 29 unicom unicom PROPN fbem-26173 151 30 , , PUNCT fbem-26173 151 31 601628 601628 NUM fbem-26173 151 32 china china PROPN fbem-26173 151 33 life life NOUN fbem-26173 151 34 , , PUNCT fbem-26173 151 35 600941 600941 NUM fbem-26173 151 36 china china PROPN fbem-26173 151 37 mobile mobile NOUN fbem-26173 151 38 , , PUNCT fbem-26173 151 39 and and CCONJ fbem-26173 151 40 688111 688111 NUM fbem-26173 151 41 kingsoft kingsoft PROPN fbem-26173 151 42 office office NOUN fbem-26173 151 43 , , PUNCT fbem-26173 151 44 respectively respectively ADV fbem-26173 151 45 , , PUNCT fbem-26173 151 46 for for ADP fbem-26173 151 47 june june PROPN fbem-26173 151 48 - - PUNCT fbem-26173 151 49 december december PROPN fbem-26173 151 50 2022 2022 NUM fbem-26173 151 51 , , PUNCT fbem-26173 151 52 based base VERB fbem-26173 151 53 on on ADP fbem-26173 151 54 the the DET fbem-26173 151 55 data data NOUN fbem-26173 151 56 test test NOUN fbem-26173 151 57 set set NOUN fbem-26173 151 58 obtained obtain VERB fbem-26173 151 59 from from ADP fbem-26173 151 60 the the DET fbem-26173 151 61 cnn cnn PROPN fbem-26173 151 62 - - PUNCT fbem-26173 151 63 bilstmattention bilstmattention NOUN fbem-26173 151 64 model model NOUN fbem-26173 151 65 prediction prediction NOUN fbem-26173 151 66 , , PUNCT fbem-26173 151 67 and and CCONJ fbem-26173 151 68 the the DET fbem-26173 151 69 respective respective ADJ fbem-26173 151 70 stock stock NOUN fbem-26173 151 71 the the DET fbem-26173 151 72 macd macd NOUN fbem-26173 151 73 strategy strategy NOUN fbem-26173 151 74 backtest backt ADJ fbem-26173 151 75 total total ADJ fbem-26173 151 76 return return NOUN fbem-26173 151 77 results result NOUN fbem-26173 151 78 are be AUX fbem-26173 151 79 shown show VERB fbem-26173 151 80 in in ADP fbem-26173 151 81 figure figure NOUN fbem-26173 151 82 8 8 NUM fbem-26173 151 83 - - PUNCT fbem-26173 151 84 figure figure NOUN fbem-26173 151 85 11 11 NUM fbem-26173 151 86 . . PUNCT fbem-26173 152 1 cumulative_returns cumulative_return NOUN fbem-26173 152 2 denote denote VERB fbem-26173 152 3 the the DET fbem-26173 152 4 total total ADJ fbem-26173 152 5 returns return NOUN fbem-26173 152 6 obtained obtain VERB fbem-26173 152 7 based base VERB fbem-26173 152 8 on on ADP fbem-26173 152 9 the the DET fbem-26173 152 10 strategy strategy NOUN fbem-26173 152 11 and and CCONJ fbem-26173 152 12 tcumulative_returns tcumulative_return NOUN fbem-26173 152 13 denote denote VERB fbem-26173 152 14 the the DET fbem-26173 152 15 actual actual ADJ fbem-26173 152 16 total total ADJ fbem-26173 152 17 returns return NOUN fbem-26173 152 18 . . PUNCT fbem-26173 153 1 figure figure NOUN fbem-26173 153 2 8 8 NUM fbem-26173 153 3 . . PUNCT fbem-26173 154 1 688111 688111 NUM fbem-26173 154 2 kingsford kingsford PROPN fbem-26173 154 3 office office NOUN fbem-26173 154 4 based base VERB fbem-26173 154 5 on on ADP fbem-26173 154 6 strategy strategy NOUN fbem-26173 154 7 and and CCONJ fbem-26173 154 8 true true ADJ fbem-26173 154 9 cumulative cumulative ADJ fbem-26173 154 10 returns return NOUN fbem-26173 154 11 figure figure VERB fbem-26173 154 12 9 9 NUM fbem-26173 154 13 . . NUM fbem-26173 155 1 600050 600050 NUM fbem-26173 156 1 china china PROPN fbem-26173 156 2 unicom unicom PROPN fbem-26173 156 3 based base VERB fbem-26173 156 4 on on ADP fbem-26173 156 5 strategy strategy NOUN fbem-26173 156 6 and and CCONJ fbem-26173 156 7 true true ADJ fbem-26173 156 8 cumulative cumulative ADJ fbem-26173 156 9 returns return NOUN fbem-26173 156 10 figure figure VERB fbem-26173 156 11 10 10 NUM fbem-26173 156 12 . . PUNCT fbem-26173 157 1 601628 601628 NUM fbem-26173 157 2 china china PROPN fbem-26173 157 3 life life NOUN fbem-26173 157 4 based base VERB fbem-26173 157 5 on on ADP fbem-26173 157 6 strategy strategy NOUN fbem-26173 157 7 and and CCONJ fbem-26173 157 8 true true ADJ fbem-26173 157 9 cumulative cumulative ADJ fbem-26173 157 10 returns return NOUN fbem-26173 157 11 82 82 NUM fbem-26173 157 12 figure figure NOUN fbem-26173 157 13 11 11 NUM fbem-26173 157 14 . . PUNCT fbem-26173 158 1 600941 600941 NUM fbem-26173 158 2 china china PROPN fbem-26173 158 3 mobile mobile NOUN fbem-26173 158 4 based base VERB fbem-26173 158 5 on on ADP fbem-26173 158 6 strategy strategy NOUN fbem-26173 158 7 and and CCONJ fbem-26173 158 8 true true ADJ fbem-26173 158 9 cumulative cumulative ADJ fbem-26173 158 10 returns return NOUN fbem-26173 158 11 it it PRON fbem-26173 158 12 can can AUX fbem-26173 158 13 be be AUX fbem-26173 158 14 observed observe VERB fbem-26173 158 15 that that SCONJ fbem-26173 158 16 the the DET fbem-26173 158 17 backtesting backtesting NOUN fbem-26173 158 18 of of ADP fbem-26173 158 19 the the DET fbem-26173 158 20 four four NUM fbem-26173 158 21 stocks stock NOUN fbem-26173 158 22 selected select VERB fbem-26173 158 23 above above ADP fbem-26173 158 24 using use VERB fbem-26173 158 25 the the DET fbem-26173 158 26 macd macd NOUN fbem-26173 158 27 trading trading NOUN fbem-26173 158 28 strategy strategy NOUN fbem-26173 158 29 yields yield NOUN fbem-26173 158 30 that that PRON fbem-26173 158 31 most most ADJ fbem-26173 158 32 of of ADP fbem-26173 158 33 the the DET fbem-26173 158 34 stocks stock NOUN fbem-26173 158 35 β€˜ ' PUNCT fbem-26173 158 36 total total ADJ fbem-26173 158 37 returns return NOUN fbem-26173 158 38 based base VERB fbem-26173 158 39 on on ADP fbem-26173 158 40 the the DET fbem-26173 158 41 strategy strategy NOUN fbem-26173 158 42 are be AUX fbem-26173 158 43 lower low ADJ fbem-26173 158 44 than than ADP fbem-26173 158 45 their their PRON fbem-26173 158 46 actual actual ADJ fbem-26173 158 47 total total ADJ fbem-26173 158 48 returns return NOUN fbem-26173 158 49 , , PUNCT fbem-26173 158 50 with with ADP fbem-26173 158 51 the the DET fbem-26173 158 52 exception exception NOUN fbem-26173 158 53 of of ADP fbem-26173 158 54 china china PROPN fbem-26173 158 55 unicom unicom PROPN fbem-26173 158 56 , , PUNCT fbem-26173 158 57 a a DET fbem-26173 158 58 stock stock NOUN fbem-26173 158 59 whose whose DET fbem-26173 158 60 own own ADJ fbem-26173 158 61 closing closing NOUN fbem-26173 158 62 price price NOUN fbem-26173 158 63 base base NOUN fbem-26173 158 64 is be AUX fbem-26173 158 65 relatively relatively ADV fbem-26173 158 66 small small ADJ fbem-26173 158 67 , , PUNCT fbem-26173 158 68 resulting result VERB fbem-26173 158 69 in in ADP fbem-26173 158 70 the the DET fbem-26173 158 71 total total ADJ fbem-26173 158 72 return return NOUN fbem-26173 158 73 based base VERB fbem-26173 158 74 on on ADP fbem-26173 158 75 the the DET fbem-26173 158 76 strategy strategy NOUN fbem-26173 158 77 nearly nearly ADV fbem-26173 158 78 overlapping overlap VERB fbem-26173 158 79 with with ADP fbem-26173 158 80 its its PRON fbem-26173 158 81 actual actual ADJ fbem-26173 158 82 total total ADJ fbem-26173 158 83 return return NOUN fbem-26173 158 84 curve curve NOUN fbem-26173 158 85 , , PUNCT fbem-26173 158 86 and and CCONJ fbem-26173 158 87 most most ADJ fbem-26173 158 88 of of ADP fbem-26173 158 89 the the DET fbem-26173 158 90 remaining remain VERB fbem-26173 158 91 stocks stock NOUN fbem-26173 158 92 ’ ' PUNCT fbem-26173 158 93 returns return NOUN fbem-26173 158 94 based base VERB fbem-26173 158 95 on on ADP fbem-26173 158 96 the the DET fbem-26173 158 97 strategy strategy NOUN fbem-26173 158 98 curve curve NOUN fbem-26173 158 99 are be AUX fbem-26173 158 100 all all ADV fbem-26173 158 101 below below ADP fbem-26173 158 102 the the DET fbem-26173 158 103 actual actual ADJ fbem-26173 158 104 return return NOUN fbem-26173 158 105 curve curve NOUN fbem-26173 158 106 , , PUNCT fbem-26173 158 107 indicating indicate VERB fbem-26173 158 108 that that SCONJ fbem-26173 158 109 when when SCONJ fbem-26173 158 110 we we PRON fbem-26173 158 111 can can AUX fbem-26173 158 112 get get VERB fbem-26173 158 113 a a DET fbem-26173 158 114 return return NOUN fbem-26173 158 115 based base VERB fbem-26173 158 116 on on ADP fbem-26173 158 117 the the DET fbem-26173 158 118 trading trading NOUN fbem-26173 158 119 strategy strategy NOUN fbem-26173 158 120 , , PUNCT fbem-26173 158 121 we we PRON fbem-26173 158 122 can can AUX fbem-26173 158 123 actually actually ADV fbem-26173 158 124 get get VERB fbem-26173 158 125 a a DET fbem-26173 158 126 higher high ADJ fbem-26173 158 127 return return NOUN fbem-26173 158 128 , , PUNCT fbem-26173 158 129 which which PRON fbem-26173 158 130 can can AUX fbem-26173 158 131 be be AUX fbem-26173 158 132 traded trade VERB fbem-26173 158 133 at at ADP fbem-26173 158 134 that that DET fbem-26173 158 135 time time NOUN fbem-26173 158 136 . . PUNCT fbem-26173 159 1 and and CCONJ fbem-26173 159 2 from from ADP fbem-26173 159 3 table table NOUN fbem-26173 159 4 5 5 NUM fbem-26173 159 5 , , PUNCT fbem-26173 159 6 we we PRON fbem-26173 159 7 can can AUX fbem-26173 159 8 see see VERB fbem-26173 159 9 that that SCONJ fbem-26173 159 10 the the DET fbem-26173 159 11 total total ADJ fbem-26173 159 12 return return NOUN fbem-26173 159 13 based base VERB fbem-26173 159 14 on on ADP fbem-26173 159 15 the the DET fbem-26173 159 16 strategy strategy NOUN fbem-26173 159 17 is be AUX fbem-26173 159 18 more more ADJ fbem-26173 159 19 than than ADP fbem-26173 159 20 40 40 NUM fbem-26173 159 21 % % NOUN fbem-26173 159 22 for for ADP fbem-26173 159 23 most most ADJ fbem-26173 159 24 of of ADP fbem-26173 159 25 the the DET fbem-26173 159 26 stocks stock NOUN fbem-26173 159 27 , , PUNCT fbem-26173 159 28 and and CCONJ fbem-26173 159 29 the the DET fbem-26173 159 30 loss loss NOUN fbem-26173 159 31 is be AUX fbem-26173 159 32 controlled control VERB fbem-26173 159 33 within within ADP fbem-26173 159 34 10 10 NUM fbem-26173 159 35 % % NOUN fbem-26173 159 36 , , PUNCT fbem-26173 159 37 which which PRON fbem-26173 159 38 indicates indicate VERB fbem-26173 159 39 that that SCONJ fbem-26173 159 40 the the DET fbem-26173 159 41 strategy strategy NOUN fbem-26173 159 42 is be AUX fbem-26173 159 43 still still ADV fbem-26173 159 44 effective effective ADJ fbem-26173 159 45 and and CCONJ fbem-26173 159 46 feasible feasible ADJ fbem-26173 159 47 . . PUNCT fbem-26173 160 1 table table NOUN fbem-26173 160 2 5 5 NUM fbem-26173 160 3 . . PUNCT fbem-26173 160 4 total total ADJ fbem-26173 160 5 return return NOUN fbem-26173 160 6 on on ADP fbem-26173 160 7 four four NUM fbem-26173 160 8 stocks stock NOUN fbem-26173 160 9 stock stock NOUN fbem-26173 160 10 total total NOUN fbem-26173 160 11 return return NOUN fbem-26173 160 12 maximum maximum ADJ fbem-26173 160 13 gross gross ADJ fbem-26173 160 14 yield yield NOUN fbem-26173 160 15 minimum minimum NOUN fbem-26173 160 16 total total ADJ fbem-26173 160 17 return return NOUN fbem-26173 160 18 688111 688111 NUM fbem-26173 160 19 kingsford kingsford PROPN fbem-26173 160 20 office office PROPN fbem-26173 160 21 28.00 28.00 NUM fbem-26173 160 22 % % NOUN fbem-26173 160 23 45.19 45.19 NUM fbem-26173 160 24 % % NOUN fbem-26173 161 1 -18.66 -18.66 ADP fbem-26173 161 2 % % NOUN fbem-26173 161 3 600050 600050 NUM fbem-26173 161 4 china china PROPN fbem-26173 161 5 unicom unicom PROPN fbem-26173 161 6 44.75 44.75 NUM fbem-26173 161 7 % % NOUN fbem-26173 161 8 44.98 44.98 NUM fbem-26173 161 9 % % NOUN fbem-26173 161 10 -2.52 -2.52 ADJ fbem-26173 161 11 % % NOUN fbem-26173 161 12 600941 600941 NUM fbem-26173 161 13 china china PROPN fbem-26173 161 14 mobile mobile NOUN fbem-26173 161 15 13.80 13.80 NUM fbem-26173 161 16 % % NOUN fbem-26173 161 17 16.46 16.46 NUM fbem-26173 161 18 % % NOUN fbem-26173 161 19 -8.66 -8.66 NUM fbem-26173 161 20 % % NOUN fbem-26173 161 21 601628 601628 NUM fbem-26173 161 22 china china PROPN fbem-26173 161 23 life life NOUN fbem-26173 161 24 44.39 44.39 NUM fbem-26173 161 25 % % NOUN fbem-26173 161 26 47.61 47.61 NUM fbem-26173 161 27 % % NOUN fbem-26173 161 28 0.56 0.56 NUM fbem-26173 161 29 % % NOUN fbem-26173 161 30 the the DET fbem-26173 161 31 above above ADJ fbem-26173 161 32 introduces introduce VERB fbem-26173 161 33 the the DET fbem-26173 161 34 quantitative quantitative ADJ fbem-26173 161 35 trading trading NOUN fbem-26173 161 36 strategy strategy NOUN fbem-26173 161 37 based base VERB fbem-26173 161 38 on on ADP fbem-26173 161 39 macd macd NOUN fbem-26173 161 40 single single ADJ fbem-26173 161 41 stock stock NOUN fbem-26173 161 42 investment investment NOUN fbem-26173 161 43 , , PUNCT fbem-26173 161 44 but but CCONJ fbem-26173 161 45 in in ADP fbem-26173 161 46 real real ADJ fbem-26173 161 47 life life NOUN fbem-26173 161 48 , , PUNCT fbem-26173 161 49 we we PRON fbem-26173 161 50 may may AUX fbem-26173 161 51 invest invest VERB fbem-26173 161 52 in in ADP fbem-26173 161 53 more more ADJ fbem-26173 161 54 than than ADP fbem-26173 161 55 one one NUM fbem-26173 161 56 stock stock NOUN fbem-26173 161 57 when when SCONJ fbem-26173 161 58 investing invest VERB fbem-26173 161 59 , , PUNCT fbem-26173 161 60 at at ADP fbem-26173 161 61 this this DET fbem-26173 161 62 time time NOUN fbem-26173 161 63 the the DET fbem-26173 161 64 amount amount NOUN fbem-26173 161 65 of of ADP fbem-26173 161 66 investment investment NOUN fbem-26173 161 67 in in ADP fbem-26173 161 68 each each DET fbem-26173 161 69 stock stock NOUN fbem-26173 161 70 has have AUX fbem-26173 161 71 become become VERB fbem-26173 161 72 a a DET fbem-26173 161 73 difficult difficult ADJ fbem-26173 161 74 point point NOUN fbem-26173 161 75 in in ADP fbem-26173 161 76 our our PRON fbem-26173 161 77 investment investment NOUN fbem-26173 161 78 , , PUNCT fbem-26173 161 79 based base VERB fbem-26173 161 80 on on ADP fbem-26173 161 81 this this PRON fbem-26173 161 82 we we PRON fbem-26173 161 83 continue continue VERB fbem-26173 161 84 to to PART fbem-26173 161 85 take take VERB fbem-26173 161 86 the the DET fbem-26173 161 87 above above ADJ fbem-26173 161 88 four four NUM fbem-26173 161 89 stocks stock NOUN fbem-26173 161 90 as as ADP fbem-26173 161 91 an an DET fbem-26173 161 92 example example NOUN fbem-26173 161 93 , , PUNCT fbem-26173 161 94 select select VERB fbem-26173 161 95 four four NUM fbem-26173 161 96 kinds kind NOUN fbem-26173 161 97 of of ADP fbem-26173 161 98 investment investment NOUN fbem-26173 161 99 portfolio portfolio NOUN fbem-26173 161 100 , , PUNCT fbem-26173 161 101 through through ADP fbem-26173 161 102 the the DET fbem-26173 161 103 calculation calculation NOUN fbem-26173 161 104 of of ADP fbem-26173 161 105 the the DET fbem-26173 161 106 total total ADJ fbem-26173 161 107 return return NOUN fbem-26173 161 108 on on ADP fbem-26173 161 109 each each DET fbem-26173 161 110 portfolio portfolio NOUN fbem-26173 161 111 to to PART fbem-26173 161 112 determine determine VERB fbem-26173 161 113 the the DET fbem-26173 161 114 portfolio portfolio NOUN fbem-26173 161 115 mode mode NOUN fbem-26173 161 116 , , PUNCT fbem-26173 161 117 the the DET fbem-26173 161 118 amount amount NOUN fbem-26173 161 119 of of ADP fbem-26173 161 120 investment investment NOUN fbem-26173 161 121 in in ADP fbem-26173 161 122 the the DET fbem-26173 161 123 stock stock NOUN fbem-26173 161 124 is be AUX fbem-26173 161 125 determined determine VERB fbem-26173 161 126 by by ADP fbem-26173 161 127 its its PRON fbem-26173 161 128 corresponding corresponding ADJ fbem-26173 161 129 weights weight NOUN fbem-26173 161 130 . . PUNCT fbem-26173 162 1 the the DET fbem-26173 162 2 amount amount NOUN fbem-26173 162 3 of of ADP fbem-26173 162 4 stock stock NOUN fbem-26173 162 5 investment investment NOUN fbem-26173 162 6 is be AUX fbem-26173 162 7 determined determine VERB fbem-26173 162 8 by by ADP fbem-26173 162 9 its its PRON fbem-26173 162 10 corresponding corresponding ADJ fbem-26173 162 11 weight weight NOUN fbem-26173 162 12 . . PUNCT fbem-26173 163 1 the the DET fbem-26173 163 2 four four NUM fbem-26173 163 3 portfolio portfolio NOUN fbem-26173 163 4 styles style NOUN fbem-26173 163 5 will will AUX fbem-26173 163 6 be be AUX fbem-26173 163 7 introduced introduce VERB fbem-26173 163 8 next next ADV fbem-26173 163 9 : : PUNCT fbem-26173 163 10 ( ( PUNCT fbem-26173 163 11 1 1 X fbem-26173 163 12 ) ) PUNCT fbem-26173 163 13 equal equal ADJ fbem-26173 163 14 weighted weight VERB fbem-26173 163 15 portfolio portfolio NOUN fbem-26173 163 16 : : PUNCT fbem-26173 163 17 the the DET fbem-26173 163 18 weight weight NOUN fbem-26173 163 19 of of ADP fbem-26173 163 20 each each DET fbem-26173 163 21 stock stock NOUN fbem-26173 163 22 is be AUX fbem-26173 163 23 assigned assign VERB fbem-26173 163 24 equally equally ADV fbem-26173 163 25 , , PUNCT fbem-26173 163 26 in in ADP fbem-26173 163 27 this this DET fbem-26173 163 28 paper paper NOUN fbem-26173 163 29 the the DET fbem-26173 163 30 weight weight NOUN fbem-26173 163 31 of of ADP fbem-26173 163 32 each each DET fbem-26173 163 33 stock stock NOUN fbem-26173 163 34 is be AUX fbem-26173 163 35 assigned assign VERB fbem-26173 163 36 as as ADP fbem-26173 163 37 0.25 0.25 NUM fbem-26173 163 38 , , PUNCT fbem-26173 163 39 this this DET fbem-26173 163 40 investment investment NOUN fbem-26173 163 41 method method NOUN fbem-26173 163 42 is be AUX fbem-26173 163 43 the the DET fbem-26173 163 44 simplest simple ADJ fbem-26173 163 45 . . PUNCT fbem-26173 164 1 ( ( PUNCT fbem-26173 164 2 2 2 X fbem-26173 164 3 ) ) PUNCT fbem-26173 164 4 market market NOUN fbem-26173 164 5 capitalisation capitalisation NOUN fbem-26173 164 6 weighted weight VERB fbem-26173 164 7 portfolio portfolio NOUN fbem-26173 164 8 : : PUNCT fbem-26173 164 9 the the DET fbem-26173 164 10 weight weight NOUN fbem-26173 164 11 of of ADP fbem-26173 164 12 this this DET fbem-26173 164 13 portfolio portfolio NOUN fbem-26173 164 14 is be AUX fbem-26173 164 15 determined determine VERB fbem-26173 164 16 by by ADP fbem-26173 164 17 the the DET fbem-26173 164 18 percentage percentage NOUN fbem-26173 164 19 of of ADP fbem-26173 164 20 total total ADJ fbem-26173 164 21 market market NOUN fbem-26173 164 22 capitalisation capitalisation NOUN fbem-26173 164 23 , , PUNCT fbem-26173 164 24 which which PRON fbem-26173 164 25 is be AUX fbem-26173 164 26 positively positively ADV fbem-26173 164 27 proportional proportional ADJ fbem-26173 164 28 . . PUNCT fbem-26173 165 1 ( ( PUNCT fbem-26173 165 2 3 3 X fbem-26173 165 3 ) ) PUNCT fbem-26173 165 4 investment investment NOUN fbem-26173 165 5 risk risk NOUN fbem-26173 165 6 minimising minimise VERB fbem-26173 165 7 portfolio portfolio NOUN fbem-26173 165 8 : : PUNCT fbem-26173 165 9 this this DET fbem-26173 165 10 strategy strategy NOUN fbem-26173 165 11 is be AUX fbem-26173 165 12 to to PART fbem-26173 165 13 select select VERB fbem-26173 165 14 the the DET fbem-26173 165 15 portfolio portfolio NOUN fbem-26173 165 16 with with ADP fbem-26173 165 17 the the DET fbem-26173 165 18 lowest low ADJ fbem-26173 165 19 investment investment NOUN fbem-26173 165 20 risk risk NOUN fbem-26173 165 21 and and CCONJ fbem-26173 165 22 the the DET fbem-26173 165 23 highest high ADJ fbem-26173 165 24 return return NOUN fbem-26173 165 25 under under ADP fbem-26173 165 26 that that DET fbem-26173 165 27 condition condition NOUN fbem-26173 165 28 . . PUNCT fbem-26173 166 1 in in ADP fbem-26173 166 2 this this DET fbem-26173 166 3 paper paper NOUN fbem-26173 166 4 , , PUNCT fbem-26173 166 5 according accord VERB fbem-26173 166 6 to to ADP fbem-26173 166 7 monte monte PROPN fbem-26173 166 8 carlo carlo PROPN fbem-26173 166 9 simulation simulation PROPN fbem-26173 166 10 markowitz markowitz PROPN fbem-26173 166 11 model model PROPN fbem-26173 166 12 , , PUNCT fbem-26173 166 13 the the DET fbem-26173 166 14 weight weight NOUN fbem-26173 166 15 of of ADP fbem-26173 166 16 each each DET fbem-26173 166 17 group group NOUN fbem-26173 166 18 is be AUX fbem-26173 166 19 randomly randomly ADV fbem-26173 166 20 generated generate VERB fbem-26173 166 21 , , PUNCT fbem-26173 166 22 calculate calculate VERB fbem-26173 166 23 the the DET fbem-26173 166 24 investment investment NOUN fbem-26173 166 25 return return NOUN fbem-26173 166 26 and and CCONJ fbem-26173 166 27 return return VERB fbem-26173 166 28 standard standard ADJ fbem-26173 166 29 deviation deviation NOUN fbem-26173 166 30 ( ( PUNCT fbem-26173 166 31 risk risk NOUN fbem-26173 166 32 standard standard ADJ fbem-26173 166 33 deviation deviation NOUN fbem-26173 166 34 ) ) PUNCT fbem-26173 166 35 of of ADP fbem-26173 166 36 each each DET fbem-26173 166 37 group group NOUN fbem-26173 166 38 , , PUNCT fbem-26173 166 39 the the DET fbem-26173 166 40 process process NOUN fbem-26173 166 41 is be AUX fbem-26173 166 42 repeated repeat VERB fbem-26173 166 43 a a DET fbem-26173 166 44 number number NOUN fbem-26173 166 45 of of ADP fbem-26173 166 46 times time NOUN fbem-26173 166 47 , , PUNCT fbem-26173 166 48 this this DET fbem-26173 166 49 paper paper NOUN fbem-26173 166 50 repeated repeat VERB fbem-26173 166 51 10,000 10,000 NUM fbem-26173 166 52 times time NOUN fbem-26173 166 53 , , PUNCT fbem-26173 166 54 the the DET fbem-26173 166 55 investment investment NOUN fbem-26173 166 56 return return NOUN fbem-26173 166 57 and and CCONJ fbem-26173 166 58 risk risk VERB fbem-26173 166 59 standard standard ADJ fbem-26173 166 60 deviation deviation NOUN fbem-26173 166 61 of of ADP fbem-26173 166 62 each each DET fbem-26173 166 63 group group NOUN fbem-26173 166 64 is be AUX fbem-26173 166 65 plotted plot VERB fbem-26173 166 66 as as ADP fbem-26173 166 67 a a DET fbem-26173 166 68 scatterplot scatterplot NOUN fbem-26173 166 69 ultimately ultimately ADV fbem-26173 166 70 as as SCONJ fbem-26173 166 71 shown show VERB fbem-26173 166 72 in in ADP fbem-26173 166 73 figure figure NOUN fbem-26173 166 74 12 12 NUM fbem-26173 166 75 , , PUNCT fbem-26173 166 76 and and CCONJ fbem-26173 166 77 then then ADV fbem-26173 166 78 find find VERB fbem-26173 166 79 out out ADP fbem-26173 166 80 the the DET fbem-26173 166 81 point point NOUN fbem-26173 166 82 of of ADP fbem-26173 166 83 the the DET fbem-26173 166 84 portfolio portfolio NOUN fbem-26173 166 85 with with ADP fbem-26173 166 86 the the DET fbem-26173 166 87 minimum minimum ADJ fbem-26173 166 88 risk risk NOUN fbem-26173 166 89 , , PUNCT fbem-26173 166 90 as as SCONJ fbem-26173 166 91 shown show VERB fbem-26173 166 92 in in ADP fbem-26173 166 93 figure figure NOUN fbem-26173 166 94 13 13 NUM fbem-26173 166 95 . . PUNCT fbem-26173 167 1 figure figure NOUN fbem-26173 167 2 12 12 NUM fbem-26173 167 3 . . PUNCT fbem-26173 168 1 markowitz markowitz PROPN fbem-26173 168 2 simulated simulate VERB fbem-26173 168 3 standard standard ADJ fbem-26173 168 4 deviation deviation NOUN fbem-26173 168 5 of of ADP fbem-26173 168 6 risk risk NOUN fbem-26173 168 7 and and CCONJ fbem-26173 168 8 scatterplot scatterplot NOUN fbem-26173 168 9 of of ADP fbem-26173 168 10 returns return NOUN fbem-26173 168 11 for for ADP fbem-26173 168 12 the the DET fbem-26173 168 13 10,000 10,000 NUM fbem-26173 168 14 group group NOUN fbem-26173 168 15 figure figure NOUN fbem-26173 168 16 13 13 NUM fbem-26173 168 17 . . PUNCT fbem-26173 169 1 the the DET fbem-26173 169 2 point point NOUN fbem-26173 169 3 in in ADP fbem-26173 169 4 the the DET fbem-26173 169 5 return return NOUN fbem-26173 169 6 - - PUNCT fbem-26173 169 7 risk risk NOUN fbem-26173 169 8 scatterplot scatterplot NOUN fbem-26173 169 9 that that PRON fbem-26173 169 10 minimizes minimize VERB fbem-26173 169 11 risk risk NOUN fbem-26173 169 12 ( ( PUNCT fbem-26173 169 13 4 4 X fbem-26173 169 14 ) ) PUNCT fbem-26173 169 15 sharpe sharpe PROPN fbem-26173 169 16 ratio ratio PROPN fbem-26173 169 17 portfolio portfolio NOUN fbem-26173 169 18 : : PUNCT fbem-26173 169 19 the the DET fbem-26173 169 20 sharpe sharpe PROPN fbem-26173 169 21 ratio ratio NOUN fbem-26173 169 22 is be AUX fbem-26173 169 23 a a DET fbem-26173 169 24 commonly commonly ADV fbem-26173 169 25 used use VERB fbem-26173 169 26 risk risk NOUN fbem-26173 169 27 - - PUNCT fbem-26173 169 28 adjusted adjust VERB fbem-26173 169 29 rate rate NOUN fbem-26173 169 30 of of ADP fbem-26173 169 31 return return NOUN fbem-26173 169 32 metric metric ADJ fbem-26173 169 33 designed design VERB fbem-26173 169 34 to to PART fbem-26173 169 35 measure measure VERB fbem-26173 169 36 the the DET fbem-26173 169 37 amount amount NOUN fbem-26173 169 38 of of ADP fbem-26173 169 39 excess excess ADJ fbem-26173 169 40 reward reward NOUN fbem-26173 169 41 that that PRON fbem-26173 169 42 can can AUX fbem-26173 169 43 be be AUX fbem-26173 169 44 achieved achieve VERB fbem-26173 169 45 for for ADP fbem-26173 169 46 each each DET fbem-26173 169 47 unit unit NOUN fbem-26173 169 48 of of ADP fbem-26173 169 49 total total ADJ fbem-26173 169 50 risk risk NOUN fbem-26173 169 51 taken take VERB fbem-26173 169 52 . . PUNCT fbem-26173 170 1 higher high ADJ fbem-26173 170 2 sharpe sharpe PROPN fbem-26173 170 3 ratios ratio NOUN fbem-26173 170 4 indicate indicate VERB fbem-26173 170 5 higher high ADJ fbem-26173 170 6 excess excess ADJ fbem-26173 170 7 returns return NOUN fbem-26173 170 8 per per ADP fbem-26173 170 9 unit unit NOUN fbem-26173 170 10 of of ADP fbem-26173 170 11 total total ADJ fbem-26173 170 12 risk risk NOUN fbem-26173 170 13 taken take VERB fbem-26173 170 14 and and CCONJ fbem-26173 170 15 better well ADJ fbem-26173 170 16 risk risk NOUN fbem-26173 170 17 - - PUNCT fbem-26173 170 18 adjusted adjust VERB fbem-26173 170 19 returns return NOUN fbem-26173 170 20 . . PUNCT fbem-26173 171 1 again again ADV fbem-26173 171 2 based base VERB fbem-26173 171 3 on on ADP fbem-26173 171 4 monte monte PROPN fbem-26173 171 5 carlo carlo PROPN fbem-26173 171 6 simulation simulation PROPN fbem-26173 171 7 of of ADP fbem-26173 171 8 the the DET fbem-26173 171 9 markowitz markowitz PROPN fbem-26173 171 10 model model PROPN fbem-26173 171 11 , , PUNCT fbem-26173 171 12 the the DET fbem-26173 171 13 sharpe sharpe PROPN fbem-26173 171 14 ratio ratio NOUN fbem-26173 171 15 corresponding correspond VERB fbem-26173 171 16 to to ADP fbem-26173 171 17 each each DET fbem-26173 171 18 group group NOUN fbem-26173 171 19 is be AUX fbem-26173 171 20 calculated calculate VERB fbem-26173 171 21 and and CCONJ fbem-26173 171 22 plotted plot VERB fbem-26173 171 23 as as ADP fbem-26173 171 24 the the DET fbem-26173 171 25 third third ADJ fbem-26173 171 26 variable variable NOUN fbem-26173 171 27 in in ADP fbem-26173 171 28 a a DET fbem-26173 171 29 scatter scatter NOUN fbem-26173 171 30 plot plot NOUN fbem-26173 171 31 of of ADP fbem-26173 171 32 return return NOUN fbem-26173 171 33 - - PUNCT fbem-26173 171 34 risk risk NOUN fbem-26173 171 35 , , PUNCT fbem-26173 171 36 as as SCONJ fbem-26173 171 37 shown show VERB fbem-26173 171 38 in in ADP fbem-26173 171 39 figure figure NOUN fbem-26173 171 40 14 14 NUM fbem-26173 171 41 . . PUNCT fbem-26173 172 1 the the DET fbem-26173 172 2 portfolio portfolio NOUN fbem-26173 172 3 corresponding correspond VERB fbem-26173 172 4 to to ADP fbem-26173 172 5 the the DET fbem-26173 172 6 largest large ADJ fbem-26173 172 7 sharpe sharpe NOUN fbem-26173 172 8 ratio ratio NOUN fbem-26173 172 9 is be AUX fbem-26173 172 10 then then ADV fbem-26173 172 11 found find VERB fbem-26173 172 12 , , PUNCT fbem-26173 172 13 as as SCONJ fbem-26173 172 14 shown show VERB fbem-26173 172 15 in in ADP fbem-26173 172 16 figure figure NOUN fbem-26173 172 17 15 15 NUM fbem-26173 172 18 . . PUNCT fbem-26173 173 1 figure figure NOUN fbem-26173 173 2 14 14 NUM fbem-26173 173 3 . . PUNCT fbem-26173 174 1 sharpe sharpe PROPN fbem-26173 174 2 ratio ratio PROPN fbem-26173 174 3 depicts depict VERB fbem-26173 174 4 scatterplot scatterplot NOUN fbem-26173 174 5 of of ADP fbem-26173 174 6 returns return NOUN fbem-26173 174 7 - - PUNCT fbem-26173 174 8 standard standard ADJ fbem-26173 174 9 deviation deviation NOUN fbem-26173 174 10 83 83 NUM fbem-26173 174 11 figure figure NOUN fbem-26173 174 12 15 15 NUM fbem-26173 174 13 . . PUNCT fbem-26173 175 1 the the DET fbem-26173 175 2 point point NOUN fbem-26173 175 3 in in ADP fbem-26173 175 4 the the DET fbem-26173 175 5 return return NOUN fbem-26173 175 6 - - PUNCT fbem-26173 175 7 risk risk NOUN fbem-26173 175 8 scatterplot scatterplot NOUN fbem-26173 175 9 with with ADP fbem-26173 175 10 the the DET fbem-26173 175 11 largest large ADJ fbem-26173 175 12 sharpe sharpe NOUN fbem-26173 175 13 ratio ratio NOUN fbem-26173 175 14 the the DET fbem-26173 175 15 final final ADJ fbem-26173 175 16 stock stock NOUN fbem-26173 175 17 weights weight NOUN fbem-26173 175 18 and and CCONJ fbem-26173 175 19 total total ADJ fbem-26173 175 20 returns return NOUN fbem-26173 175 21 corresponding correspond VERB fbem-26173 175 22 to to ADP fbem-26173 175 23 the the DET fbem-26173 175 24 four four NUM fbem-26173 175 25 portfolio portfolio NOUN fbem-26173 175 26 approaches approach NOUN fbem-26173 175 27 are be AUX fbem-26173 175 28 obtained obtain VERB fbem-26173 175 29 as as SCONJ fbem-26173 175 30 shown show VERB fbem-26173 175 31 in in ADP fbem-26173 175 32 table table NOUN fbem-26173 175 33 6 6 NUM fbem-26173 175 34 and and CCONJ fbem-26173 175 35 figure figure VERB fbem-26173 175 36 16 16 NUM fbem-26173 175 37 : : PUNCT fbem-26173 175 38 table table NOUN fbem-26173 175 39 6 6 NUM fbem-26173 175 40 . . PUNCT fbem-26173 176 1 equity equity NOUN fbem-26173 176 2 weights weight NOUN fbem-26173 176 3 corresponding correspond VERB fbem-26173 176 4 to to ADP fbem-26173 176 5 the the DET fbem-26173 176 6 four four NUM fbem-26173 176 7 portfolio portfolio NOUN fbem-26173 176 8 approaches approach VERB fbem-26173 176 9 stock stock NOUN fbem-26173 176 10 600050 600050 NUM fbem-26173 176 11 china china PROPN fbem-26173 176 12 unicom unicom PROPN fbem-26173 176 13 600941 600941 NUM fbem-26173 176 14 china china PROPN fbem-26173 176 15 mobile mobile PROPN fbem-26173 176 16 601628 601628 NUM fbem-26173 176 17 china china PROPN fbem-26173 176 18 life life PROPN fbem-26173 176 19 688111 688111 NUM fbem-26173 176 20 kingsoft kingsoft PROPN fbem-26173 176 21 office office PROPN fbem-26173 176 22 equal equal ADJ fbem-26173 176 23 weighted weight VERB fbem-26173 176 24 portfolio portfolio NOUN fbem-26173 176 25 0.25 0.25 NUM fbem-26173 176 26 0.25 0.25 NUM fbem-26173 176 27 0.25 0.25 NUM fbem-26173 176 28 0.25 0.25 NUM fbem-26173 176 29 market market NOUN fbem-26173 176 30 capitalizationweighted capitalizationweighte VERB fbem-26173 176 31 portfolio portfolio NOUN fbem-26173 176 32 0.04591 0.04591 NUM fbem-26173 176 33 965 965 NUM fbem-26173 177 1 0.56786 0.56786 NUM fbem-26173 177 2 866 866 NUM fbem-26173 177 3 0.34554 0.34554 NUM fbem-26173 177 4 871 871 NUM fbem-26173 177 5 0.040662 0.040662 NUM fbem-26173 177 6 98 98 NUM fbem-26173 177 7 investment investment NOUN fbem-26173 177 8 risk risk NOUN fbem-26173 177 9 minimization minimization NOUN fbem-26173 177 10 portfolio portfolio NOUN fbem-26173 177 11 0.43547 0.43547 NUM fbem-26173 177 12 867 867 NUM fbem-26173 177 13 0.36178 0.36178 NUM fbem-26173 177 14 263 263 NUM fbem-26173 177 15 0.11778 0.11778 NUM fbem-26173 177 16 557 557 NUM fbem-26173 177 17 0.084953 0.084953 NUM fbem-26173 177 18 12 12 NUM fbem-26173 177 19 sharpe sharpe PROPN fbem-26173 177 20 ratio ratio PROPN fbem-26173 177 21 portfolio portfolio NOUN fbem-26173 177 22 0.67685 0.67685 NUM fbem-26173 177 23 089 089 NUM fbem-26173 177 24 0.00755 0.00755 NUM fbem-26173 177 25 021 021 NUM fbem-26173 177 26 0.21619 0.21619 NUM fbem-26173 177 27 762 762 NUM fbem-26173 177 28 0.099401 0.099401 NUM fbem-26173 177 29 29 29 NUM fbem-26173 177 30 figure figure NOUN fbem-26173 177 31 16 16 NUM fbem-26173 177 32 . . PUNCT fbem-26173 178 1 total total ADJ fbem-26173 178 2 return return NOUN fbem-26173 178 3 graphs graph NOUN fbem-26173 178 4 corresponding correspond VERB fbem-26173 178 5 to to ADP fbem-26173 178 6 the the DET fbem-26173 178 7 four four NUM fbem-26173 178 8 portfolio portfolio NOUN fbem-26173 178 9 approaches approach NOUN fbem-26173 178 10 table table NOUN fbem-26173 178 11 7 7 NUM fbem-26173 178 12 . . PUNCT fbem-26173 179 1 portfolio portfolio NOUN fbem-26173 179 2 projected project VERB fbem-26173 179 3 total total ADJ fbem-26173 179 4 return return NOUN fbem-26173 179 5 stock stock NOUN fbem-26173 179 6 total total NOUN fbem-26173 179 7 return return NOUN fbem-26173 179 8 minimum minimum ADJ fbem-26173 179 9 total total ADJ fbem-26173 179 10 return return NOUN fbem-26173 179 11 maximum maximum ADJ fbem-26173 179 12 gross gross ADJ fbem-26173 179 13 yield yield NOUN fbem-26173 179 14 equal equal ADJ fbem-26173 179 15 weighted weight VERB fbem-26173 179 16 portfolio portfolio NOUN fbem-26173 179 17 33.64 33.64 NUM fbem-26173 179 18 % % NOUN fbem-26173 179 19 -5.60 -5.60 NUM fbem-26173 179 20 % % NOUN fbem-26173 179 21 35.80 35.80 NUM fbem-26173 179 22 % % NOUN fbem-26173 179 23 market market NOUN fbem-26173 179 24 capitalizationweighted capitalizationweighte VERB fbem-26173 179 25 portfolio portfolio NOUN fbem-26173 179 26 25.62 25.62 NUM fbem-26173 179 27 % % NOUN fbem-26173 179 28 -4.07 -4.07 NUM fbem-26173 179 29 % % NOUN fbem-26173 179 30 29.27 29.27 NUM fbem-26173 179 31 % % NOUN fbem-26173 179 32 investment investment NOUN fbem-26173 179 33 risk risk NOUN fbem-26173 179 34 minimization minimization NOUN fbem-26173 179 35 portfolio portfolio NOUN fbem-26173 179 36 32.64 32.64 NUM fbem-26173 179 37 % % NOUN fbem-26173 180 1 -4.59 -4.59 NUM fbem-26173 181 1 % % NOUN fbem-26173 181 2 34.33 34.33 NUM fbem-26173 181 3 % % NOUN fbem-26173 181 4 sharpe sharpe PROPN fbem-26173 181 5 ratio ratio PROPN fbem-26173 181 6 portfolio portfolio NOUN fbem-26173 181 7 43.72 43.72 NUM fbem-26173 181 8 % % NOUN fbem-26173 181 9 -1.93 -1.93 NUM fbem-26173 181 10 % % NOUN fbem-26173 181 11 44.90 44.90 NUM fbem-26173 181 12 % % NOUN fbem-26173 181 13 as as SCONJ fbem-26173 181 14 can can AUX fbem-26173 181 15 be be AUX fbem-26173 181 16 seen see VERB fbem-26173 181 17 from from ADP fbem-26173 181 18 table table NOUN fbem-26173 181 19 7 7 NUM fbem-26173 181 20 , , PUNCT fbem-26173 181 21 the the DET fbem-26173 181 22 highest high ADJ fbem-26173 181 23 total total ADJ fbem-26173 181 24 returns return NOUN fbem-26173 181 25 obtained obtain VERB fbem-26173 181 26 through through ADP fbem-26173 181 27 the the DET fbem-26173 181 28 four four NUM fbem-26173 181 29 portfolio portfolio NOUN fbem-26173 181 30 approaches approach NOUN fbem-26173 181 31 are be AUX fbem-26173 181 32 basically basically ADV fbem-26173 181 33 above above ADP fbem-26173 181 34 30 30 NUM fbem-26173 181 35 % % NOUN fbem-26173 181 36 , , PUNCT fbem-26173 181 37 with with ADP fbem-26173 181 38 losses loss NOUN fbem-26173 181 39 controlled control VERB fbem-26173 181 40 within within ADP fbem-26173 181 41 5 5 NUM fbem-26173 181 42 % % NOUN fbem-26173 181 43 , , PUNCT fbem-26173 181 44 indicating indicate VERB fbem-26173 181 45 that that SCONJ fbem-26173 181 46 the the DET fbem-26173 181 47 returns return NOUN fbem-26173 181 48 that that PRON fbem-26173 181 49 can can AUX fbem-26173 181 50 be be AUX fbem-26173 181 51 obtained obtain VERB fbem-26173 181 52 from from ADP fbem-26173 181 53 these these DET fbem-26173 181 54 four four NUM fbem-26173 181 55 portfolios portfolio NOUN fbem-26173 181 56 are be AUX fbem-26173 181 57 all all PRON fbem-26173 181 58 relatively relatively ADV fbem-26173 181 59 stable stable ADJ fbem-26173 181 60 and and CCONJ fbem-26173 181 61 less less ADV fbem-26173 181 62 risky risky ADJ fbem-26173 181 63 . . PUNCT fbem-26173 182 1 returns return NOUN fbem-26173 182 2 from from ADP fbem-26173 182 3 figure figure NOUN fbem-26173 182 4 16 16 NUM fbem-26173 182 5 can can AUX fbem-26173 182 6 be be AUX fbem-26173 182 7 seen see VERB fbem-26173 182 8 , , PUNCT fbem-26173 182 9 in in ADP fbem-26173 182 10 these these DET fbem-26173 182 11 four four NUM fbem-26173 182 12 kinds kind NOUN fbem-26173 182 13 of of ADP fbem-26173 182 14 investment investment NOUN fbem-26173 182 15 portfolio portfolio NOUN fbem-26173 182 16 approach approach NOUN fbem-26173 182 17 , , PUNCT fbem-26173 182 18 portfoliof portfoliof NOUN fbem-26173 182 19 investment investment NOUN fbem-26173 182 20 risk risk NOUN fbem-26173 182 21 of of ADP fbem-26173 182 22 the the DET fbem-26173 182 23 smallest small ADJ fbem-26173 182 24 portfolio portfolio NOUN fbem-26173 182 25 in in ADP fbem-26173 182 26 these these DET fbem-26173 182 27 four four NUM fbem-26173 182 28 portfolio portfolio NOUN fbem-26173 182 29 approach approach NOUN fbem-26173 182 30 to to PART fbem-26173 182 31 obtain obtain VERB fbem-26173 182 32 the the DET fbem-26173 182 33 return return NOUN fbem-26173 182 34 has have AUX fbem-26173 182 35 been be AUX fbem-26173 182 36 relatively relatively ADV fbem-26173 182 37 small small ADJ fbem-26173 182 38 , , PUNCT fbem-26173 182 39 in in ADP fbem-26173 182 40 the the DET fbem-26173 182 41 long long ADJ fbem-26173 182 42 run run NOUN fbem-26173 182 43 , , PUNCT fbem-26173 182 44 portfolioiox portfolioiox PROPN fbem-26173 182 45 sharpe sharpe PROPN fbem-26173 182 46 ratio ratio PROPN fbem-26173 182 47 investment investment PROPN fbem-26173 182 48 portfolio portfolio NOUN fbem-26173 182 49 is be AUX fbem-26173 182 50 relatively relatively ADV fbem-26173 182 51 more more ADV fbem-26173 182 52 potential potential ADJ fbem-26173 182 53 , , PUNCT fbem-26173 182 54 in in ADP fbem-26173 182 55 the the DET fbem-26173 182 56 later later ADJ fbem-26173 182 57 stage stage NOUN fbem-26173 182 58 can can AUX fbem-26173 182 59 obtain obtain VERB fbem-26173 182 60 a a DET fbem-26173 182 61 relatively relatively ADV fbem-26173 182 62 high high ADJ fbem-26173 182 63 return return NOUN fbem-26173 182 64 . . PUNCT fbem-26173 183 1 investors investor NOUN fbem-26173 183 2 can can AUX fbem-26173 183 3 therefore therefore ADV fbem-26173 183 4 choose choose VERB fbem-26173 183 5 the the DET fbem-26173 183 6 right right ADJ fbem-26173 183 7 portfolio portfolio NOUN fbem-26173 183 8 to to PART fbem-26173 183 9 trade trade VERB fbem-26173 183 10 at at ADP fbem-26173 183 11 the the DET fbem-26173 183 12 right right ADJ fbem-26173 183 13 time time NOUN fbem-26173 183 14 depending depend VERB fbem-26173 183 15 on on ADP fbem-26173 183 16 the the DET fbem-26173 183 17 length length NOUN fbem-26173 183 18 of of ADP fbem-26173 183 19 time time NOUN fbem-26173 183 20 they they PRON fbem-26173 183 21 have have AUX fbem-26173 183 22 been be AUX fbem-26173 183 23 buying buy VERB fbem-26173 183 24 shares share NOUN fbem-26173 183 25 . . PUNCT fbem-26173 184 1 4 4 X fbem-26173 184 2 . . X fbem-26173 184 3 conclusion conclusion VERB fbem-26173 184 4 this this DET fbem-26173 184 5 paper paper NOUN fbem-26173 184 6 addresses address NOUN fbem-26173 184 7 the the DET fbem-26173 184 8 time time NOUN fbem-26173 184 9 series series PROPN fbem-26173 184 10 stock stock NOUN fbem-26173 184 11 price price NOUN fbem-26173 184 12 prediction prediction NOUN fbem-26173 184 13 problem problem NOUN fbem-26173 184 14 . . PUNCT fbem-26173 185 1 firstly firstly ADV fbem-26173 185 2 , , PUNCT fbem-26173 185 3 the the DET fbem-26173 185 4 most most ADV fbem-26173 185 5 important important ADJ fbem-26173 185 6 10 10 NUM fbem-26173 185 7 factors factor NOUN fbem-26173 185 8 affecting affect VERB fbem-26173 185 9 stock stock NOUN fbem-26173 185 10 returns return NOUN fbem-26173 185 11 are be AUX fbem-26173 185 12 selected select VERB fbem-26173 185 13 through through ADP fbem-26173 185 14 random random ADJ fbem-26173 185 15 forest forest NOUN fbem-26173 185 16 , , PUNCT fbem-26173 185 17 and and CCONJ fbem-26173 185 18 the the DET fbem-26173 185 19 constructed construct VERB fbem-26173 185 20 xgboost xgboost NOUN fbem-26173 185 21 regression regression NOUN fbem-26173 185 22 model model NOUN fbem-26173 185 23 is be AUX fbem-26173 185 24 used use VERB fbem-26173 185 25 for for ADP fbem-26173 185 26 four four NUM fbem-26173 185 27 highquality highquality NOUN fbem-26173 185 28 stocks stock NOUN fbem-26173 185 29 selection selection NOUN fbem-26173 185 30 . . PUNCT fbem-26173 186 1 secondly secondly ADV fbem-26173 186 2 , , PUNCT fbem-26173 186 3 for for ADP fbem-26173 186 4 the the DET fbem-26173 186 5 prediction prediction NOUN fbem-26173 186 6 of of ADP fbem-26173 186 7 stock stock NOUN fbem-26173 186 8 closing closing NOUN fbem-26173 186 9 price price NOUN fbem-26173 186 10 , , PUNCT fbem-26173 186 11 this this DET fbem-26173 186 12 paper paper NOUN fbem-26173 186 13 proposes propose VERB fbem-26173 186 14 a a DET fbem-26173 186 15 new new ADJ fbem-26173 186 16 framework framework NOUN fbem-26173 186 17 cnnbilstm cnnbilstm NOUN fbem-26173 186 18 - - PUNCT fbem-26173 186 19 attention attention NOUN fbem-26173 186 20 model model NOUN fbem-26173 186 21 based base VERB fbem-26173 186 22 on on ADP fbem-26173 186 23 deep deep ADJ fbem-26173 186 24 learning learning NOUN fbem-26173 186 25 method method NOUN fbem-26173 186 26 , , PUNCT fbem-26173 186 27 which which PRON fbem-26173 186 28 extracts extract VERB fbem-26173 186 29 the the DET fbem-26173 186 30 historical historical ADJ fbem-26173 186 31 information information NOUN fbem-26173 186 32 features feature NOUN fbem-26173 186 33 of of ADP fbem-26173 186 34 the the DET fbem-26173 186 35 stock stock NOUN fbem-26173 186 36 through through ADP fbem-26173 186 37 cnn cnn PROPN fbem-26173 186 38 , , PUNCT fbem-26173 186 39 and and CCONJ fbem-26173 186 40 trains train NOUN fbem-26173 186 41 and and CCONJ fbem-26173 186 42 predicts predict VERB fbem-26173 186 43 the the DET fbem-26173 186 44 stock stock NOUN fbem-26173 186 45 dataset dataset VERB fbem-26173 186 46 by by ADP fbem-26173 186 47 using use VERB fbem-26173 186 48 the the DET fbem-26173 186 49 bidirectional bidirectional ADJ fbem-26173 186 50 lstm lstm NOUN fbem-26173 186 51 network network NOUN fbem-26173 186 52 , , PUNCT fbem-26173 186 53 meanwhile meanwhile ADV fbem-26173 186 54 , , PUNCT fbem-26173 186 55 an an DET fbem-26173 186 56 attention attention NOUN fbem-26173 186 57 mechanism mechanism NOUN fbem-26173 186 58 is be AUX fbem-26173 186 59 added add VERB fbem-26173 186 60 in in ADP fbem-26173 186 61 the the DET fbem-26173 186 62 process process NOUN fbem-26173 186 63 of of ADP fbem-26173 186 64 training training NOUN fbem-26173 186 65 to to PART fbem-26173 186 66 give give VERB fbem-26173 186 67 every every DET fbem-26173 186 68 stock stock NOUN fbem-26173 186 69 features feature NOUN fbem-26173 186 70 , , PUNCT fbem-26173 186 71 so so SCONJ fbem-26173 186 72 that that SCONJ fbem-26173 186 73 the the DET fbem-26173 186 74 network network NOUN fbem-26173 186 75 can can AUX fbem-26173 186 76 learn learn VERB fbem-26173 186 77 the the DET fbem-26173 186 78 feature feature NOUN fbem-26173 186 79 information information NOUN fbem-26173 186 80 more more ADV fbem-26173 186 81 effectively effectively ADV fbem-26173 186 82 and and CCONJ fbem-26173 186 83 improve improve VERB fbem-26173 186 84 the the DET fbem-26173 186 85 prediction prediction NOUN fbem-26173 186 86 accuracy accuracy NOUN fbem-26173 186 87 . . PUNCT fbem-26173 187 1 finally finally ADV fbem-26173 187 2 , , PUNCT fbem-26173 187 3 the the DET fbem-26173 187 4 cumulative cumulative ADJ fbem-26173 187 5 returns return NOUN fbem-26173 187 6 of of ADP fbem-26173 187 7 the the DET fbem-26173 187 8 four four NUM fbem-26173 187 9 selected select VERB fbem-26173 187 10 stocks stock NOUN fbem-26173 187 11 are be AUX fbem-26173 187 12 calculated calculate VERB fbem-26173 187 13 and and CCONJ fbem-26173 187 14 backtested backteste VERB fbem-26173 187 15 using use VERB fbem-26173 187 16 the the DET fbem-26173 187 17 single single ADJ fbem-26173 187 18 stock stock NOUN fbem-26173 187 19 quantitative quantitative ADJ fbem-26173 187 20 trading trading NOUN fbem-26173 187 21 strategy strategy NOUN fbem-26173 187 22 of of ADP fbem-26173 187 23 macd macd NOUN fbem-26173 187 24 , , PUNCT fbem-26173 187 25 and and CCONJ fbem-26173 187 26 the the DET fbem-26173 187 27 highest high ADJ fbem-26173 187 28 total total ADJ fbem-26173 187 29 returns return NOUN fbem-26173 187 30 of of ADP fbem-26173 187 31 most most ADJ fbem-26173 187 32 stocks stock NOUN fbem-26173 187 33 are be AUX fbem-26173 187 34 over over ADP fbem-26173 187 35 40 40 NUM fbem-26173 187 36 % % NOUN fbem-26173 187 37 , , PUNCT fbem-26173 187 38 and and CCONJ fbem-26173 187 39 the the DET fbem-26173 187 40 losses loss NOUN fbem-26173 187 41 are be AUX fbem-26173 187 42 controlled control VERB fbem-26173 187 43 within within ADP fbem-26173 187 44 10 10 NUM fbem-26173 187 45 % % NOUN fbem-26173 187 46 , , PUNCT fbem-26173 187 47 which which PRON fbem-26173 187 48 indicates indicate VERB fbem-26173 187 49 that that SCONJ fbem-26173 187 50 the the DET fbem-26173 187 51 trading trading NOUN fbem-26173 187 52 strategy strategy NOUN fbem-26173 187 53 based base VERB fbem-26173 187 54 on on ADP fbem-26173 187 55 this this DET fbem-26173 187 56 model model NOUN fbem-26173 187 57 is be AUX fbem-26173 187 58 effective effective ADJ fbem-26173 187 59 and and CCONJ fbem-26173 187 60 feasible feasible ADJ fbem-26173 187 61 . . PUNCT fbem-26173 188 1 moreover moreover ADV fbem-26173 188 2 , , PUNCT fbem-26173 188 3 when when SCONJ fbem-26173 188 4 the the DET fbem-26173 188 5 stock stock NOUN fbem-26173 188 6 portfolio portfolio NOUN fbem-26173 188 7 returns return NOUN fbem-26173 188 8 are be AUX fbem-26173 188 9 studied study VERB fbem-26173 188 10 , , PUNCT fbem-26173 188 11 the the DET fbem-26173 188 12 portfoliox portfoliox NOUN fbem-26173 188 13 sharpe sharpe PROPN fbem-26173 188 14 ratio ratio PROPN fbem-26173 188 15 portfolio portfolio PROPN fbem-26173 188 16 has have VERB fbem-26173 188 17 a a DET fbem-26173 188 18 relatively relatively ADV fbem-26173 188 19 higher high ADJ fbem-26173 188 20 potential potential NOUN fbem-26173 188 21 to to PART fbem-26173 188 22 achieve achieve VERB fbem-26173 188 23 higher high ADJ fbem-26173 188 24 returns return NOUN fbem-26173 188 25 at at ADP fbem-26173 188 26 a a DET fbem-26173 188 27 later later ADJ fbem-26173 188 28 stage stage NOUN fbem-26173 188 29 . . PUNCT fbem-26173 189 1 this this PRON fbem-26173 189 2 shows show VERB fbem-26173 189 3 that that SCONJ fbem-26173 189 4 the the DET fbem-26173 189 5 proposed propose VERB fbem-26173 189 6 model model NOUN fbem-26173 189 7 has have VERB fbem-26173 189 8 a a DET fbem-26173 189 9 certain certain ADJ fbem-26173 189 10 predictive predictive ADJ fbem-26173 189 11 ability ability NOUN fbem-26173 189 12 for for ADP fbem-26173 189 13 stock stock NOUN fbem-26173 189 14 prices price NOUN fbem-26173 189 15 , , PUNCT fbem-26173 189 16 which which PRON fbem-26173 189 17 can can AUX fbem-26173 189 18 be be AUX fbem-26173 189 19 used use VERB fbem-26173 189 20 as as ADP fbem-26173 189 21 a a DET fbem-26173 189 22 reference reference NOUN fbem-26173 189 23 for for SCONJ fbem-26173 189 24 investors investor NOUN fbem-26173 189 25 to to PART fbem-26173 189 26 make make VERB fbem-26173 189 27 faster fast ADJ fbem-26173 189 28 decisions decision NOUN fbem-26173 189 29 when when SCONJ fbem-26173 189 30 investing invest VERB fbem-26173 189 31 and and CCONJ fbem-26173 189 32 reduce reduce VERB fbem-26173 189 33 the the DET fbem-26173 189 34 risk risk NOUN fbem-26173 189 35 of of ADP fbem-26173 189 36 investment investment NOUN fbem-26173 189 37 . . PUNCT fbem-26173 190 1 acknowledgment acknowledgment NOUN fbem-26173 190 2 this this DET fbem-26173 190 3 work work NOUN fbem-26173 190 4 is be AUX fbem-26173 190 5 supported support VERB fbem-26173 190 6 by by ADP fbem-26173 190 7 the the DET fbem-26173 190 8 natural natural ADJ fbem-26173 190 9 science science NOUN fbem-26173 190 10 fund fund NOUN fbem-26173 190 11 for for ADP fbem-26173 190 12 colleges college NOUN fbem-26173 190 13 and and CCONJ fbem-26173 190 14 universities university NOUN fbem-26173 190 15 , , PUNCT fbem-26173 190 16 department department NOUN fbem-26173 190 17 of of ADP fbem-26173 190 18 education education NOUN fbem-26173 190 19 of of ADP fbem-26173 190 20 anhui anhui PROPN fbem-26173 190 21 province province PROPN fbem-26173 190 22 ( ( PUNCT fbem-26173 190 23 kj2021a0481 kj2021a0481 PROPN fbem-26173 190 24 ) ) PUNCT fbem-26173 190 25 , , PUNCT fbem-26173 190 26 and and CCONJ fbem-26173 190 27 anhui anhui PROPN fbem-26173 190 28 university university PROPN fbem-26173 190 29 of of ADP fbem-26173 190 30 finance finance PROPN fbem-26173 190 31 and and CCONJ fbem-26173 190 32 economics economic NOUN fbem-26173 190 33 graduate graduate NOUN fbem-26173 190 34 student student NOUN fbem-26173 190 35 research research NOUN fbem-26173 190 36 and and CCONJ fbem-26173 190 37 innovation innovation NOUN fbem-26173 190 38 fund fund NOUN fbem-26173 190 39 project project NOUN fbem-26173 190 40 ( ( PUNCT fbem-26173 190 41 acyc2023173 acyc2023173 PROPN fbem-26173 190 42 ) ) PUNCT fbem-26173 190 43 . . PUNCT fbem-26173 191 1 references reference NOUN fbem-26173 191 2 [ [ X fbem-26173 191 3 1 1 NUM fbem-26173 191 4 ] ] PUNCT fbem-26173 191 5 box box NOUN fbem-26173 191 6 g g PROPN fbem-26173 191 7 e e PROPN fbem-26173 191 8 p p PROPN fbem-26173 191 9 , , PUNCT fbem-26173 191 10 jenkins jenkins PROPN fbem-26173 191 11 g g PROPN fbem-26173 191 12 m. m. NOUN fbem-26173 191 13 time time PROPN fbem-26173 191 14 series series PROPN fbem-26173 191 15 analysis analysis NOUN fbem-26173 191 16 : : PUNCT fbem-26173 191 17 forecasting forecasting NOUN fbem-26173 191 18 and and CCONJ fbem-26173 191 19 control control NOUN fbem-26173 192 1 [ [ X fbem-26173 193 1 m m X fbem-26173 193 2 ] ] X fbem-26173 193 3 . . PUNCT fbem-26173 194 1 san san PROPN fbem-26173 194 2 francisco francisco PROPN fbem-26173 194 3 : : PUNCT fbem-26173 194 4 holden holden PROPN fbem-26173 194 5 - - PUNCT fbem-26173 194 6 day day NOUN fbem-26173 194 7 , , PUNCT fbem-26173 194 8 1970 1970 NUM fbem-26173 194 9 . . PUNCT fbem-26173 195 1 [ [ X fbem-26173 195 2 2 2 NUM fbem-26173 195 3 ] ] PUNCT fbem-26173 195 4 ariyo ariyo NOUN fbem-26173 195 5 a a DET fbem-26173 195 6 a a NOUN fbem-26173 195 7 , , PUNCT fbem-26173 195 8 adewumi adewumi NOUN fbem-26173 195 9 a a DET fbem-26173 195 10 o o NOUN fbem-26173 195 11 , , PUNCT fbem-26173 195 12 ayo ayo PROPN fbem-26173 195 13 c c PROPN fbem-26173 195 14 k. k. PROPN fbem-26173 195 15 stock stock PROPN fbem-26173 195 16 price price NOUN fbem-26173 195 17 prediction prediction NOUN fbem-26173 195 18 using use VERB fbem-26173 195 19 the the DET fbem-26173 195 20 arimamodel arimamodel NOUN fbem-26173 195 21 [ [ X fbem-26173 195 22 c]//2014 c]//2014 NOUN fbem-26173 195 23 uksim uksim ADJ fbem-26173 195 24 - - PUNCT fbem-26173 195 25 amss amss NOUN fbem-26173 195 26 16th 16th ADJ fbem-26173 195 27 intemational intemational ADJ fbem-26173 195 28 conference conference NOUN fbem-26173 195 29 on on ADP fbem-26173 195 30 computermodelling computermodelling NOUN fbem-26173 195 31 and and CCONJ fbem-26173 195 32 simulation simulation NOUN fbem-26173 195 33 . . PUNCT fbem-26173 196 1 ieee ieee PROPN fbem-26173 196 2 , , PUNCT fbem-26173 196 3 2014:106 2014:106 NUM fbem-26173 196 4 - - PUNCT fbem-26173 196 5 112 112 NUM fbem-26173 196 6 . . PUNCT fbem-26173 197 1 [ [ X fbem-26173 197 2 3 3 X fbem-26173 197 3 ] ] X fbem-26173 197 4 wen wen PROPN fbem-26173 197 5 f f PROPN fbem-26173 197 6 , , PUNCT fbem-26173 197 7 xiao xiao PROPN fbem-26173 197 8 j j PROPN fbem-26173 197 9 , , PUNCT fbem-26173 197 10 zhifang zhifang PROPN fbem-26173 197 11 h h PROPN fbem-26173 197 12 e e PROPN fbem-26173 197 13 , , PUNCT fbem-26173 197 14 et et PROPN fbem-26173 197 15 al al PROPN fbem-26173 197 16 . . PUNCT fbem-26173 197 17 stock stock NOUN fbem-26173 197 18 price price NOUN fbem-26173 197 19 prediction prediction NOUN fbem-26173 197 20 based base VERB fbem-26173 197 21 on on ADP fbem-26173 197 22 ssa ssa NOUN fbem-26173 197 23 and and CCONJ fbem-26173 197 24 svm svm VERB fbem-26173 197 25 [ [ X fbem-26173 197 26 j j X fbem-26173 197 27 ] ] X fbem-26173 197 28 . . PUNCT fbem-26173 198 1 procedia procedia PROPN fbem-26173 198 2 computer computer NOUN fbem-26173 198 3 science science NOUN fbem-26173 198 4 , , PUNCT fbem-26173 198 5 2014 2014 NUM fbem-26173 198 6 , , PUNCT fbem-26173 198 7 31 31 NUM fbem-26173 198 8 : : SYM fbem-26173 198 9 625 625 NUM fbem-26173 198 10 - - SYM fbem-26173 198 11 631 631 NUM fbem-26173 198 12 . . PUNCT fbem-26173 199 1 [ [ X fbem-26173 199 2 4 4 X fbem-26173 199 3 ] ] PUNCT fbem-26173 199 4 lahmiri lahmiri PROPN fbem-26173 199 5 s. s. PROPN fbem-26173 199 6 minute minute PROPN fbem-26173 199 7 - - PUNCT fbem-26173 199 8 ahead ahead NOUN fbem-26173 199 9 stock stock NOUN fbem-26173 199 10 price price NOUN fbem-26173 199 11 forecasting forecasting NOUN fbem-26173 199 12 based base VERB fbem-26173 199 13 on on ADP fbem-26173 199 14 singular singular PROPN fbem-26173 199 15 spectrum spectrum NOUN fbem-26173 199 16 analysis analysis NOUN fbem-26173 199 17 and and CCONJ fbem-26173 199 18 support support VERB fbem-26173 199 19 vector vector NOUN fbem-26173 199 20 regression regression NOUN fbem-26173 199 21 [ [ X fbem-26173 199 22 j j X fbem-26173 199 23 ] ] X fbem-26173 199 24 . . PUNCT fbem-26173 200 1 applied apply VERB fbem-26173 200 2 mathematics mathematic NOUN fbem-26173 200 3 and and CCONJ fbem-26173 200 4 computation computation NOUN fbem-26173 200 5 , , PUNCT fbem-26173 200 6 2018 2018 NUM fbem-26173 200 7 , , PUNCT fbem-26173 200 8 320 320 NUM fbem-26173 200 9 : : PUNCT fbem-26173 200 10 444 444 NUM fbem-26173 200 11 - - SYM fbem-26173 200 12 451 451 NUM fbem-26173 200 13 . . PUNCT fbem-26173 201 1 84 84 NUM fbem-26173 202 1 [ [ X fbem-26173 202 2 5 5 NUM fbem-26173 202 3 ] ] PUNCT fbem-26173 202 4 md md PROPN fbem-26173 202 5 a a DET fbem-26173 202 6 q q PROPN fbem-26173 202 7 , , PUNCT fbem-26173 202 8 kapoor kapoor PROPN fbem-26173 202 9 s s PROPN fbem-26173 202 10 , , PUNCT fbem-26173 202 11 av av PROPN fbem-26173 202 12 c c PROPN fbem-26173 202 13 j j PROPN fbem-26173 202 14 , , PUNCT fbem-26173 202 15 et et PROPN fbem-26173 203 1 al al PROPN fbem-26173 203 2 . . PROPN fbem-26173 203 3 novel novel ADJ fbem-26173 203 4 optimization optimization NOUN fbem-26173 203 5 approach approach NOUN fbem-26173 203 6 for for ADP fbem-26173 203 7 stock stock NOUN fbem-26173 203 8 price price NOUN fbem-26173 203 9 forecasting forecasting NOUN fbem-26173 203 10 using use VERB fbem-26173 203 11 multi multi ADJ fbem-26173 203 12 - - ADJ fbem-26173 203 13 layered layered ADJ fbem-26173 203 14 sequential sequential ADJ fbem-26173 203 15 lstm lstm NOUN fbem-26173 204 1 [ [ X fbem-26173 204 2 j j X fbem-26173 204 3 ] ] X fbem-26173 204 4 . . PUNCT fbem-26173 205 1 applied apply VERB fbem-26173 205 2 soft soft ADJ fbem-26173 205 3 computing computing NOUN fbem-26173 205 4 , , PUNCT fbem-26173 205 5 2023 2023 NUM fbem-26173 205 6 , , PUNCT fbem-26173 205 7 134 134 NUM fbem-26173 205 8 : : SYM fbem-26173 205 9 109830 109830 NUM fbem-26173 205 10 . . PUNCT fbem-26173 206 1 [ [ X fbem-26173 206 2 6 6 NUM fbem-26173 206 3 ] ] X fbem-26173 206 4 ren ren PROPN fbem-26173 206 5 s s PROPN fbem-26173 206 6 , , PUNCT fbem-26173 206 7 wang wang PROPN fbem-26173 206 8 x x PROPN fbem-26173 206 9 , , PUNCT fbem-26173 206 10 zhou zhou PROPN fbem-26173 206 11 x x PROPN fbem-26173 206 12 , , PUNCT fbem-26173 206 13 et et PROPN fbem-26173 206 14 al al PROPN fbem-26173 206 15 . . PUNCT fbem-26173 207 1 a a DET fbem-26173 207 2 novel novel ADJ fbem-26173 207 3 hybrid hybrid NOUN fbem-26173 207 4 model model NOUN fbem-26173 207 5 for for ADP fbem-26173 207 6 stock stock NOUN fbem-26173 207 7 price price NOUN fbem-26173 207 8 forecasting forecasting NOUN fbem-26173 207 9 integrating integrate VERB fbem-26173 207 10 encoder encoder NOUN fbem-26173 207 11 forest forest NOUN fbem-26173 207 12 and and CCONJ fbem-26173 207 13 informer informer NOUN fbem-26173 208 1 [ [ X fbem-26173 208 2 j j X fbem-26173 208 3 ] ] X fbem-26173 208 4 . . PUNCT fbem-26173 209 1 expert expert NOUN fbem-26173 209 2 systems system NOUN fbem-26173 209 3 with with ADP fbem-26173 209 4 applications application NOUN fbem-26173 209 5 , , PUNCT fbem-26173 209 6 2023 2023 NUM fbem-26173 209 7 , , PUNCT fbem-26173 209 8 234 234 NUM fbem-26173 209 9 : : SYM fbem-26173 209 10 121080 121080 NUM fbem-26173 209 11 . . PUNCT fbem-26173 210 1 [ [ X fbem-26173 210 2 7 7 NUM fbem-26173 210 3 ] ] X fbem-26173 210 4 long long PROPN fbem-26173 210 5 w w PROPN fbem-26173 210 6 , , PUNCT fbem-26173 210 7 gao gao PROPN fbem-26173 210 8 j j PROPN fbem-26173 210 9 , , PUNCT fbem-26173 210 10 bai bai PROPN fbem-26173 210 11 k k PROPN fbem-26173 210 12 , , PUNCT fbem-26173 210 13 et et PROPN fbem-26173 210 14 al al PROPN fbem-26173 210 15 . . PUNCT fbem-26173 211 1 a a DET fbem-26173 211 2 hybrid hybrid ADJ fbem-26173 211 3 model model NOUN fbem-26173 211 4 for for ADP fbem-26173 211 5 stock stock NOUN fbem-26173 211 6 price price NOUN fbem-26173 211 7 prediction prediction NOUN fbem-26173 211 8 based base VERB fbem-26173 211 9 on on ADP fbem-26173 211 10 multi multi ADJ fbem-26173 211 11 - - ADJ fbem-26173 211 12 view view ADJ fbem-26173 211 13 heterogeneous heterogeneous ADJ fbem-26173 211 14 data datum NOUN fbem-26173 211 15 [ [ X fbem-26173 211 16 j j X fbem-26173 211 17 ] ] X fbem-26173 211 18 . . PUNCT fbem-26173 212 1 financial financial ADJ fbem-26173 212 2 innovation innovation NOUN fbem-26173 212 3 , , PUNCT fbem-26173 212 4 2024 2024 NUM fbem-26173 212 5 , , PUNCT fbem-26173 212 6 10(1 10(1 NUM fbem-26173 212 7 ): ): PUNCT fbem-26173 212 8 48 48 NUM fbem-26173 212 9 . . PUNCT fbem-26173 213 1 [ [ X fbem-26173 213 2 8 8 NUM fbem-26173 213 3 ] ] PUNCT fbem-26173 213 4 breiman breiman NOUN fbem-26173 213 5 l. l. PROPN fbem-26173 213 6 random random ADJ fbem-26173 213 7 forests forest NOUN fbem-26173 214 1 [ [ X fbem-26173 214 2 j j X fbem-26173 214 3 ] ] X fbem-26173 214 4 . . PUNCT fbem-26173 215 1 machine machine NOUN fbem-26173 215 2 learning learning NOUN fbem-26173 215 3 , , PUNCT fbem-26173 215 4 2001,45(1):5 2001,45(1):5 NUM fbem-26173 215 5 - - SYM fbem-26173 215 6 32 32 NUM fbem-26173 215 7 . . PUNCT fbem-26173 216 1 [ [ X fbem-26173 216 2 9 9 NUM fbem-26173 216 3 ] ] X fbem-26173 216 4 iranzad iranzad PROPN fbem-26173 216 5 r r PROPN fbem-26173 216 6 , , PUNCT fbem-26173 216 7 liu liu PROPN fbem-26173 216 8 x. x. PROPN fbem-26173 216 9 a a DET fbem-26173 216 10 review review NOUN fbem-26173 216 11 of of ADP fbem-26173 216 12 random random ADJ fbem-26173 216 13 forest forest NOUN fbem-26173 216 14 - - PUNCT fbem-26173 216 15 based base VERB fbem-26173 216 16 feature feature NOUN fbem-26173 216 17 selection selection NOUN fbem-26173 216 18 methods method NOUN fbem-26173 216 19 for for ADP fbem-26173 216 20 data datum NOUN fbem-26173 216 21 science science NOUN fbem-26173 216 22 education education NOUN fbem-26173 216 23 and and CCONJ fbem-26173 216 24 applications application NOUN fbem-26173 216 25 [ [ X fbem-26173 216 26 j j X fbem-26173 216 27 ] ] X fbem-26173 216 28 . . PUNCT fbem-26173 217 1 international international ADJ fbem-26173 217 2 journal journal PROPN fbem-26173 217 3 of of ADP fbem-26173 217 4 data data PROPN fbem-26173 217 5 science science NOUN fbem-26173 217 6 and and CCONJ fbem-26173 217 7 analytics analytic NOUN fbem-26173 217 8 , , PUNCT fbem-26173 217 9 2024 2024 NUM fbem-26173 217 10 : : PUNCT fbem-26173 217 11 1 1 NUM fbem-26173 217 12 - - SYM fbem-26173 217 13 15 15 NUM fbem-26173 217 14 . . PUNCT fbem-26173 218 1 [ [ X fbem-26173 218 2 10 10 NUM fbem-26173 218 3 ] ] X fbem-26173 218 4 chen chen PROPN fbem-26173 218 5 t t PROPN fbem-26173 218 6 , , PUNCT fbem-26173 218 7 guestrin guestrin PROPN fbem-26173 218 8 c. c. PROPN fbem-26173 218 9 xgboost xgboost PROPN fbem-26173 218 10 : : PUNCT fbem-26173 218 11 a a DET fbem-26173 218 12 scalable scalable ADJ fbem-26173 218 13 tree tree NOUN fbem-26173 218 14 boosting boost VERB fbem-26173 218 15 svstem svstem NOUN fbem-26173 219 1 [ [ X fbem-26173 219 2 c]//in c]//in PROPN fbem-26173 219 3 proceedings proceeding NOUN fbem-26173 219 4 of of ADP fbem-26173 219 5 the the DET fbem-26173 219 6 22nd 22nd PROPN fbem-26173 219 7 acm acm PROPN fbem-26173 219 8 sigkdd sigkdd PROPN fbem-26173 219 9 international international ADJ fbem-26173 219 10 conference conference NOUN fbem-26173 219 11 on on ADP fbem-26173 219 12 knowledge knowledge NOUN fbem-26173 219 13 discovery discovery PROPN fbem-26173 219 14 and and CCONJ fbem-26173 219 15 data datum NOUN fbem-26173 219 16 mining mining NOUN fbem-26173 219 17 . . PUNCT fbem-26173 220 1 acm acm PROPN fbem-26173 220 2 , , PUNCT fbem-26173 220 3 2016 2016 NUM fbem-26173 220 4 : : PUNCT fbem-26173 220 5 785 785 NUM fbem-26173 220 6 - - SYM fbem-26173 220 7 794 794 NUM fbem-26173 220 8 . . PUNCT fbem-26173 221 1 [ [ X fbem-26173 221 2 11 11 NUM fbem-26173 221 3 ] ] X fbem-26173 221 4 niazkar niazkar PRON fbem-26173 221 5 m m NOUN fbem-26173 221 6 , , PUNCT fbem-26173 221 7 menapace menapace NOUN fbem-26173 221 8 a a NOUN fbem-26173 221 9 , , PUNCT fbem-26173 221 10 brentan brentan PROPN fbem-26173 221 11 b b PROPN fbem-26173 221 12 , , PUNCT fbem-26173 221 13 et et PROPN fbem-26173 221 14 al al PROPN fbem-26173 221 15 . . PUNCT fbem-26173 221 16 applications application NOUN fbem-26173 221 17 of of ADP fbem-26173 221 18 xgboost xgboost ADV fbem-26173 221 19 in in ADP fbem-26173 221 20 water water NOUN fbem-26173 221 21 resources resource NOUN fbem-26173 221 22 engineering engineering NOUN fbem-26173 221 23 : : PUNCT fbem-26173 221 24 a a DET fbem-26173 221 25 systematic systematic ADJ fbem-26173 221 26 literature literature NOUN fbem-26173 221 27 review review NOUN fbem-26173 221 28 ( ( PUNCT fbem-26173 221 29 dec dec PROPN fbem-26173 221 30 2018 2018 NUM fbem-26173 221 31 – – PUNCT fbem-26173 221 32 may may NOUN fbem-26173 221 33 2023 2023 NUM fbem-26173 221 34 ) ) PUNCT fbem-26173 222 1 [ [ X fbem-26173 222 2 j j X fbem-26173 222 3 ] ] X fbem-26173 222 4 . . PUNCT fbem-26173 223 1 environmental environmental ADJ fbem-26173 223 2 modelling modelling NOUN fbem-26173 223 3 & & CCONJ fbem-26173 223 4 software software NOUN fbem-26173 223 5 , , PUNCT fbem-26173 223 6 2024 2024 NUM fbem-26173 223 7 : : PUNCT fbem-26173 223 8 105971 105971 NUM fbem-26173 223 9 . . PUNCT fbem-26173 224 1 [ [ X fbem-26173 224 2 12 12 NUM fbem-26173 224 3 ] ] X fbem-26173 224 4 joshi joshi PROPN fbem-26173 224 5 a a PROPN fbem-26173 224 6 , , PUNCT fbem-26173 224 7 vishnu vishnu ADJ fbem-26173 224 8 c c NOUN fbem-26173 224 9 , , PUNCT fbem-26173 224 10 mohan mohan PROPN fbem-26173 224 11 c c PROPN fbem-26173 224 12 k k PROPN fbem-26173 224 13 , , PUNCT fbem-26173 224 14 et et PROPN fbem-26173 224 15 al al PROPN fbem-26173 224 16 . . PUNCT fbem-26173 224 17 application application NOUN fbem-26173 224 18 of of ADP fbem-26173 224 19 xgboost xgboost PROPN fbem-26173 224 20 model model NOUN fbem-26173 224 21 for for ADP fbem-26173 224 22 early early ADJ fbem-26173 224 23 prediction prediction NOUN fbem-26173 224 24 of of ADP fbem-26173 224 25 earthquake earthquake NOUN fbem-26173 224 26 magnitude magnitude NOUN fbem-26173 224 27 from from ADP fbem-26173 224 28 waveform waveform NOUN fbem-26173 224 29 data datum NOUN fbem-26173 225 1 [ [ X fbem-26173 225 2 j j X fbem-26173 225 3 ] ] X fbem-26173 225 4 . . PUNCT fbem-26173 226 1 journal journal PROPN fbem-26173 226 2 of of ADP fbem-26173 226 3 earth earth NOUN fbem-26173 226 4 system system NOUN fbem-26173 226 5 science science NOUN fbem-26173 226 6 , , PUNCT fbem-26173 226 7 2023 2023 NUM fbem-26173 226 8 , , PUNCT fbem-26173 226 9 133(1 133(1 NUM fbem-26173 226 10 ): ): PUNCT fbem-26173 226 11 5 5 NUM fbem-26173 226 12 . . PUNCT fbem-26173 227 1 [ [ X fbem-26173 227 2 13 13 NUM fbem-26173 227 3 ] ] X fbem-26173 227 4 kang kang PROPN fbem-26173 227 5 m. m. PROPN fbem-26173 227 6 stock stock PROPN fbem-26173 227 7 price price NOUN fbem-26173 227 8 prediction prediction NOUN fbem-26173 227 9 with with ADP fbem-26173 227 10 heavy heavy ADJ fbem-26173 227 11 - - PUNCT fbem-26173 227 12 tailed tail VERB fbem-26173 227 13 distribution distribution NOUN fbem-26173 227 14 time time NOUN fbem-26173 227 15 - - PUNCT fbem-26173 227 16 series series NOUN fbem-26173 227 17 generation generation NOUN fbem-26173 227 18 based base VERB fbem-26173 227 19 on on ADP fbem-26173 227 20 wganbilstm wganbilstm NOUN fbem-26173 227 21 [ [ X fbem-26173 227 22 j j X fbem-26173 227 23 ] ] X fbem-26173 227 24 . . PUNCT fbem-26173 228 1 computational computational ADJ fbem-26173 228 2 economics economic NOUN fbem-26173 228 3 , , PUNCT fbem-26173 228 4 2024 2024 NUM fbem-26173 228 5 : : PUNCT fbem-26173 228 6 1 1 NUM fbem-26173 228 7 - - SYM fbem-26173 228 8 20 20 NUM fbem-26173 228 9 . . PUNCT fbem-26173 229 1 [ [ X fbem-26173 229 2 14 14 NUM fbem-26173 229 3 ] ] X fbem-26173 229 4 bukhari bukhari NOUN fbem-26173 229 5 a a DET fbem-26173 229 6 h h NOUN fbem-26173 229 7 , , PUNCT fbem-26173 229 8 raja raja PROPN fbem-26173 229 9 m m PROPN fbem-26173 229 10 a a DET fbem-26173 229 11 z z PROPN fbem-26173 229 12 , , PUNCT fbem-26173 229 13 sulaiman sulaiman PROPN fbem-26173 229 14 m m PROPN fbem-26173 229 15 , , PUNCT fbem-26173 229 16 et et PROPN fbem-26173 229 17 al al PROPN fbem-26173 229 18 . . PROPN fbem-26173 230 1 fractional fractional PROPN fbem-26173 230 2 neurose⁃quential neurose⁃quential ADJ fbem-26173 230 3 arfima arfima NOUN fbem-26173 230 4 - - PUNCT fbem-26173 230 5 lstm lstm NOUN fbem-26173 230 6 for for ADP fbem-26173 230 7 financial financial ADJ fbem-26173 230 8 market market NOUN fbem-26173 230 9 forecasting forecasting NOUN fbem-26173 231 1 [ [ X fbem-26173 231 2 j j X fbem-26173 231 3 ] ] X fbem-26173 231 4 . . PUNCT fbem-26173 232 1 ieee ieee PROPN fbem-26173 232 2 access,2020,(8 access,2020,(8 NOUN fbem-26173 232 3 ) ) PUNCT fbem-26173 232 4 . . PUNCT fbem-26173 233 1 [ [ X fbem-26173 233 2 15 15 NUM fbem-26173 233 3 ] ] X fbem-26173 233 4 khetarpal khetarpal ADJ fbem-26173 233 5 p p X fbem-26173 233 6 , , PUNCT fbem-26173 233 7 nagpal nagpal ADJ fbem-26173 233 8 n n CCONJ fbem-26173 233 9 , , PUNCT fbem-26173 233 10 siano siano PROPN fbem-26173 233 11 p p NOUN fbem-26173 233 12 , , PUNCT fbem-26173 233 13 et et PROPN fbem-26173 233 14 al al PROPN fbem-26173 233 15 . . PUNCT fbem-26173 234 1 power power NOUN fbem-26173 234 2 quality quality NOUN fbem-26173 234 3 disturbance disturbance NOUN fbem-26173 234 4 signal signal NOUN fbem-26173 234 5 segmentation segmentation NOUN fbem-26173 234 6 and and CCONJ fbem-26173 234 7 classification classification NOUN fbem-26173 234 8 based base VERB fbem-26173 234 9 on on ADP fbem-26173 234 10 modified modified ADJ fbem-26173 234 11 bi‐lstm bi‐lstm NOUN fbem-26173 234 12 with with ADP fbem-26173 234 13 double double ADJ fbem-26173 234 14 attention attention NOUN fbem-26173 234 15 mechanism mechanism NOUN fbem-26173 235 1 [ [ X fbem-26173 235 2 j j X fbem-26173 235 3 ] ] X fbem-26173 235 4 . . PUNCT fbem-26173 236 1 iet iet PROPN fbem-26173 236 2 generation generation NOUN fbem-26173 236 3 , , PUNCT fbem-26173 236 4 transmission transmission NOUN fbem-26173 236 5 & & CCONJ fbem-26173 236 6 distribution distribution NOUN fbem-26173 236 7 , , PUNCT fbem-26173 236 8 2024 2024 NUM fbem-26173 236 9 , , PUNCT fbem-26173 236 10 18(1 18(1 NUM fbem-26173 236 11 ): ): PUNCT fbem-26173 236 12 50 50 NUM fbem-26173 236 13 - - SYM fbem-26173 236 14 62 62 NUM fbem-26173 236 15 . . PUNCT fbem-26173 237 1 [ [ X fbem-26173 237 2 16 16 NUM fbem-26173 237 3 ] ] PUNCT fbem-26173 237 4 ye ye PROPN fbem-26173 237 5 z z PROPN fbem-26173 237 6 , , PUNCT fbem-26173 237 7 zuo zuo PROPN fbem-26173 237 8 t t PROPN fbem-26173 237 9 , , PUNCT fbem-26173 237 10 chen chen PROPN fbem-26173 237 11 w w PROPN fbem-26173 237 12 , , PUNCT fbem-26173 237 13 et et PROPN fbem-26173 237 14 al al PROPN fbem-26173 237 15 . . PROPN fbem-26173 238 1 textual textual ADJ fbem-26173 238 2 emotion emotion NOUN fbem-26173 238 3 recognition recognition NOUN fbem-26173 238 4 method method NOUN fbem-26173 238 5 based base VERB fbem-26173 238 6 on on ADP fbem-26173 238 7 albert albert NOUN fbem-26173 238 8 - - PUNCT fbem-26173 238 9 bilstm bilstm NOUN fbem-26173 238 10 model model NOUN fbem-26173 238 11 and and CCONJ fbem-26173 238 12 svm svm ADJ fbem-26173 238 13 - - ADJ fbem-26173 238 14 nb nb ADJ fbem-26173 238 15 classification classification NOUN fbem-26173 239 1 [ [ X fbem-26173 239 2 j j X fbem-26173 239 3 ] ] X fbem-26173 239 4 . . PUNCT fbem-26173 240 1 soft soft ADJ fbem-26173 240 2 computing computing NOUN fbem-26173 240 3 , , PUNCT fbem-26173 240 4 2023 2023 NUM fbem-26173 240 5 , , PUNCT fbem-26173 240 6 27(8 27(8 NUM fbem-26173 240 7 ): ): PUNCT fbem-26173 240 8 5063 5063 NUM fbem-26173 240 9 - - SYM fbem-26173 240 10 5075 5075 NUM fbem-26173 240 11 . . PUNCT fbem-26173 241 1 [ [ X fbem-26173 241 2 17 17 NUM fbem-26173 241 3 ] ] PUNCT fbem-26173 241 4 thumilvannan thumilvannan NOUN fbem-26173 241 5 s s PROPN fbem-26173 241 6 , , PUNCT fbem-26173 241 7 balamanigandan balamanigandan PROPN fbem-26173 241 8 r. r. PROPN fbem-26173 241 9 a a DET fbem-26173 241 10 novel novel ADJ fbem-26173 241 11 adaptive adaptive ADJ fbem-26173 241 12 weight weight NOUN fbem-26173 241 13 bi bi ADJ fbem-26173 241 14 - - ADJ fbem-26173 241 15 directional directional ADJ fbem-26173 241 16 long long ADJ fbem-26173 241 17 short short ADJ fbem-26173 241 18 - - PUNCT fbem-26173 241 19 term term NOUN fbem-26173 241 20 memory memory NOUN fbem-26173 241 21 ( ( PUNCT fbem-26173 241 22 awbi awbi NOUN fbem-26173 241 23 - - PUNCT fbem-26173 241 24 lstm lstm PROPN fbem-26173 241 25 ) ) PUNCT fbem-26173 241 26 classifier classifier NOUN fbem-26173 241 27 model model NOUN fbem-26173 241 28 for for ADP fbem-26173 241 29 heart heart NOUN fbem-26173 241 30 stroke stroke NOUN fbem-26173 241 31 risk risk NOUN fbem-26173 241 32 level level NOUN fbem-26173 241 33 prediction prediction NOUN fbem-26173 241 34 in in ADP fbem-26173 241 35 iot iot PROPN fbem-26173 242 1 [ [ X fbem-26173 242 2 j j X fbem-26173 242 3 ] ] X fbem-26173 242 4 . . PUNCT fbem-26173 243 1 peerj peerj PROPN fbem-26173 243 2 computer computer NOUN fbem-26173 243 3 science science NOUN fbem-26173 243 4 , , PUNCT fbem-26173 243 5 2024 2024 NUM fbem-26173 243 6 , , PUNCT fbem-26173 243 7 10 10 NUM fbem-26173 243 8 : : SYM fbem-26173 243 9 e2196 e2196 NUM fbem-26173 243 10 . . PUNCT fbem-26173 244 1 [ [ X fbem-26173 244 2 18 18 NUM fbem-26173 244 3 ] ] X fbem-26173 244 4 li li PROPN fbem-26173 244 5 h h PROPN fbem-26173 244 6 , , PUNCT fbem-26173 244 7 wu wu PROPN fbem-26173 244 8 x x PROPN fbem-26173 244 9 j. j. PROPN fbem-26173 244 10 crossfuse crossfuse PROPN fbem-26173 244 11 : : PUNCT fbem-26173 244 12 a a DET fbem-26173 244 13 novel novel ADJ fbem-26173 244 14 cross cross NOUN fbem-26173 244 15 attention attention NOUN fbem-26173 244 16 mechanism mechanism NOUN fbem-26173 244 17 based base VERB fbem-26173 244 18 infrared infrared ADJ fbem-26173 244 19 and and CCONJ fbem-26173 244 20 visible visible ADJ fbem-26173 244 21 image image NOUN fbem-26173 244 22 fusion fusion NOUN fbem-26173 244 23 approach approach NOUN fbem-26173 244 24 [ [ X fbem-26173 244 25 j j X fbem-26173 244 26 ] ] X fbem-26173 244 27 . . PUNCT fbem-26173 245 1 information information NOUN fbem-26173 245 2 fusion fusion NOUN fbem-26173 245 3 , , PUNCT fbem-26173 245 4 2024 2024 NUM fbem-26173 245 5 , , PUNCT fbem-26173 245 6 103 103 NUM fbem-26173 245 7 : : SYM fbem-26173 245 8 102147 102147 NUM fbem-26173 245 9 . . PUNCT fbem-26173 246 1 [ [ X fbem-26173 246 2 19 19 NUM fbem-26173 246 3 ] ] PUNCT fbem-26173 246 4 agudelo agudelo NOUN fbem-26173 246 5 aguirre aguirre PROPN fbem-26173 246 6 a a DET fbem-26173 246 7 a a PROPN fbem-26173 246 8 , , PUNCT fbem-26173 246 9 duque duque PROPN fbem-26173 246 10 mΓ©ndez mΓ©ndez PROPN fbem-26173 246 11 n n PROPN fbem-26173 246 12 d d PROPN fbem-26173 246 13 , , PUNCT fbem-26173 246 14 rojas rojas PROPN fbem-26173 246 15 medina medina PROPN fbem-26173 246 16 r r PROPN fbem-26173 246 17 a. a. NOUN fbem-26173 246 18 artificial artificial ADJ fbem-26173 246 19 intelligence intelligence NOUN fbem-26173 246 20 applied apply VERB fbem-26173 246 21 to to ADP fbem-26173 246 22 investment investment NOUN fbem-26173 246 23 in in ADP fbem-26173 246 24 variable variable ADJ fbem-26173 246 25 income income NOUN fbem-26173 246 26 through through ADP fbem-26173 246 27 the the DET fbem-26173 246 28 macd macd NOUN fbem-26173 246 29 ( ( PUNCT fbem-26173 246 30 moving move VERB fbem-26173 246 31 average average ADJ fbem-26173 246 32 convergence convergence NOUN fbem-26173 246 33 / / SYM fbem-26173 246 34 divergence divergence NOUN fbem-26173 246 35 ) ) PUNCT fbem-26173 246 36 indicator indicator NOUN fbem-26173 247 1 [ [ X fbem-26173 247 2 j j X fbem-26173 247 3 ] ] X fbem-26173 247 4 . . PUNCT fbem-26173 248 1 journal journal PROPN fbem-26173 248 2 of of ADP fbem-26173 248 3 economics economic NOUN fbem-26173 248 4 , , PUNCT fbem-26173 248 5 finance finance NOUN fbem-26173 248 6 and and CCONJ fbem-26173 248 7 administrative administrative ADJ fbem-26173 248 8 science science NOUN fbem-26173 248 9 , , PUNCT fbem-26173 248 10 2021 2021 NUM fbem-26173 248 11 , , PUNCT fbem-26173 248 12 26(52 26(52 NUM fbem-26173 248 13 ): ): PUNCT fbem-26173 248 14 268 268 NUM fbem-26173 248 15 - - SYM fbem-26173 248 16 281 281 NUM fbem-26173 248 17 . . PUNCT fbem-26173 249 1 [ [ X fbem-26173 249 2 20 20 NUM fbem-26173 249 3 ] ] X fbem-26173 249 4 yang yang PROPN fbem-26173 249 5 x x PROPN fbem-26173 249 6 , , PUNCT fbem-26173 249 7 liu liu PROPN fbem-26173 249 8 w w PROPN fbem-26173 249 9 , , PUNCT fbem-26173 249 10 zhou zhou PROPN fbem-26173 249 11 d d PROPN fbem-26173 249 12 , , PUNCT fbem-26173 249 13 et et PROPN fbem-26173 249 14 al al PROPN fbem-26173 249 15 . . PUNCT fbem-26173 250 1 qlib qlib PROPN fbem-26173 250 2 : : PUNCT fbem-26173 250 3 an an DET fbem-26173 250 4 ai ai ADV fbem-26173 250 5 - - PUNCT fbem-26173 250 6 oriented orient VERB fbem-26173 250 7 quantitative quantitative ADJ fbem-26173 250 8 investment investment NOUN fbem-26173 250 9 platform platform NOUN fbem-26173 251 1 [ [ X fbem-26173 251 2 j j X fbem-26173 251 3 ] ] X fbem-26173 251 4 . . PUNCT fbem-26173 252 1 arxiv arxiv PROPN fbem-26173 252 2 preprint preprint PROPN fbem-26173 252 3 arxiv:2009.11189 arxiv:2009.11189 PROPN fbem-26173 252 4 , , PUNCT fbem-26173 252 5 2020 2020 NUM fbem-26173 252 6 . . PUNCT