Copyright © 2016 The Authors. Published by VGTU Press. This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 (CC BY-NC 4.0) license, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. The material cannot be used for commercial purposes. FOrMATION OF AN INTEGrATED STOCK PrICE FOrECAST MODEl IN lITHUANIA Audrius DZIKEVIČIUS1, Svetlana ŠArANDA2 1,2Department of Finance Engineering, Faculty of Business Management, Vilnius Gediminas Technical University, Saulėtekio al. 11, 10223 Vilnius, Lithuania E-mails: 1audrius.dzikevicius@vgtu.lt (corresponding author); 2svetlana.saranda@gmail.com Received 16 November 2016; accepted 22 December 2016 Abstract. Technical and fundamental analyses are widely used to forecast stock prices due to lack of knowledge of other modern models and methods such as Residual Income Model, ANN-APGARCH, Support Vector Machine, Probabilistic Neural Network and Genetic Fuzzy Systems. Although stock price forecast models integrating both technical and fundamental analyses are currently used widely, their integration is not justified comprehensively enough. This paper discusses theoretical one-factor and multi-factor stock price forecast models already applied by investors at a global level and determines possibility to create and apply prac- tically a stock price forecast model which integrates fundamental and technical analysis with the reference to the Lithuanian stock market. The research is aimed to determine the relationship between stock prices of the 14 Lithuanian compa- nies listed in the Main List by the Nasdaq OMX Baltic and various fundamental variables. Based on correlation and regression analysis results and application of c-Squared Test, ANOVA method, a general stock price forecast model is gener- ated. This paper discusses practical implications how the developed model can be used to forecast stock prices by individual investors and suggests additional check measures. Keywords: Stock price, forecast, correlation analysis, regression analysis, funda- mental analysis, technical analysis, model, investment. JEL Classification: G1, G12, G2, G30. 1. Introduction Nowadays individual and institutional investors are surrounded by a rapidly changing external environment (political, social, economic and technological) which results in volatility in the financial markets. Inability to manage investments properly is consid- ered to be one of the causes of the global 2008–2009 financial crisis. Investments in securities, especially to companies’ stocks, are the most common type of investments. However, investors, especially if they are individual persons, often lack B u s i n e s s, Ma n ag e M e n t a n d e d u c at i o n ISSN 2029-7491 / eISSN 2029-6169 2016, 14(2): 292–307 doi:10.3846/bme.2016.337 mailto:1audrius.dzikevicius@vgtu.lt doi:10.3846/bme.2016.337 293 Business, Management and Education, 2016, 14(2): 292–307 knowledge needed to make investment decisions, which can lead to financial losses. If these investors could make more accurate stock price forecasts, financial resources could be allocated more effectively. If this were the case, the development of stock markets could result in increasing confidence among individual investors. In order to forecast stock prices investors usually use either fundamental or technical analysis. Some investors may try to apply models which include both fundamental and technical analysis with a basis on neural networks. However, this requires a sophisti- cated understanding of neural networking. The problem is that since global financial markets and economic have recovered after the financial crisis of 2008–2009, the academia started to pay less attention to re- search that focuses on stock markets and their forecasts. Several studies of high quality have been done to forecast stock prices, probably since it seemed attractive due to high benefits (Zavadskas, Turskis 2011; Li et al. 2014). Research aimed to help investors in the application of both fundamental and techni- cal analysis without using neural networks is at early development phase. Due to this reason a stock price forecast model which integrates both analyses is being developed and tested in the Lithuanian stock market with a possibility to apply it at global level further. It is assumed that stock prices in various foreign markets can be also predicted using the same stock price forecast model. The purpose of this paper is to clarify if it is possible to form and practically apply a stock price forecast model with reference to previously conducted research based on technical analysis and possibilities to apply financial ratio analysis when forecasting stock prices. Moving Average (MA), one of the widely used technical analysis tools, is not useful when forecasting stock price movement trends. For specific time series its lengths dif- fer by forecast accuracy level. Thus this method can generate significant forecast value errors and deviations from real stock prices. Meanwhile, Exponential Moving Average (EMA) with an appropriate α Constanta level is more relevant to forecast stock prices. The highest α level decreases the prob- ability of a higher bias level for investors. However, longer period of the EMA means higher bias level and less accurate forecast of stock prices but it does not mean that this method is unsuitable to predict stock markets fluctuations. Previous research proves that a company analysis should be conducted, after that a sector analysis and then an economic analysis should be conducted. By so doing the links between the company and the external factors can be identified and used to evalu- ate their impact on the financial company’s performance and its stock prices. That’s why it is necessary to obtain various variables (macroeconomic factors, financial ratios, sec- tor indexes etc.) which potentially have an influence on companies’ stock prices. This can help in the investment decisions making. 294 A. Dzikevičius, S. Šaranda. Formation of an integrated stock price forecast model in Lithuania 2. review of theoretical stock price forecast methods and models While most investors believe that markets in general, and stock markets in particular, will deliver increased value in the future, during times of economic and financial instability they face significant losses. Most investors relate their expectations to increasing asset value, supported by future profit, productivity, population growth and similar factors. The global financial and economic crisis of 2008–2009 demonstrated that investors’ false belief in unstoppable asset value growth is the main reason for the financial market downturn. Today econometric and economic science make possible the application of various methods and models aimed at predicting stock market processes but not all of already available theoretical forecast methods and models can be applied effectively. For instances, in Lithuania, the most common methods used are quite simple, name- ly, fundamental and technical analysis. Fundamental analysis is based on the in-depth analysis of companies’ financial per- formance, surrounding industrial environment, macroeconomic factors, as well as indus- trial news. All this supports decision making when forecasting stock prices (Chen 2013). Its application also requires the assessment of companies’ future growth prospects and competitive landscape (Bonga 2015) as well as potential changes in GDP, sales strate- gies and other different indicators relevant not only for individual businesses but for the industries as a whole in which they operate (Kartašova, Venclauskienė 2014). In order to forecast stock prices the Residual Income Model (RIM) with its strong theoretical background can be applied with reference to the financial data sourced from account- ing (Sarikhani, Ebrahimi 2012; Tareq 2012; Kariuki, Oyugi 2013). As Higgins (2011) states, its successful application contributes to a fundamental perspective when making decisions regarding forecast of stock prices. Proponents of technical analysis criticize the application of fundamental analysis by claiming that the stock market historical data creates problematic assumptions for the development of rules used to predict future changes in stock prices (Kimoto et al. 1990; Olson, Mossman 2003). Indeed, investors sometimes ignore fundamental analysis and rely solely on technical analysis results. In their opinion, application of fundamental analysis is useful for a long-term investment while in the short-term it may generate losses. For instance, Autoregressive and Moving Average (ARMA) model is widely used by the proponents of technical analysis. Autoregressive Integrated Moving Average (ARIMA) as an improved ARMA model can be also used effectively when forecasting stock prices. The main difference between the models is that ARIMA model converts a non-stationary data to a stationary data before its application. ARIMA model is widely used to make linear data series forecasts (Box, Tiao 1975). Mondal et al. (2014) claim that the accuracy of stock price forecasts achieved with ARIMA model is quite high i.e. above 85%, making the model suitable for practical use with a relatively small bias. In addition to these methods and models aimed to support companies’ stock price forecasts, other various methods and models exist. For instances, artificial neural networks 295 Business, Management and Education, 2016, 14(2): 292–307 (ANN) method is widely applied (Öğüt et al. 2009; Tseng et al. 2012; Guersen et al. 2011; Mostafa 2010; Cheng et al. 2012; Cimpoeru 2011). Bildirici and Ersin (2009) also sug- gest to use an ANN-APGARCH model as an improved APGARCH model to make stock prices forecasts more accurate by reflecting the real situation in the market. Data mining and Neuro Fuzzy system methods are based on a possibility of extract- ing useful information from various data sets and play an important part when fore- casting stock markets and making investment decisions. These approaches are applied in order to monitor the entire the stock market price behavior and market fluctuations. Meanwhile, Markov Model is developed exclusively for financial markets and stock price forecasting (Preethi, Santhi 2012). Sometimes integrated or hybrid models such as the hybrid model of artificial intel- ligent methods proposed by Wu et al. (2015) can be also be applied. It is a combina- tion of ARMA, the support vector machine method (SVM) and a probabilistic neural network (PNN). Another offered model includes application of genetic fuzzy systems (GFS) and ANN (Hadavandi 2010). Despite the fact that a lot of theoretical and practically tested stock price forecast methods and models exist, the majority of Lithuanian investors believe that stock prices can be described by one factor. For example, it is assumed that yields depend on the GDP and its dynamics (Duca 2007). In 2011, Hsing proposed a stock price forecast model suitable for a global and most developed stock markets as well as less developed markets such as Lithuanian one. The model is based on an expansion of the previously proposed models (Bulmash, Trivoli 1991; Abdullah, Hayworth 1993; Kim 2003; Ratanapakorn, Sharma 2007; Humpe, Mac- millan 2009; Pilinkus 2009, 2010). Unfortunately, only few studies suggest a stock price forecast model that includes not only the fundamental factors but also the technical elements of the analysis. One such model has been proposed by Bettman et al. (2009). Ferreira and Santa-Clara (2008) reviewed the relevant academic literature and concluded that price multiples, macro- economic factors, financial ratios and different measures of risk are the most common variables taken into account when forecasting stock returns. For example, Campbell and Shiller (1988), Lamont (1998) used the P/E ratio to forecast stock returns. This ratio enables an analysis of investors’ willingness to pay for the company’s earnings. It makes sense to purchase stocks with low P/E ratio value only in case if the earnings are real (Kartašova, Venclauskienė 2014). Meanwhile Kothari and Shanken (1997), Pontiff and Schall (1998) use the P/BV ratio. Another financial ratio to forecast companies’ stock prices suggested by the aca- demia is return on equity (ROE) (Petcharabul, Romprasert 2014). It measures company’s performance efficiency (Higgins 2009). Elleuch (2009) also noticed that the ratio is a predictive measure to forecast future stock prices. Omran and Ragab (2004) discovered that ROE plays a significant role in investment decision making. 296 A. Dzikevičius, S. Šaranda. Formation of an integrated stock price forecast model in Lithuania Return on assets (ROA) ratio also indicates the level of company’s performance (Purnamasari 2015). Ball and Brown (1968), Hou et al. (2014) indicate that in addition to the mentioned financial ratios, earning per share ratio (EPS) may serve as a variable when forecasting stock prices. Analysts estimate that the stock price increases when reported ratio is higher than expected and vice versa – if it is lower, then the stock price decreases (Renfro 2015). A company’s net profit margin (NPM) is also an important driver of stock perfor- mance and price trends. Financial analysts and investors refer to earnings, as well as NPM, as stock price is often related to a company’s expected future earnings (Amir et al. 2012). Due to this reason the ratio may serve as a useful variable when forecasting stock prices, for example in the Lithuanian market. With the reference to previously conducted research, Dutta et al. (2012) concluded that financial ratios can used by investors to assess future performance of companies’ stock prices and their trends. Therefore, financial ratios can be also used by to classify the performance of differ- ent companies and forecast their stock prices in the market. Both linear and non-linear types of relationship between financial ratios and stock prices can be used as a back- ground. This finding is also supported by Jabbari and Fathi (2014). 3. research methodology For the research purposes the annual data of 14 Lithuanian companies in accordance to the Baltic Main List of stocks for the period of 2010–2014 was used (Table 1). Table 1. List of researched companies and the industries they operate in Symbol Company’s name Industry APG1L Apranga 5000 Consumer Services CTS1L City Service 2000 Industrials GRG1L Grigiškės 1000 Basic Materials IVL1L Invalda INVL 8000 Financials LDJ1L Lietuvos dujos 7000 Utilities LNA1L Linas Agro Group 3000 Consumer Goods PTR1L Panevėžio statybos trestas 2000 Industrials PZV1L Pieno žvaigždės 3000 Consumer Goods RSU1L Rokiškio sūris 3000 Consumer Goods SAB1L Šiaulių bankas 8000 Financials TEO1L TEO LT 6000 Telecommunications UTR1L Utenos trikotažas 3000 Consumer Goods VLP1L Vilkyškių pieninė 3000 Consumer Goods VBL1L Vilniaus baldai 3000 Consumer Goods 297 Business, Management and Education, 2016, 14(2): 292–307 Only last (close) annual stock prices of the listed companies were analysed. Statisti- cal data reflecting indexes provided in the Table 1 were also used. In order to reflect the real stock market performance dividends are reinvested in the gross index. The relevant financial data for these companies is sourced from the officially pub- lished financial statements. Based on the in-depth literature analysis, seven financial ratios were selected for the further analysis (Table 2). Table 2. Calculation methodology and description of the financial ratios used Financial ratio Calculation methodology Description EPS (Earnings Per Share) Net profit – Preference share dividend Weighted average number of ordinary outstanding EPS ratio shows how much of the company’s earned net profit is attributed to an ordinary share. P/E (Price / Earnings Ratio) Current share price Earnings per share P/E ratio reflects how much an investor pays for 1 euro of the company’s 1 net profit euro. It also relates the stock’s market value to the profit reflected in financial statements. P/S (Price to Sales Ratio) Current share price Sales per share P/S ratio shows how much an investor pays for 1 euro of sales revenue generated by one share. P/BV (Price to Book Value Ratio) Current share price Book value per share P/BV reflects how much an investor pays for 1 euro of equity. ROE (Return on Equity) Net profit margin x Total assets Total assets turnover x Equity ROE shows how much net profit is generated by 1 euro of equity. ROA (Return on Assets) Net profit margin x Total assets turnover ROA shows how much net profit is generated by 1 euro of assets. NPM (Net Profit Margin) Net profit Sales NPM indicates the percentage of net profit with 1 euro of sales revenue i.e. it shows the company’s operating efficiency. 298 A. Dzikevičius, S. Šaranda. Formation of an integrated stock price forecast model in Lithuania The financial ratios are calculated in accordance with the methodology accepted by NASDAQ OMX. Data indicating annual time series of macroeconomic factors is also used in this research. Based on the analysis of the scientific literature it was concluded that for this research purpose only the most important macroeconomic factors are analyzed. They are outlined as follows: Gross Domestic Product (GDP), value in EUR; Foreign Di- rect Investments (FDI), value in EUR; Consumer Price Indexes (CPI); Unemployment Rate (UR), %; Interest Rate of loans issued to households and non-financial institutions (IR), %; Import (I), value in EUR; Export (E), value in EUR. Relevant data was sourced from the Official Statistics Portal and the Bank of Lithu- ania. During the first phase of the research a data set combining companies’ stock prices, financial ratios, macroeconomic variables, and sector indexes is built. The set is used to perform correlation analysis. Only variables which correlation with companies’ stock prices R ≥ |0.95| are selected. Further a regression analysis is conducted to equate a linear relationship between companies’ stock prices and selected variables as well as analysis of variance (ANOVA). For decision making purposes, additional measure of forecast accuracy evaluation is introduced. c-Squared Test is applied to check whether forecasted stock prices reflect real market prices. To check whether forecasted companies’ stock prices differ from the real ones, the analysis of the distribution of index returns using c-Squared Test method is used (1): ( )22 1 k i i ii O E E= − χ = ∑ , (1) where: Oi – historical companies’ stock prices; Ei – forecasted companies’ stock prices; k – range of variables; Companies’ stock price forecast model is acceptable when c- Squared Test p value >0.05. If this condition is satisfied than a created model based on linear regression maybe used practically. 4. Discussion of the research results During the first stage of the research variables (i.e. stock prices of other companies, fi- nancial ratios, macroeconomic factors, sector indexes) which correlate with companies’ stock prices R ≥ |0.95| are selected (Table 3). During the research it was established that most of the companies’ stock prices cor- relate with one or several variables: one variable (PZV1L, RSU1L, UTR1L, VLB1L); three variables (APG1L, LNA1L, TEO1L, LDJ1L); four variables (INV1L, SAB1L); five and more variables (PTR1L, CTS1L). 299 Business, Management and Education, 2016, 14(2): 292–307 Table 3. Correlation analysis results Company Variable R ≥ |0.95| Company Variable R ≥ |0.95| PZV1L SAB1L P/E 0.9833 TEO1L LNA1L ROA LNA1L ROE TEO1L P/BV –0.9583 –0.9531 0.9940 RSU1L INV1L P/BV 0.9843 LDJ1L SAB1L UTR1L P/E VLP1L P/BV 0.9753 0.9985 0.9576 UTR1L RSU1L EPS –0.9515 INV1L UTR1L P/E VBL1L EPS VBL1L P/E VBL1L ROA –0.9977 –0.9949 0.9662 –0.9664 VBL1L APG1L EPS 0.9531 SAB1L APG1L P/E RSU1L P/E UTR1L P/E VLP1L P/BV 0.9661 0.9573 0.9555 0.9700 APG1L LNA1L PZV1L P/E PZV1L P/BV 0.9801 –0.9936 –0.9826 PTR1L APG1L P/E LDJ1L ROA LDJ1L ROE PTR1L P/BV RSU1L P/E RSU1L P/BV VLP1L GP B2000GI B8000GI 0.9592 0.9865 0.9925 0.9978 0.9811 0.9842 0.9521 0.9869 0.9680 LNA1L LNA1L GP PZV1L P/E PZV1L P/BV –0.9528 –0.9579 –0.9658 CTS1L PTR1L LDJ1L ROA LDJ1L ROE PTR1L P/BV PTR1L GP RSU1L P/E RSU1L P/BV SAB1L P/E B2000GI B8000GI E I 0.9758 0.9782 0.9604 0.9839 0.9641 0.9817 0.9863 –0.9893 0.9718 0.9682 –0.9538 –0.9764 VLP1L is the only company stock prices of which do not correlate with any vari- ables. Due to this reason it is singled out for further analysis. PTR1L and CTS1L are the companies’ stock prices of which correlate with sector indexes, namely B2000GI Industrials and B8000GI Financials. Moreover, in the case of company CTS1L the stock prices correlate with macroeconomic variables such as import and export. The research reveals that in the Lithuanian market environment the companies’ stock prices correlate with their own financial ratio very rarely. It is more likely that they cor- relate with the financial ratios of other companies (Fig. 1). 300 A. Dzikevičius, S. Šaranda. Formation of an integrated stock price forecast model in Lithuania Fig. 1. Share (%) of the analysed variables correlating with companies’ stock prices P/E, P/BV, ROA and ROE are the most important financial ratios of the companies other than analyzed. The results exhibit both negative and positive correlations. There- fore, further research is necessary to clarify why these correlations exist. This will enable a deeper understanding of investment logics, and help in the developing of an improved stock price forecast model, which is detailed later on in this paper. With reference to the findings presented in the Figure 1 regression analysis and ANOVA method are applied to equate the relationship between companies’ stock prices and selected variables mentioned in the Table 1. Essentially, companies’ stock price forecast models for PZV1L, RSU1L, UTR1L, VLB1L can be supported by a linear regression when one variable is applied (2): 1 1tP x= α + β + ε, (2) where: α and β1 – linear regression coefficients; ε – bias for the forecast value; t – time period; Stock price forecast models for these companies are presented in the Table 4. Table 4. Stock price forecast models with one variable Company Stock price forecast model for the period 2010–2014 General stock price forecast model PTR1L ( )1 2261 0 0398 1 SAB L P E+ + ε. . / ( )1 SAB L P Eα + β + ε/ RSU1L 0 9673 0 4660 1 IVL L P BV+ + ε. . ( / ) 1 IVL L P BVα + β + ε( / ) UTR1L 0 5939 1 4455 1 RSU L EPS− + ε. . ( ) 1 RSU L EPSα + β + ε( ) VBL1L 5 9058 42 5329 1 APG L EPS+ + ε. . ( ) 1 APG L EPSα + β + ε( ) 301 Business, Management and Education, 2016, 14(2): 292–307 Stock price forecast model supported by a linear regression with three variables is applied using equation (3): 1 1 2 2 3 3tP x x x= α + β + β + β + ε, (3) where: α and b1, b2, b3 – linear regression coefficients; ε – bias for the forecast value; t – time period. However, stock price forecast models only for three companies i.e. APG1L, LNA1L, TEO1L are presented in Table 5. Table 5. Stock price forecast models with three variables Company Stock price forecast model for the period 2010–2014 APG1L 2 0574 2 2581 1 0 0247 1 1 8083 1 LNA L PZV L P E PZV L P BV− + − + + ε. . ( ) . ( / ) . ( / ) LNA1L 1 2770 0 0884 1 0 0002 1 0 1807 1 LNA L GP PZV L P E PZV L P BV− − − + ε. . ( ) . ( / ) . ( / ) TEO1L 0 4882 0 4924 1 0 2076 1 0 2269 1 LNA L ROA LNA L ROE TEO L P BV− + + + ε. . ( ) . ( ) . ( / ) Company General stock price forecast model APG1L 1 2 31 1 1 LNA L PZV L P E PZV L P BVα + β − β + β + ε( ) ( / ) ( / ) LNA1L 1 2 31 1 1 LNA L GP PZV L P E PZV L P BVα + β + β + β + ε( ) ( / ) ( / ) TEO1L 1 2 31 1 1 LNA L ROA LNA L ROE TEO L P BVα + β + β + β + ε( ) ( ) ( / ) During correlation analysis stage it appeared that stock prices of the LDJ1L company correlate with three variables. However, after regression analysis and application of the ANOVA method it was concluded that only one variable is significant i.e. UTR1L P/E (Table 6). Table 6. Stock price forecast model for LDJ1L Company Stock price forecast model for the period 2010–2014 LDJ1L 0 5832 0 0029 1 UTR L P E+ + ε. . ( / ) Company General stock price forecast model LDJ1L 2 1 UTR L P Eα + β + ε( / ) Meanwhile variables SAB1L and VLP1L P/BV are excluded. Although stock prices of INV1L and SAB1L correlate with them, they should be excluded when equating the relationship between stock prices and variables for these companies. The results are presented in Table 7. 302 A. Dzikevičius, S. Šaranda. Formation of an integrated stock price forecast model in Lithuania Table 7. Stock price forecast model for INV1L and SAB1L Company Stock price forecast model for the period 2010–2014 INV1L 3 2887 0 0338 1 0 0480 1 UTR L P E VBL L P E− + + ε. . ( / ) . ( / ) SAB1L 0 2344 0 023 1 UTR L P E+ + ε. . ( / ) Company General stock price forecast model INV1L 1 31 1 UTR L P E VBL L P Eα + β + β + ε( / ) ( / ) SAB1L 3 1 VBL L P Eα + β + ε( / ) Similar to the LDJ1L case, unnecessary β coefficient values are eliminated together with the relevant variables. For the stock price forecast model linear regression with five and more variables is applied as follows: 1 1 2 2t i iP x x x= α + β + β + + β + ε... , (4) where: α and b1, b2, ..., bi – linear regression coefficients; ε – bias for the forecast value; t – time period. Although during correlation analysis phase it was concluded that stock prices of the companies PTR1L and CTS1L correlate with more than five variables, the application of the regression analysis and ANOVA method provides slightly different results as presented in the Table 8. Table 8. Correlation and regression analysis results for the stock price forecast model when cor- relation analysis is misleading Company Stock price forecast model for the period 2010–2014 PTR1L 0 3154 0 1560 1 0 0009 2000 0 0004 8000VLP L GP B GI B GI− + + + + ε. . ( ) . ( ) . ( ) CTS1L 3 45025 0 0005528 1 0 0000001 2000 0 0000001 8000 SAB L P E B GI B GI + + − + ε . . ( / ) . ( ) . ( ) Company General stock price forecast model PTR1L 7 8 91 2000 8000VLP L GP B GI B GIα + β + β + β + ε( ) ( ) ( ) CTS1L 10 11 121 2000 8000SAB L P E B GI B GIα + β + β − β + ε( / ) ( ) ( ) It was established that stock prices of some companies, namely PTR1L and CTS11L, depend on sector indexes i.e. B2000GI (Industrials) and B8000GI (Financials). Due to this reason it is recommended to integrate additional variables i.e. EMA of different 303 Business, Management and Education, 2016, 14(2): 292–307 length (n days) into the stock price forecast model as presented in Table 9 to make more accurate stock price forecasts as it was determined during previous research (Table 9). Table 9. Multiple EMAs to be used as additional variables to forecast stock prices of PTR1L and CTS1L B2000GI (Industrials) Length (days) Length (days) Length (days) Length (days) Δ n-1 48 72 92 54 N 49 73 93 55 Δ n+1 50 74 94 56 B8000GI (Financials) Length (days) Length (days) Length (days) Δ n-1 48 72 92 N 49 73 93 Δ n+1 50 74 94 When adding the variables into the stock price forecast model, three EMA lengths (for example, of 48, 49 and 50 days) should be used. After the research in reference to historical data covering period 2010–2014 was conducted, it was concluded that a theoretical stock price forecast model for companies listed by Nasdaq OMX can be generated. Furthermore, such model can be improved in accordance with its practical application possibilities at the global level with reference to already developed international stock markets. Based on the research results a general stock price forecast model is created (5): 1t i i k k m m n nP P FR MF EMA SI+ = α + β + β + β + β + ε( ) , (5) where: α and bi, bk, bm, bn – linear regression coefficients; Pi – i – company’s stock price; FRk – k – financial ratios of the company under analysis or another company performing in the stock market; MFm – m – macroeconomic factor; EMA – EMA of n days for the selected sector index; SIn – n – sector index; ε – bias for the forecast value. To support the created stock price forecast model it is also useful to understand why the relationship between stock prices of particular companies and sector indexes exists. In light of findings outlined above, there are all assumptions to carry out further research aimed at developing the created stock price forecast model at the global level, which will also enhance its practical application various stock markets. 5. Conclusions This paper examined whether it is possible to create a stock price forecast model which integrates already known methods and models, namely fundamental and technical 304 A. Dzikevičius, S. Šaranda. Formation of an integrated stock price forecast model in Lithuania analyses, can be used practically by investors. With the reference to the companies listed by Nasdaq OMX Baltic, stock prices of 14 selected Lithuanian companies where analyzed, covering the period of 2010–2014 OMX. The researched discovered a strong correlation (R ≥ |0.95|) between stock prices and several variables, such as: stock prices of other companies than these under analysis; financial ratios (in most cases financial ratios of companies other than under review but also listed by Nasdaq OMX); macroeconomic factors; sector indexes. It was established that stock prices of the majority of the companies examined cor- relate with one or several variables, among which P/E, P/BV, ROA and ROE are the generally held to be the most important financial ratios. In some cases the correlation between stock prices and the ratios is positive, while in other cases a negative correla- tion was found. Other important variables include the stock price of companies other than analyzed, EPS and NPM ratios, country’s import and export values as well as B2000GI (Industri- als) and B8000GI (Financials) indexes. For companies stock prices forecast models of which include sector indexes it is rec- ommended to integrate an additional variables EMA of specific length (n days) selected particularly for these sectors. When adding the EMA, at least three different length (n- days) should be used to make stock forecasts more accurate. However, application of the correlation analysis only is not reliable when creating a stock price forecast model because the selected variables may not be significant statis- tically and should be not integrated into the model. For these purposes the regression analysis and ANOVA method coupled with c-Squared Test results is recommended. Potentially, the general stock price forecast model described in this paper can be applied practically to forecast stock prices in various. However, further research is needed in order to understand the drivers behind the relationship between stock prices of particular companies and sector indexes, financial ratios etc. In light of findings outlined above, there are grounds for conducting further research towards developing stock price forecast model. All this will allow to drive and support growth of the economic and strengthen companies’ growth perspectives. In this case, for different stock markets the use of additional check measures is recommended. They include: Goodness of Fit Test (GFI); Tucker-Lewis Index (TLI); Normed Fit Index (NFI). The stock price forecast model is suitable for application when the GFI, TLI or NFI value is close to 0.95. Root Means Square Error Approximation (RMSEA) can be also used as an ad- ditional check measure for the created stock price forecast models. The stock price forecast model can be applied when RMSEA<0.95. 305 Business, Management and Education, 2016, 14(2): 292–307 references Abdullah, D. A.; Hayworth, S. C. 1993. Macroeconometrics of stock price fluctuations, Quarterly Journal of Business and Economics 32: 50–67. Amir, E.; Einhorn, E.; Kama, I. 2012. Extracting sustainable earnings from profit margins, Journal of Accounting, Auditing and Finance 6 (9): 1–47. Bettman, J. L., Sault, S.; Schultz, E. 2009. Fundamental and technical analysis: substitutes or com- plements?, Accounting and Finance 49(1): 21–36. https://doi.org/10.1111/j.1467–629X.2008.00277.x Bildirici, M.; Ersin, Ö. Ö. 2009. Improving forecasts of GARCH family models with the artificial neural networks: an application to the daily returns in Istanbul Stock Exchange, Expert Systems with Applications 36(4): 7355–7362. https://doi.org/10.1016/j.eswa.2008.09.051 Bonga, W. G. 2015. The need for efficient investment: fundamental analysis and technical analysis [online], [cited 21 April 2015]. Social Science Research Network. Available from Internet: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2593315 Ball, R.; Brown, P. 1968. An empirical evaluation of accounting income numbers, Journal of Account- ing Research 6(2): 159–178. https://doi.org/10.2307/2490232 Box, G. E. P.; Tiao, G. C. 1975. Intervention analysis with applications to economic and environmental problems, Journal of the American Statistical Association 70(349): 70–79. https://doi.org/10.1080/01621459.1975.10480264 Bulmash, T. G.; Trivoli, G. W. 1991. Time-lagged interactions between stock prices and selected economic variables, The Journal of Portfolio Management 17: 61–67. https://doi.org/10.3905/jpm.1991.409351 Campbell, J.; Shiller, R. 1988. Stock prices, earnings, and expected dividends, Journal of Finance 43: 661–676. https://doi.org/10.1111/j.1540–6261.1988.tb04598.x Chen, Y. J. 2013. A fundamental analysis-based method for stock market forecasting, in Fourth In- ternational Conference on Intelligent Control and Information Processing, 9–11 June 2013, IEEE, 354–359. https://doi.org/10.1109/ICICIP.2013.6568097 Cheng, M.; Su, C-W.; Tsai, M.-G. 2012. Data pre-processing for artificial neural network applications in prioritizing railroad projects: a practical experience in Taiwan, Journal of Civil Engineering and Management 18(4): 483–494. https://doi.org/10.3846/13923730.2012.699914 Cimpoeru, S. S. 2011. Neural networks and their application in credit risk assessment. Evidence from the Romanian market, Technological and Economic Development of Economy 17(3): 519–534. https://doi.org/10.3846/20294913.2011.606339 Duca, G. 2007. The relationship between the stock market and the economy: experience from inter- national financial markets, Bank of Valetta Review 36: 1–12. Dutta, A.; Bandopadhyay, G.; Sengupta, S. 2012. Prediction of stock performance in the Indian stock market using logistic regression, International Journal of Business and Information 7(1): 105–136. Elleuch, J. 2009. Fundamental analysis strategy and the prediction of stock returns, International Research Journal of Finance and Economics 30(1): 95–107. Ferreira, M. A.; Santa-Clara, P. 2011. Forecasting stock market returns: The Sum of the Parts is More than the Whole, Journal of Financial Economics 100(3): 514–537. https://doi.org/10.1016/j.jfineco.2011.02.003 Guersen, E.; Kayakutlu, G.; Daim, T. U. 2011. Using artificial neural network models in stock market index prediction, Expert Systems with Applications 38: 10389–10397. https://doi.org/10.1016/j.eswa.2011.02.068 https://doi.org/10.1111/j.1467-629X.2008.00277.x https://doi.org/10.1016/j.eswa.2008.09.051 http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2593315 https://doi.org/10.2307/2490232 https://doi.org/10.1080/01621459.1975.10480264 https://doi.org/10.3905/jpm.1991.409351 https://doi.org/10.1111/j.1540-6261.1988.tb04598.x https://doi.org/10.1109/ICICIP.2013.6568097 https://doi.org/10.3846/13923730.2012.699914 https://doi.org/10.3846/20294913.2011.606339 https://doi.org/10.1016/j.jfineco.2011.02.003 https://doi.org/10.1016/j.eswa.2011.02.068 306 A. Dzikevičius, S. Šaranda. Formation of an integrated stock price forecast model in Lithuania Hadavandi, E. 2010. Integration of genetic fuzzy systems and artificial neural networks for stock price forecasting, Knowledge-Based Systems 23(8): 800–808. https://doi.org/10.1016/j.knosys.2010.05.004 Humpe, A.; Macmillan, P. 2009. Can macroeconomic variables explain long-term stock market move- ments? A comparison of the US and Japan, Applied Financial Economics 19: 111–119. https://doi.org/10.1080/09603100701748956 Higgins, R. C. 2009. Analysis for financial management. 9th ed. McGrew Hill. Higgins, H. N. 2011. Forecasting stock price with the residual income model, Review of Quantitative Finance and Accounting May 36(4): 583–604. Hou, T.; Hung, W.; Gao, S. S. 2014. Investors’ reactions to analysts’ forecast revisions and informa- tion uncertainty: evidence of stock price drift, Journal of Accounting, Auditing and Finance 29(3): 238–259. https://doi.org/10.1177/0148558X14530129 Hsing, Y. 2011. Macroeconomic variables and the stock market: the case of Lithuania, The Review of Finance and Banking 3(1): 31–37. Jabbari. E.; Fathi, Z. 2014. Prediction of stock returns using financial ratios based on historical cost, compared with adjusted prices (accounting for inflation) with neural network approach, Indian Journal of Fundamental and Applied Life Sciences 4: 1064–1078. Kariuki, B. W.; Oyugi, L. A. 2013. Testing the residual income valuation model in a Nascent stock market: the case of Nairobi securities exchange, International Journal of Business and Social Science 4(10): 69–78. Kartašova, J.; Venclauskienė, D. 2014. Valuation of fundamental analysis reliability in stock pricing: theoretical approach, in International Scientific Conference Business and Management 2014, 15–16 May 2014, Vilnius, Lithuania. Kim, K. 2003. Dollar exchange rate and stock price: evidence from multivariate cointegration and error correction model, Review of Financial Economics 12: 301–313. https://doi.org/10.1016/S1058–3300(03)00026–0 Kimoto, T.; Asakawa, K.; Yoda, M.; Takeoka, M. 1990. Stock market prediction system with modular neural networks, in International Joint Conference on Neural Networks, 17–21 June 1990, IEEE, 1: 1–6. https://doi.org/10.1109/ijcnn.1990.137535 Kothari, S. P.; Shanken, J. 1997. Book-to-market, dividend yield, and expected market returns: a time- series analysis, Journal of Financial Economics 44: 169–203. https://doi.org/10.1016/S0304–405X(97)00002–0 Lamont, O. 1998, Earnings and expected returns, Journal of Finance 53: 1563–1587. https://doi.org/10.1111/0022–1082.00065 Li, Y. et al. 2014. A combination prediction model of stock composite index based on artificial intelli- gent methods and multi-agent simulation, International Journal of Computational Intelligence Systems 7(5): 853–864. https://doi.org/10.1080/18756891.2013.876722 Mondal, P. et al. 2014. Study of effectiveness of time series modeling (ARIMA) in forecasting stock prices, International Journal of Computer Science, Engineering and Applications 4(2): 13–29. Mostafa, M. 2010. Forecasting stock exchange movements using neural networks: empirical evidence from Kuwait, Expert Systems with Applications 37: 6302–6309. https://doi.org/10.1016/j.eswa.2010.02.091 Öğüt, H.; Doğanay, E. E.; Aktaş, R. 2009. Detecting stock-price manipulation in an emerging market: the case of Turkey, Expert Systems with Applications 36(9): 11944–11949. https://doi.org/10.1016/j.eswa.2009.03.065 https://doi.org/10.1016/j.knosys.2010.05.004 https://doi.org/10.1080/09603100701748956 https://doi.org/10.1177/0148558X14530129 https://doi.org/10.1016/S1058-3300(03)00026-0 https://doi.org/10.1109/ijcnn.1990.137535 https://doi.org/10.1016/S0304-405X(97)00002-0 https://doi.org/10.1111/0022-1082.00065 https://doi.org/10.1080/18756891.2013.876722 https://doi.org/10.1016/j.eswa.2010.02.091 https://doi.org/10.1016/j.eswa.2009.03.065 307 Business, Management and Education, 2016, 14(2): 292–307 Olson, D.; Mossman, C. 2003. Neural network forecasts of Canadian stock returns using accounting ratios, International Journal of Forecasting 19(3): 453–465. https://doi.org/10.1016/S0169–2070(02)00058–4 Omran, M.; Ragab, A. 2004. Linear versus non-linear relationships between financial ratios and stock returns: empirical evidence from Egyptian firms, Review of Accounting and Finance 3(2): 84–102. https://doi.org/10.1108/eb043404 Petcharabul, P.; Romprasert, S. 2014. Technology industry on financial ratios and stock returns, Jour- nal of Business and Economics 5(5): 739–746. Pilinkus, D. 2009. Stock market and macroeconomic variables: evidences from Lithuania, Economics and Management 14: 884–891. Pilinkus, D. 2010. Stock market and macroeconomic variables evidences from Lithuania, Economics and Management 15: 884–889. Pontiff, J.; Schall, L. 1998. Book-to-market ratios as predictors of market returns, Journal of Financial Economics 49: 141–160. https://doi.org/10.1016/S0304–405X(98)00020–8 Preethi, G.; Santhi, B. 2012. Stock market forecasting techniques: a survey, Journal of Theoretical and Applied Information Technology 46(1): 24–30. Purnamasari, D. 2015. The effect of changes in return on assets, return on equity, and economic value added to the stock price changes and its impact on earnings per share, Research Journal of Finance and Accounting 6 (6): 80–90. Ratanapakorn, O.; Sharma, C. 2007. Dynamic analysis between the US stock returns and the macro- economic variables, Applied Financial Economics 17(5): 369–377. Renfro, B. K. 2015. Stock price reaction to analysts’ EPS forecast error, Journal of Finance and Ac- countancy 15: 1–6. Sarikhani, M.; Ebrahimi, F. 2012. An empirical evaluation of using the residual income model for prediction of stock price, African Journal of Business Management 6(5): 2043–2047. Tareq, M.A. 2012. Is residual income model (RIM) really superior to dividend discount model (DDM)?, Journal of Business and Management 5(6): 36–44. https://doi.org/10.2139/ssrn.2179059 Tseng, K.-C.; Kwon, O.; Tjung, L. C. 2012. Time series and neural network forecast of daily stock prices, Investment Management and Financial Innovations 9(1): 32–54. Wu, C.; Luo, P.; Li, Y.; Chen, K. 2015. Stock price forecasting: hybrid model of artificial intelligent methods, Engineering Economics 26(1): 40–48. https://doi.org/10.5755/j01.ee.26.1.3836 Zavadskas, E. K.; Turskis, Z. 2011. Multiple criteria decision making (MCDM) methods in economics: an overview, Technological and Economic Development of Economy 17(2): 397–427. https://doi.org/10.3846/20294913.2011.593291 Audrius DZIKEVIČIUS is Associated Professor at the Vilnius Gediminas Technical University, Department of Finance Engineering, since 2007. His scholarly interests cover areas such as manage- ment of portfolio risk, forecasting and modelling of financial markets, appraisal of business by qualita- tive methods, strategic corporate finance management decisions. Svetlana ŠArANDA completed master studies at the Vilnius Gediminas Technical University, Department of Finance Engineering, in 2012. Her scholarly interests cover areas such as forecasting and modelling of financial markets and market analysis. https://doi.org/10.1016/S0169-2070(02)00058-4 https://doi.org/10.1108/eb043404 https://doi.org/10.1016/S0304-405X(98)00020-8 https://doi.org/10.2139/ssrn.2179059 https://doi.org/10.5755/j01.ee.26.1.3836 https://doi.org/10.3846/20294913.2011.593291