Share Price Prediction and Analysis - Essay Example

? Share Price Prediction and Analysis Introduction Share valuation is the process of calculated the actual value of a given company through the use of its shares. Share valuation helps investors understand not only the performance of a given organization but also make appropriate decisions in as far as investment is concerned (Chen, et al., 2003). Predicting and analyzing share prices of various organization have ignited a lot of interest especially amongst investors who wish to put their funds in shares that they are sure will grow in future. As a result, there are, in the literature, various approaches to using data in financial statements to predict share prices over the medium term (for example one to three years). The following is a critical review of such literature. In addition, the discussion applies a synthesis of two approaches/models identified to predict the share prices for Tesco Plc from the publications of the firm’s financial statements for 2008 and 2009. Lastly, this discussion attempts to test the approach by comparing these two sets of predictions with actual share prices. A concluding remark, which comments on the results, winds up the paper. Approaches/Models for Predicting Share Prices In short-term or medium-term, different models or approaches are used in predicting the future prices of shares of various companies. Share prices of companies may take different forms such as linear, horizontal, cyclic, or seasonal as influenced by prevailing market and environmental factors (Hassan, et al., 2007). Due to lack of prediction methods that provide least prediction error, investors tend to apply numerous methods thereby comparing their results in a bid to finding the best model or approach to use (Chen, et al., 2003). Examples of models or approaches commonly used in predicting the prices of share are regression method, Artificial Neural Networks (ANN), extended Kalman filter data fusion, decision tree (DT) model, and Residual Income Model (RIM) amongst others. Different valuation models have also been used to estimate and predict the share prices of different companies. For instance, Dividend Valuation Models, Market-Value-Added, and Multiplier Methods have been used by investors in order to predict the share prices of different companies within a given stock exchange market. Artificial Neural Network (ANN) is a share price prediction method that is commonly used. For many years, ANN has been developed and restructured in order to provide efficient and effective performances on predicting share prices of firms in a stock exchange for purposes of investment (Tom, et al, 2000). Nonetheless, most predictors used single dosage of ANN (Kim and Shin, 2007). Application of single dosage in predicting share prices rarely provides an opportunity to discover the decision rule that the model uses while making the predictions (Hassan, et al, 2007). Artificial Neural Network is a share price prediction model or approach, which is created through stimulation of biological central nervous system of investors or predictors (Swales and Yoon, 2002). One of the reasons explaining its extensive application is the ability to predict share prices from large databases (Olson and Mossman, 2003). The idea of back-propagation algorithm is the basis of Artificial Neural Network in predicting share prices of firms. ANN back propagation function is usually represented by the following function: Where, xi is the sum of inputs, which is multiplied by their respective weights wji; Aj is the predicted share value under the ANN model; and n is the end period in which the valuation is carried out. Decision tree (DT) model on the other hand is a data mining model or approach used in predicting or forecasting share prices within a stock exchange market. One of the reasons for its extensive application is the fact that DT has an excellent ability and capability of describing cause as well as effect relationships of various stock prices. From the concepts or application of DT, investors are able to explain various causes and effects of their predictions in respect to share prices. In addition, decision tree has the ability of providing reasonable classification of various performances related to prediction of share prices. Unlike the ANN, which has a back propagation algorithm, decision tree has the tree structure with branches that are derived from the situation under analysis (Chen, et al, 2003). In every tree structure of decision tree model, there are the nodes and branches each representing an output or target class and process or decision that will be involved in making classification respectively. At the end of every node within the decision tree structure is the prediction results that an investor decides to use (Hassan, et al, 2007). Nevertheless, decision tree is ineffective without pruning; a process whereby error rate is computed for purposes of improving prediction and classification abilities of decisions in respect to making investments. Another commonly applied model in predicting share prices of firms within a stock exchange is the Kalman filter model, which was developed by Rudolf E. Kalman. Kalman filter model represents mathematical equations encompassing optimal estimators, predictors, as well as correctors (Eugene and Houston, 2004). The idea of combining these three aspects (estimator, predictor, and corrector) is to minimize possibility of having large estimation error covariance (Wang and Chan, 2006). Notably, unlike other models used in predicting share prices of various companies, Kalman filter model is effectively and efficiently applicable to data that is considered to be having normal distribution. In order to compute or predict share prices through application of Kalman filter model, there is the need to fusion two parameters, which contribute to effectiveness and efficiency of the prediction (Ramazan, 2006). The two parameters, which are fused to come up with a model for forecasting or predicting share prices are ; And In this case, is the weighted average obtained from two measuring devices, is the variance of error that is obtained from measuring various devices, and represents the estimated value of parameter that is under measurement within time of t2. Despite being an effective model in measuring and predicting share prices, Kalman filter model is only applicable to normally distributed share prices (Eugene and Daves, 2002). In this perspective, investors or predictors must evaluate whether the share prices represented are normally distributed before engaging in predicting share prices (Chen, et al, 2003). When share prices that take other forms of distributions are predicted through the use of Kalman filter model, the obtained results are unreliable and not valid as when normally distributed shares are predicted through the model (Eugene and Houston, 2004). In this regards, Kalman filter model should strictly be used within share prices that are normally distributed. Residual Income Model (RIM) is another way of predicting or forecasting share prices of various firms within a given stock exchange market (Chen, et al, 2003). Being a theoretical model, RIM is able to link share prices to book value thereby providing a vista to predict or forecast future share prices (Eugene and Daves, 2002). In most cases, RIM derives its approach from the concept of Dividend Discount Model (DDM). From the concepts of DDM, the present value of a share with expected future dividends is given by the following function: , Where Pt is the share price, rt is the discount rate, and dt is the dividend all at a given time, t. In this function RIM attempts to correlate cumulative dividend at a given time t, with a series of discounted dividends (Kunhuang & Yu, 2006). From the concept of RIM, it is obvious that one should be able to predict dividends in order to be able to forecast or predict estimates of share prices of a given company within a specific duration (Eugene and Daves, 2002). There are cases where Residual Income Model applies the concept of regression analysis especially in explaining and analyzing the idea of correlation between dividends and the share prices of a given organization (Roh, 2007). Residual Income Model in Predicting Shares of Tesco Plc Residual Income Model is represented by For instance, in predicting the value of shares for January 2008, the December 2008 dividend will be used and time will be 1 year since we are predicting shares of the following year. In this case, the predicted share will be as follows: = 417.5 Method If you look at it keenly then you will be able to see that the time, k starts from 0 to infinity. The sum of all the 1+0.07 raised to power -1 are meant to be obtained from when k is 0 to infinity as no one knows when the company will stop operations. From the concept of finance, the ending time is ? (infinity), the calculated value of the above formula will be 417.5 as the book value of the share in January 2008. This is done for all the years represented in the table below. From the discussed models or approaches to predicting shares of firms within a given stock exchange, RIM will be used to predict shares of Tesco Plc between 2008 and 2009. The following table indicates the predicted values as well as the actual values of share prices for Tesco Plc between 2008 and 2009: Date Predicted Share Price Actual Share Price Dec-09 425.15 428 Nov-09 427.8 423 Oct-09 408.95 407.35 Sep-09 398.4 399.6 Aug-09 376.9 375.9 Jul-09 374.5 367.5 Jun-09 359.8 353.6 May-09 372.1 364.9 Apr-09 346.9 337.2 Mar-09 321 333.4 Feb-09 336.6 333.2 Jan-09 361.4 358.2 Dec-08 361.9 360 Nov-08 304 295.3 Oct-08 318 339.4 Sep-08 369 387.6 Aug-08 384.8 381.5 Jul-08 373.3 360.4 Jun-08 371 378 May-08 419.5 411.8 Apr-08 418 429 Mar-08 386 379 Feb-08 398.5 400.5 Jan-08 417.5 410.75 Residual income model has been chosen due to its simplicity, accuracy, and the fact that it takes into account all factors within the entire period. In addition, the model has been chosen given its theoretical nature hence the possibility of linking the share value (predicted) to book value. Nonetheless, it should be noted that RIM’s idea of being theoretical makes it prone to errors of exaggerations. Conclusion From the above discussions, it is evident that there are different models applicable in predicting shares of a given firm. Some of these models include Artificial Neural Networks (ANN), extended Kalman filter data fusion, decision tree (DT) model, and Residual Income Model (RIM) amongst others. Nonetheless, there have been problems in identifying the best model to apply due to lack of prediction methods that provide least prediction error, investors tend to apply numerous methods thereby comparing their results in a bid to finding the best model or approach to use. Though there are differences in between predicted values and actual share prices, there is a possibility that RIM can be used in forecasting shares of a given company. Therefore, investors should not shy away from using RIM as a model for predicting share prices. Bibliography Chen, A., Leung, M. and Daouk, H., 2003, “Application of neural networks to an emerging financial market: forecasting and trading the Taiwan Stock Index“ ,Computers & Operations Research, Vol.30, Pp. 901-923. Eugene, B. and Daves, P., 2002, Intermediate Financial Management. Seventh Edition, South Western.Copeland, Eugene, B., and Houston, R., 2004, Fundamentals of Financial Management. Thomson/South-Western. Hassan, H., Hamadi, K. & Saleem, S., 2007, “Towards Evaluation of Phonics Method for Teaching of Reading Using Artificial Neural Networks (A Cognitive Modeling Approach)”, IEEE International Symposium on Signal Processing and Information Technology, Pp. 855-862. Kim, H. and Shin, K, 2007, “A hybrid approach based on neural networks and genetic algorithms for detecting temporal patterns in stock markets”, Applied Soft Computing, Vol. 7, pp. 569–576. Kunhuang, H. & Yu, T., 2006, “The application of neural networks to forecast fuzzy time series”, Physical A: Statistical Mechanics and its Applications, Vol. 363, No. 2, Pp. 481-491. Olson, D. and Mossman, C., 2003, “Neural network forecasts of Canadian stock returns using accounting ratios”, International Journal of Forecasting, Vol. 19, No. 3, Pp. 453-465. Ramazan, G., 2006, “Non-linear prediction of security returns with moving average rules”, Journal of Forecasting, Vol. 15, No. 3, Pp. 165-174. Roh, T., 2007, “Forecasting the volatility of stock price index”, Expert Systems with Applications: An International Journal, Vol. 33, Pp. 916-922. Swales, G. and Yoon, Y., 2002, “Applying artificial neural network to investment analysis”, Financial Analysts Journal, Pp. 78-80. Tom, T., Koller, T., and Murin, J, 2000, Valuation: Measuring and Managing the Value of Companies. Third Edition, John Wiley& Sons, New York Wang, J. and Chan, H., 2006, “Stock market trading rule discovery using two-layer bias decision tree”, Expert Systems with Applications, Vol. 30, Pp. 605–611. Read More