Authors : E. Montakhabai Nodeh, M. Kheirandish and M.H. Ranjbar
Abstract: Predicting the future is always a necessity in everyday life and as a common sphere in several disciplines has been discussed. One of the areas that forecast in it has the particular importance in matters related to financial and economic fields. Present study compared investigates the linear and neural network model ability to predict the returns of listed companies on the Stock Exchange of Tehran. In the present study, with causal correlation method has been done the statistical population includes all companies listed in Tehran Stock Exchange which their information for the period of 2011-2013 is available. Order to prediction from daily stock returns of companies active in the stock and the independent variables net profit to asset, sale asset, the ratio of profit to sale, operating profit to sale, operating profit to gross profit and returns of investments have been used. For linear model from the multivariate linear regression method and for neural network model from the multilayer architecture with back-propagation algorithm has been used. Results of the research showed that both linear and artificial neural network models are able to predict stock returns. But, the accuracy of neural network in this forecast is higher and this shows the superiority of artificial neural network against multivariate linear regression model and artificial neural network capabilities in this forecast is confirmed.
E. Montakhabai Nodeh, M. Kheirandish and M.H. Ranjbar, 2015. Comparative Evaluation of Linear and Neural Network Model Ability to Predict the Returns of Listed Companies in Tehran Stock Exchange. International Business Management, 9: 1805-1808.