A comparison between,CAPM,Fama and French,s models and artificial neural networks in predicting the Iranian stock Market

Comparison between the Capital Asset Pricing model,Fama and Ferench three factors model and Artificial Neural Network model in predicting Tehran stock Exchange returns is discussed in this research.the first two models are linear and the following are nonlinear.Four hypotheses have been designed for this purpose.To examine these hypotheses,the expected return was calculated daily during 1383 to 1387 for 110 companies.companies in each quarter have divided to 6 portfolios by size and book to market value factors. Results showed that the performance of Fama &Ferench three factors model is better than Capital Asset pricing model.Also Univariable and Multyvariable Artificial Neural Network models have better performance in compare with their corresponding nonlinear models.