Comparing Accuracy of State Space Model and Ordinary Least Squares (OLS) in Predicting Stock Return by Fama and French Three-Factor Model in Tehran Stock Exchange

Predicting stock returns is one of the major issues to be discussed in the financial literature, and investment. Researchers have proposed various methods for predicting stock returns, that the most famous of them are the Capital Asset Pricing Model by Sharpe and Lintner, arbitrage pricing model by ross and three factors model by Fama and French. F& F three-factor model as the most significant factor models in recent years great attention has been. Despite having many strengths of this model is based on the assumption of constant beta coefficient is founded, However, this assumption does not hold absolute in any circumstances. In this study, we tried to model with constant or variable coefficients fitted separately and then compare the accuracy each of them. For this purpose, the state space model and ordinary least squares (OLS) models were fit assuming constant and variable coefficients are used. This research will be done on listed companies in Tehran Stock Exchange for a period of 72 months (October 1385 to September1391). The results show that, compared to state-space model of a linear least squares model for predicting stock returns has a better performance, this means that Beta coefficients in three-factor is on the Tehran Stock Exchange are not constant.