Hosein Maghsoud - PhD Candidate Science and Research Branch, Islamic Azad University Tehran, Iran
Fraydoon Rahnamay Roodposhti - Professor Faculty Member Department of Accounting, Science and Research Branch, Islamic Azad University Tehran, Iran (Correspond Author.)
Hamidreza Vakilifard - Assistant Professor and Faculty Member Department of Accounting, Science and Research Branch, Islamic Azad University Tehran, Iran
Taghi Torabi - Assistant Professor and Faculty Member Department of Economy, Science and Research Branch, Islamic Azad University Tehran, Iran

In this study, ۳ models of Time-Varying Parameters (TVP), Dynamic Model Selection (DMS) and Dynamic Model Averaging (DMA) and a comparison with the Ordinary Least Squares (OLS) method in MATLAB in the time period ۲۰۰۳-۲۰۱۳ (with data on a monthly basis) are discussed. In the present study, the variables of unofficial exchange rate changes, interest rate changes and inflation in oil price forecast returns for stocks in Tehran Stock Exchange are used. The study concludes that dynamic models with time-varying parameters are more accurate in predicting returns in the Stock Exchange, in a way that the MAFE and MSFE models, DMA, DMS which have complete dynamics are more efficient than other models. As a consequence, it can be said that the variability of the coefficients of the variables in the TVP model cannot lead to higher accuracy in predicting returns in the Stock Exchange, and it is required that the dynamics of time-varying variables of the model used to predict stock returns be taken into consideration

Macro indexes, optimal portfolio, Stock Returns, time-varying parameter method, dynamic models