Objective: The aim of the present study is to analyze and test the power of Conditional Capital Asset Pricing Model (CAPM) with Time Variant Beta against Standard Capital Asset Pricing Model to find the better model to explain expected return of stocks. Methods: Using monthly data, beta value was estimated using standard CAPM and Multivariate GARCH methods for companies included in the statistical sample. Based on these two methods, the expected returns of the next year to test out-of-sample performance were calculated by eliminating 12 months from the top and adding 12 months from the bottom. The same process was repeated for the following years. Then, the accuracy of each of these models was examined using criteria MAE and MSE. Results: Using paired t-test and Diebold-Mariano test, we tested the research hypotheses and the results were presented based on MAE and MSE indices. The results showed that according to both criteria in MAE and MSE, the conditional CAPM models, whether based on full rank BEKK or diagonal BEKK, can have better performance than the standard CAPM model. Conclusion: Regarding the findings and better predictive power of conditional CAPM based on full rank BEKK and/or diagonal BEKK, in terms of MAE and MSE criteria, replacing the standard model with these models can result in higher accuracy.