Prediction of stock returns has always been one of the most important issues in finance. Investors have attracted to use of Fama-French Five-Factor Model (FFFFM) as one of the powerful methods for pricing financial assets and predicting the stock returns. This research investigates the predictability of stock returns by including some important firms features namely cash holdings, dividend rate, and free cash flow to equity to FFFFM. Statistical samples consist of 75 companies listed on the Tehran Stock Exchange (TSE) during 2003-2017. The results of panel data test indicate positive significant effects of all variables in FFFFM (i.e. book to market value ratio, company size, growth opportunity, profitability, and investment) as well as newly added firms feature variables (cash holding, dividend rate, and free cash flow to equity). However, the investment has a negative impact on the returns due to the initial estimate of primary FFFFM. In addition, the results indicate that the inclusion of firms feature variables significantly improve the predictive power of stock returns. Finally, by comparing the predictive power of the models, the best prediction model is determined.