The article examines the volatility forecasting and option pricing performance of model‐free implied volatility (MFIV) in comparison to that of the forecasts based on model‐free realized volatility (RV). There is evidence that the forecasts based on RV are significantly more efficient and less biased than those based on MFIV, whereas the option prices based on MFIV are significantly more efficient and less biased than those based on RV. These contrasting results can be reconciled by accounting for the volatility risk premium (VRP), which is found to follow an autoregressive process in this study. The significant daily returns, observed for various option strategies used to exploit the VRP, are substantially reduced, when the normal transaction costs are accounted for. Although the VRP is priced in the Indian options market, it can provide economic benefits only to those option writers, who have sufficiently low transaction costs. © 2014 Wiley Periodicals, Inc. Jrl Fut Mark