The predictability of average skewness is also statistically and economically significant out of sample. When the next-month market return is predicted out of sample using a recursive window, we find that value-weighted average skewness (in combination with market excess return or not) has the highest predictive power among the macroeconomic and financial variables that we consider. Implementing a strategy based on the predictive regression including the value-weighted average skewness allows the investor to generate better performance compared with a strategy that is based on other predictors. The annualized returns are equal to 15.8% and 14.6% for the strategies based on the value-weighted and equal-weighted average skewness, respectively, and the buy-and-hold strategy and the strategy based on the historical mean produce annualized returns equal to 10.5% and 9.6%, respectively. Annualized Sharpe ratios are equal to 0.62, 0.57, 0.50, and 0.38, respectively.