Prior literature finds that variables that can forecast market returns in sample do not beat historical averages in forecasting market returns out of sample. We propose a naïve model averaging (NMA) method, which produces mostly positive out-of-sample R2s for the variables that are significant...