This paper investigates how to improve prediction accuracy of stock realized volatility using a large set of predictors. Exploiting normalized positive adjusted R-square and significant t statistic of predictor obtained from the in-sample result as weight, we develop two simple and effective...