Intraday volatility predictability in china gold futures market: The case of last half-hour realized volatility forecasting

Abstract

This study investigates the intraday realized volatility (RV) forecasting of gold futures in China. To predict the RV in the last half an hour before the day trading close (LH), we decompose the whole trading period into several intervals. The empirical results show that gold futures RVs in day trading intervals can predict the RV in LH, with RV in ROD interval (from night trading close to the last half an hour before day trading close) having stronger predictive power. Models which use gold futures RVs in separate partitions during ROD deliver better forecasting performance in and out of sample. Meanwhile, gold futures RVs in night trading are also informative even after controlling the RVs in day trading. Further, we reconfirm the predictive power of US gold futures information from the intraday perspective.