Forecasting oil futures realized range-based volatility with jumps, leverage effect, and regime switching: New evidence from MIDAS models

This study adds ample evidences on forecasting oil futures realized range-based volatility (RRV) in the mixed data sampling (MIDAS) framework. Considering the features and frequent extreme risk in oil futures trading in practice, we investigate the effect of jumps, leverage effect, and regime switching. The results highlight the great importance of combing jumps, leverage effect, and regime switching simultaneously in oil futures RRV forecasting, which still holds the best predictability during a highly volatile state. Moreover, considering regime switching can also improve forecast accuracy for longer forecasting horizons such as weekly, biweekly, and monthly. The conclusions are robust to various settings and criteria. Our findings are essential for oil enterprises and quantitative investors to have a good command of crude oil futures market characteristics and achieve risk aversion.