ABSTRACT The paper proposes a new integrated realized stochastic volatility–mixed data sampling–geopolitical risk (RSV–MIDAS–GPR) model to model and forecast crude oil futures volatility. The model jointly models returns and the realized measure of volatility, leverages contemporaneous volatility information, and captures the effects of GPR on crude oil futures volatility. The empirical results demonstrate a significant positive correlation between GPR and crude oil futures volatility. Meanwhile, the RSV–MIDAS–GPR model, which incorporates both GPR and realized volatility, exhibits a synergistic effect, leading to a substantial improvement in out‐of‐sample forecasting performance. Furthermore, the model demonstrates notable capability in identifying high‐volatility states and achieves higher forecasting accuracy than competing models during market turmoil. Finally, economic value tests confirm that the inclusion of GPR provides valuable guidance for investor decision‐making. These findings offer both methodological and empirical contributions to the related research field.
Yang et al. (Thu,) studied this question.