Forecasting Chinese Stock Market Volatility With Intraday and Overnight Volatility Components of INE Oil Futures

Abstract

This paper investigates the role of different volatility components of the Shanghai International Energy Exchange (INE) oil futures, including intraday, overnight, and the first half-hour components, in forecasting Chinese stock market volatility. Using 5-min realized volatility (RV) as realized volatility measure (RM), the log-HAR models are applied to generate one-step-ahead forecasts for three Chinese stock indices (CSI 300, SHSE and SZSE). Our out-of-sample results show that the model extended with 5-min RV of INE oil futures does not generate more accurate volatility forecasts than the baseline log-HAR model. However, the overnight volatility of INE oil futures significantly improves forecasting accuracy. Our results are robust across different estimation schemes, estimation windows, out-of-sample periods, and evaluation methods. Additionally, using Bi-Power Variation (BPV) as an alternative RM yields consistent results. Overall, the results highlight the importance of incorporating the overnight volatility component of INE oil futures in forecasting Chinese stock market volatility.