Research of Jump Durations for CSI300 Index Futures based on Realized Volatility and Improved ACH Models

we construct the ADS jump detection based on realized kernel and median realized volatility. The realized volatility of CSI300 index futures is decomposed into jump volatility and continuous volatility. Augmented ACH models including week effect ACH model and intraday ACH model are proposed to describe the dynamic dependency of jump durations. Numerous empirical studies were conducted, drawing the conclusions: daily jumps exhibit weekday effects, while intraday jumps display intraday effects, jumps and jump durations exhibit non-normality and self-correlation Our augmented ACH models can interpret the dynamic dependency of jump durations very well, and the hazard rate of augmented ACH models is volatile and sensitive. This paper presents useful conclusions for investors and regulator to understand the market volatility and grasp the market risk.