Abstract: Based on the time-varying jump of high-frequency data, the one-minute closing price of copper-aluminum futures in Shanghai Futures Exchange from 2010 to 2015 is selected as sample data, and the realized variance (RV) of copper-aluminum futures high-frequency data is decomposed into continuous sample path variance (CV) and discrete jump variance (JV). The DCC-MVGARCH model is used to calculate the dynamic correlation coefficients between the continuous sample path variance and the discrete jump variance. The results show that there is an obvious positive correlation between the volatility of copper and aluminum futures, the correlation between CV and JV is significantly different in dynamic path, and the former correlation degree is much higher than the latter; under the shock of extreme events such as the European debt crisis, the dynamic correlation between CV and JV reaches to a high point to some extent.

Key words: jump, realized variance, futures market, dynamic correlation, high-frequency data