-we build the threshold-based state-dependent autoregressive jump intensity-GARCH model to study smooth fluctuations and jump changes of individual stocks listed in China. It makes the jump intensity not only driven by idiosyncratic factors, but impacted by external state variables. Comparing with the existing models, it boasts better goodness of fit and explanatory ability. It not only can more clearly describe the time variation, clustering and threshold-based state-dependence of jump changes as well as capture the asymmetry and clustering in normal volatility, but also can better illustrate the risk contagion between cross-shareholding companies.