Excessive monetary occasions normally result in sharp jumps in inventory costs and volatilities. As well as, bounce clustering and inventory worth correlations contribute to the danger amplification acceleration mechanism through the disaster. On this paper, 4 Soar-GARCH fashions are used to forecast the bounce diffusion volatility, which is used as the danger issue. The linear and uneven nonlinear results are thought-about, and the worth prone to banks is estimated by help vector quantile regression. There are three essential findings. First, by way of the volatility means of financial institution inventory worth, the Soar Diffusion GARCH mannequin is healthier than the Steady Diffusion GARCH mannequin, and the discrete bounce volatility is important. Secondly, because of the distinction of the sensitivity of irregular info shock, the bounce habits of financial institution inventory worth is heterogeneous. Furthermore, CJ-GARCH fashions are appropriate for many banks, whereas ARJI-R2-GARCH fashions are extra appropriate for small and medium sized banks. Thirdly, primarily based on the bounce diffusion volatility info, the efficiency of the help vector quantile regression is healthier than that of the parametric quantile regression and nonparametric quantile regression.