Abstract: Selecting the 1550-day closing price data of the China Securities Liquor Index, based on the T distribution and the generalized error distribution (GED) GARCH family model, it is found that the CGARCH-M model under the GED distribution can reflect the non-normality, heteroscedasticity and volatility of returns. Leverage effect, better fitting effect of the China Securities Liquor Index return; using the weighted historical simulation method and the CGARCH-M-GED model to calculate the value at risk of the index, it is found that the weighted historical simulation method is easy to use in the low volatility period and the unstable volatility period. Overestimating market risk, the model is invalid in high volatility periods; according to the prediction failure rate and Kupiec test results, the asymmetric GARCH model can obtain better risk prediction effects for the China Securities Liquor Index in each volatility period.

Key Words: the China Securities Liquor Index;the closing price ; the risk prediction