The GARCH model is used in simulating the volatility and VaR of the financial assets. The paper established an EGARCH-GED model to calculate the time varying VaR. Compared the VaR of the EGARCH-GED model and the GARCH model under the normal distribution and T distribution respectively, The paper checked the anticipated VaR in the previous step by employing failure rate test and back-testing. The result shows that GED distribution is fitted with the fat tail feature of the financial assets. Under different confident levels, the VaR predicated by EGARCH-GED is more accurate and has more low level risk to be overestimated or underestimated.