Abstract: Preliminary data analysis shows that the return rates distribution of SSE is fat-tailed and doesn’t obey normal distribution and there is"leverage effect"in Shanghai Stock market.In this paper,we propose an APARCH model with three different distributions assumption to estimate conditional VaR.This model is then compared with the GARCH model under the corresponding three distributions assumption.Using back-testing of historical daily return series we show that the APARCH model yields statistically valid VaR measures and gives better one-day ahead estimates that the GARCH model.

Key words: Value at Risk, APARCH Model, GARCH model, fat tails, leverage effect