Abstract: This paper Appies FIAPARCH and skew student t distribution to capture conditional volatility and skew distribution in financial return respectively, then measures dynamic Value at Risk(VaR), and uses Back-testing and Dynamic Quantile Regression(DQR) to test accuracies of different risk models.Our results indicate that RiskMetrics model and GARCH-Normal model can not measure dynamic financial markets risk accurately;skew student t distribution is a more fit distribution of financial conditional return than standard student t distribution and standard Normal distribution;and, AR-FIAPARCH-SKST model is the best risk measurement model among all models studied in this paper.

Key words: financial market, stylized facts, FIAPARCH, skew distribution, dynamic VaR measure