Abstract: This paper applies Fractional Integrated Autoregressive Moving Average(ARFIMA)model and Hyperbolic Memory Generalized Autoregressive Conditional Heteroscedasticity(HYGARCH)model, Fractional Integrated Asymmetric Power Autoregressive Conditional Heteroscedasticity(FIAPARCH) model and Fractional Integrated Exponential Generalized Autoregressive Conditional Heteroscedasticity (FIEGARCH)model to capture some stylized facts of conditional volatility and conditional return of financial markets,and apply Skew Student t Distribution(SKST)to capture return distribution,and then measure dynamic risk of financial markets.At last,we use Likelihood Ratio Test(LRT)and Dynamic Quantile Regression(DQR)to test accuracy of risk measurement model as well.Our results show that all risk models used in this paper has no significant difference on accuracy for Chinese stock markets;ARFIMA(1,d,1)-FIAPARCH(1,d,1)-SKST model is no excel to ARFIMA(1,d,1)–HYGARCH(1,d,1)SKST model in developed market;ARFIMA(1,d,1)-FIEGARCH(1,d,1)-SKST model can not measure risk accurately for developed market.

Key words: financial markets, stylized facts, HYGARCH, dynamic risk, measurement