【摘要】Nonlinear time series models GARCH and Extreme value theory are employed to explore the relationship between the volatility and exreme value of stock returns.Given the pass information,we assume the innovation follows the conditional t-distribution and generalized error distribution,then we fit ARMA models to the retum level and GARCH,EGARCH and TGARCH models to the volatility of the rerurn.Finally we obtain the tails indices for the innovation and the returns by using the statistical extreme value method,and the statistical results show that the tail indices of returns coincide with the one of innovations for the EGARCH models.