The Approach Of Bivariate Extreme Theory For Aanlyzing The Dynamic Nonstationary Times Series

Financial time series,such as stock index,exchange rate income,etc.have characteristics of heavy tail, variance fluctuation and strong dependence.It is comparatively difficult to use the traditional analytical method.This text utilize the GARCH model to get rid of dependence and fluctuation among the data and then utilize the EVT(Extreme Value Theory) to analyse the iid residualswith heavy tails.After introducing the concept of Copula, the proceeding of estimation is divided into two parts, first estimate all parameter of marginal distribution and second step of consideration is dependence structure among the random variables.In a situation of multidimension, point process method is better method,can get only for the vectorial distribution law at random of the extreme value on one side.This text derives through the theory to get a set of standards method of financial assets VaR alone of calculation, and receive the quantitative theory analysis result to the relation of influencing among the financial assets. With analying the stock index income of Shenzhen, shanghai it gets good results and tests it.