Research On Extreme Value Risk Of Shanghai And Shenzhen Stock 300 Index Futures Based On GARCH Model

Since CSI 300 stock futures was officially established in 2010,not only opening another milestone in the financial markets and ending the China's conservative unilateral state,but also attractting more and more scholars for the futures market risk analysis.The quantification of market risk has become an important measurement index of financial instruments,and it is very important in theory and practice.As a kind of financial direction indicator,VaR has been widely used,but its calculation method is always changing and innovated,in order to achieve a better practical effect.Due to the inherent defects of VaR,ES which has a better effect of measuring risk has been used and proved by more and more scholars.In order to promote the healthy development of the futures market and promote the economy,the precision of the risk has become an important research content.In this paper,we review the evolution of VaR and ES,and measure the risk of the empirical data about CSI 300 stock futures.It is not difficult to find that the kind of GARCH models which as the computation of dynamic risk has become the hot research method.So in this paper GARCH?EGARCH?TGARCH and emerging Realized GARCH are selected with norm?t distribution?skewed-t and generalized hyperbolic distribution measuring the risk of data.The results show that the Realized GARCH has a better effect,and has more better effect under the generalized hyperbolic distribution.Aimed at effectively measuring the tail of the yield,the extreme value theory is used in the following chapters.The kurtosis method is used for computing threshold and the GPD test is used for testing,the paper use GARCH-EVT formula calculate the risk after testing.The addition of extreme value theory is helpful to the improvement of the risk measurement in higher confidence,which shows that the extreme value theory has a great advantage in the tail.The comprehensive analysis shows that Realized GARCH has improved the estimation effect by adding a series of realized fluctuation,and has the best effect under the generalized hyperbolic distribution.At the end of this paper,the empirical analysis of without extreme value theory and extreme value theory are summarized,and the research shortages and future research directions are discussed.