Based On The Research On Chinese Stock Index Futures Risk Var

On April16th,2010, China launched the Shanghai&Shenzhen300stock index futures. Since then, the Shanghai&Shenzhen300index futures experienced an active trading and relatively stable operation period. But the research towards its risk measurement and prediction is deficient.This paper introduces the basic principle of VaR method, and takes the daily logarithmic return rate data of Chinese stock index futures IFO as its research objects. First, we take some basic statistical analysis about this index return series. The result indicates that the return series are Leptokurtic and fat tailed, with the effect of high grade ARCH. To overcome the heteroscedasticity, we can use GARCH models.The paper introduces and uses the GARCH models including GARCH-N, GARCH-t, GARCH-GED, TARCH and EGARCH model to measure the risk of Chinese stock index futures, namely its VaR. Then we discuss Chinese stock index futures’leverage effect and its risk characteristics according to the GARCH models’ coefficients. At the same time, the article introduces the mixture density network, a relatively new model of neural network. As a branch of neural network, MDN model fits the time series data well which leads to its good predictive ability.Finally, we take accuracy test, Kupiec failure frequency test, towards the VaRs calculated by GARCH models and MDN model and find that GARCH-GED, EGARCH model and MDN model can relatively accurately predict the VaR values.