A Study On Risk Measurement Of Stock Market Based On Realized EGARCH Model

The Internet Plus has brought out a series of new financial models and financial products.Coupled with the constant changes in the political and economic environment at home and abroad,the prevention of financial risks has gradually become the focus of the government and the business community.With the development of new technologies such as cloud computing and big data,collecting and analyzing high-frequency financial data have become increasingly easier.Volatility prediction and risk measurement based on realized measures have gradually become a frontier problem.However,how to measure the risk of China’s stock market based on the R-EGARCH model still deserves deep research.In this paper,the residual distribution of Realized EGARCH model has been expanded.The non-parametric and semi-parametric methods have been used to estimate the conditional variance.There are also comparisons about predictions of CSI 300 based on R-EGARCH models under different distribution assumptions and parametric estimation methods.This article focuses on R-EGARCH models,and the representative index of the stock market-CSI 300 is selected as the main empirical analysis sample.First of all,it studies the parametric estimation of the R-EGARCH model under the assumptions of Student-t distribution and GED distribution,which is applied to forecast the volatility and VaR of CSI 300 index,and the effect of the binary realized measurement model is discussed as well.Secondly,the non-parametric and semi-parametric estimation of R-EGARCH model have been proposed,according to which volatility and VaR prediction of CSI 300 are carried out.Finally,the single-variable model is extended to a multi-variate DCC-REGARCH model,based on which the correlation between CSI 300 and the Internet financial index is studied and the VaR measure of this portfolio is obtained.Based on the R-EGARCH model,this paper carries out volatility forecasting and VaR risk measurement on CSI 300 Index.The empirical analysis results show that the thick-tailed distribution assumptions can improve the accuracy of R-EGARCH model,and the R-EGARCH model based on Student-t distribution performs well.As for the effect of volatility forecasting,the accuracy of R-EGARCH model is significantly higher than that of the non-parametric and semi-parametric methods.Besides,compared to the R-EGARCH model only relying on RK,the R-EGARCH model based on two measures,RK and DR,is efficient in both volatility and VaR predictions.The prediction effects of various kinds of models under different confidence levels are quite different,and R-EGARCH model tend to achieve the least number of failures.What’s more,applicable to analyze the asset correlation,the multi-DCC-REGARCH model can effectively measure the risk of the portfolio,CSI300 and the Internet financial index.