Var Analysis For Market Risk Of Shanghai Stock Market Based On Non-Parametric GARCH -EVT Model

Financial market is developing in China now. As an important part of financial markets, the risk of the stock market is more and more important. This paper selects Shanghai Stock Exchange Composite Index as the research sample, trying to get a more accurate and easier method to measure the market risk.For risk measurement, today’s main method is to calculate the value at risk (VaR), which is a measure of the most widely used tool for risk. This method changed the risk which is abstract to be intuitive digitals, and more concise, but the traditional VaR calculation methods require certain prerequisites assumed, in many cases these assumptions are not consistent with the reality, which reduces the reliability of the model. This paper will combine the Non-parametric GARCH model and the Extreme value theory, using the two models’advantages to improve VaR. In this article, using Non-parametric GARCH Model can overcome the clustering effect of financial data and get a better result than other GARCH models, so, it will provide an ideal data for the extreme value. At the same time, this article will use the extreme value model as a theoretical basis, only the tail of the distribution of market data characteristics, avoiding some unreality assumptions. Combining the two models, the result is a dynamic VaR, which is more accurate than other VaRs.Firstly, this paper describes the current developments in the financial markets and financial market risks. Summarizing the risk measure theory, introducing the VaR methods, and noting the advantages and disadvantages of various methods. Secondly, this paper established a non-parametric GARCH model, through theoretical and empirical tests proved its superior to other GARCH family models, especially in the calculation of VaR. Finally, based on the extreme value theory, this paper gets the dynamic VaR, and compared with the static VaR which is just calculated by the extreme value model. As compared with each other, the result of the non-parametric GARCH--EVT model significantly improves the accuracy of the VaR calculation.Main innovation of this paper is to introduce non-parametric GARCH model to measure the risk in the stock market, so that the data applied to the extreme data more accurate. Meanwhile, for the VaR estimates, this paper introduces the extreme value theory, and through the establishment of a series of more objective method to determine the threshold; Finally, the new model that the non-parametric GARCH model combined with the extreme value theory, expands the application of the conventional method, relaxes the traditional assumption and obtains a more stable and accurate estimates.