An Empirical Research On VaR's Trimodal Distribution Model Of The Return Ratio Of Shanghai Stock Market

Rapid globalization has made firms recognize the growing importance of risk management .This has led to the development of various methods and tools to measure the risks. VaR(Value at Risk) originated by J.P.Morgan is the popular technique used to measure the market risk. It is defined as the maximum expected loss over a given horizon period at a given level of confidence. Nevertheless, the VaR approach has been subjected to several criticisms. The most significant one is the normal distribution assumption which makes the estimated VaR inaccurate because of the presence of fat tails. Other studies use T distribution that can incorporate fat tails to replace the normal distribution. Those studies reveal that their VaR estimates are more accurate than those calculated under the normal distribution assumption. Nevertheless, one drawback of previous VaR measures is the focus of the estimation on central observations rather than the observations in the tails which come from the extreme events.We concentrate on analyzing the return ratio of Shanghai Stock Market and propose a trimodal distribution of returns which combines normal distribution and stochastic jumps. This model gives importance to the fat tails that are described adequately under the normal distribution and T distribution models. To test whether the trimodal exists, we perform the joint test and the individual tests on the positive and negative means; and to test the performance of our model, this model is compared with the normal distribution and T distribution through Backtesting. The studies real that the trimodal model is exist and its performance is better than the normal distribution and T distribution models.This paper is organized as follows.In section 1, we consider the background and significance of the study. Section 2 is concerned with the concept and statistics tools of VaR measuring. In section 3, we introduce the main measuring methods and their merit and drawbacks of VaR. The proposed trimodal distribution and the empirical results are presented in Section 4. Finally, Section 5 gathers some concluding remarks.