Data Analysis Of Value-at-Risk Under Adaptive Window Size Selection

In recent years,with the development of economic financial globalization,various countries have been integrating into the world economic tide.Following the financial system blending,the financial-market instability has been raised to a higher level.And then the risk management highlights a more important role.VaR(Value at Risk)method has been adopted by more and more financial institutions since it was proposed.And it has gradually become one of the main methods of risk measurement,so that how to calculate it accurately has been a hot spot in the academic circle.This paper describes the background and development of VaR,and its concept,advantages and disadvantages,three general and one empirical calculation methods Then it summarizes the related research of national and foreign scholars.The normal distribution and t-distribution parameter method,the historical and the time-weighted historical simulation method were used to calculate the VaR under the three empirical windows of 250,500 and 1000,respectively.It was found that the accuracy of the results was relatively low,and the size of window would affect the accuracy of VaR obviously.Accordingly,it proposed a model of combining the traditional methods with the sliding window and the empirical estimate method.This model can optimize the original one from the perspective of window and can find the optimal window to make the failure rate converge to the theoretical value.Then using Shanghai composite index data to calculate the failure rate and LR statistics of VaR under three confidences degrees and window lengths,it verified the validity and accuracy of the new model,which can optimize the original method significantly.Meanwhile,it proved that the effect is more obvious at a higher confidence level.In addition,this text compares the optimal sliding window based on four different methods and two different indices.The results showed that the sliding window method's advantages lied in improving the parameter method.Though the trend of Shanghai Composite Index is line with the trend of Shenzhen Component Index,there're significant differences between their optimal windows.Investors should choose appropriate methods and windows to measure risk in diverse situations.Finally,this paper summarized the conclusion and forecasted the model in further research.