The Empirical Analysis Of Univariate Time Series A Research Of The Value At Risk Based On Price Range-ARMA-GARCH-POT Model

In recent years, with the rapid development of information technology, Chinese stock market is also undergoing changes rapidly. However, the financial market of China is an emerging market; the existence of market structure is imperfect and speculative etc. In addition, with the further development of global integration,various policies and news brought great impact on the global financial market, which also have an impact on China’s financial market, and then the stock market will be affected. With the continuous development of China’s stock market, the risks will be more and more. Thus how to measure the Value at Risk and minimize the loss become more and more important. And in today’s world the mainstream method to measure the Value at Risk of the financial assets is the VaR model, a large number of domestic and foreign scholars are studying how to use VaR model to measure the risk, improve the effectiveness of Value at Risk, which has important significance on the financial risk measurement. On the basis of the former studies, the GARCH model, variance-covariance method and extreme value theory will be introduced in the paper ant these three aspects will be combined to improve the VaR Model, aiming at improving the effectiveness of VaR.The study of the paper starts from the background of financial market risk,explains the definition of the Value at Risk and the basic calculation methods in detail,introduces the most widely used combination model to estimate VaR, namely the combination of GARCH model and variance- covariance model. From this model it is quite known that two quantities should be known to assess VaR.They are the standard deviation sequence from the founding of the GARCH model to financial time series and the time sequence probability distribution quantile obtained.Thus the model can be improved from two aspects, using contains the stock closing price, the highest price and the lowest price of price range- GARCH model to improve the effectiveness of the standard deviation; it is usually under normal assumption when using the combination model of variance covariance and GARCH model to estimate VaR.But research shows that time series often has the characteristics of peak thick tail, so based on the assumption of normal distribution for the VaR method will usually underestimate the tail risk, and the extreme value theory is to deal with the time sequence data’s tail directly, and doesn’t need to follow any distribution hypothesis of loss data in advance, thus the quantile calculated by extreme valuetheory is more effective. Combine the two models together,a new model constitutes:Price Range-ARMA-GARCH-POT Model, finally through the empirical analysis, the VaR calculated by Price Range-ARMA-GARCH-POT model, is much closer to significance level than the Value at Risk of failure rate which assessed by the unimproved model. It shows that the new model Price Range-ARMA-GARCH-POT model does improve the validity of VaR, and the improvement of VaR model is feasible.Finally,this dissertation summarized the conclusions from the empirical analysis,and expounds the deficiency of this dissertation, based on the shortcomings, put forward some future research directions.