Research On Return Relevance And Var Measure Of Shenzhen Stock Market Based On Multivarite Mixed Coupla-garch Model

The paper is to research on returns and risk of the stock market. Taken six importantstocks of Shenzhen Stock Market as research objects, the paper constructed multivaritemixed Copula-GARCH model class to analyze the relevance of stock returns and VaRcalculation of stock. The main content is as follows:Firstly, we elaborate the basis and value of this paper and overview the refrences;Secondly, we summary the definition, property and classify of Copula functions whichprovide concept for copula model;Thirdly, we research the method of constructing Copula Model, including marginaldistribution and the method of choosing Copula function, parametric estimation andinspection, then give seveval important relevant measure, including Kendall rankcorrelative coefficent and tailed correlative coefficent of Copula;Fourthly, we do empirical research, choose the closing price of six stocks from2011.6.30to2012.6.30in Shenzhen Stock Market, process data to get the rate of retrunsequences of six stocks, then construct GARCH-t model of six stocks to characterize themarginal distributions based on statistic data, considering that there are fifteen different twoto two relevences, we take SFZ A and WK A as an example to analyze the relevencebetween two stocks to aviod the same, choose four different Archimede Copula function tobuild multivarite Copula-GARCH model on the same marginal distribution and makeparametic estimation and inspection, choose multivarite M-Copula-GARCH model todiscribe the strcture of relevence according toχ2inspection, the model not noly cancharacterize operation rules of two stocks, but also can get the following results: there existsthe strong relevence between SFZ A and WK A, their sequences of rate of retrun havesignificant non-symmetric tailed relevence, Kendall rank relevent coefficient is0.7123, theup tailed relevent coefficient is0.5363and the low tailed relevent coefficient is0.2757.Because the model give joint distribution of two stocks, we can master the operation rulesat any time, the discussing about the relevence of fourteen stock combinations is completelysame to the process of constructing the Copula model between SFZ A and WK A, in orderto save length and consider the need of Chapter5, we only give the results of rank releventcofficient of these stock combinations;Fifthly, we use the method of prediction to calculate the VaR of single stock returnaccording to the method and result of constructing model in Chapter4, then we compute the whole stocks and get one kind of risk measure in Shenzhen Stock Market.