Stock Market Portfolio Risk Measurement Based On A Mixed R Vine Copula-SV Model

Risk measurement of portfolio is one of the main contents of financial risk management research and the hot topics that investors continuously concern.In recent years,with the acceleration of economic globalization,the risks facing financial markets have become increasingly complex and diversified.The interdependence structure between financial assets presents characteristics such as non-linearity,asymmetry,and tail correlation.The linear correlation analysis of traditional portfolio models is no longer suitable for describing relevant information about financial risks.Therefore,a corresponding model is constructed to accurately characterize interdependence of underlying assets and measure risk of portfolio are great significance.The risk of the stock market portfolio is measured by constructing a mixed R Vine Copula-SV model.Its main content is divided into two parts: In the first part model construction and parameter is estimated.First,a SV-Skt model is constructed to analyze edge assets and its parameters estimated with the EIS algorithm.The SV-Skt model is used to analyze the characteristics of each edge asset and predict the volatility returns of each asset.Then,the SV-Skt model is used to obtain the innovation and a probability integral transformation performed under the Skt distribution to obtain a sequence on(0,1)of uniform distribution,which will pave the way for a mixed R Vine Copula model built in the next step.The dependence of the underlying asset isdescribed in the portfolio.That is going to establish a mixed R Vine Copula model by using the uniform subsequence as an estimation sample.The optimal pair Copula of each node is determined with the AIC criterion and the two-stage maximum likelihood is employed to estimate its parameters and four types of R Vine Copula models built simultaneously.In the second part is an empirical study.Take the CSI six industry indexes as an example for empirical analysis,and carrying out investment portfolios under equal weights and the Mean-ES model constraints.Monte Carlo simulation of rolling time window is used to calculate in the future holding periods.The Va R and ES of the portfolio are forecasted.The forecast results of the mixed R Vine Copula in the long position and the short position are given,which shows the limitations of Va R in measuring extreme risks and also proves the superiority of ES.Backtesting the prediction performance of each model shows that the prediction performance of mixed R Vine Copula is optimal under equal weights or under the constraints of Mean-ES model.It also further proves that the mixed R Vine Copula-SV model is effective,and it also provides a new theoretical basis for subsequent researchers to predict the risk of portfolios.