Research On Market Risk Measure For Portfolio Composed Of Stocks And Portfolio Selection

Risk management and portfolio selection has always been the market focus of attention.Risk measurement is the core issue of risk management and portfolio selection,risk measurement accuracy will not only affect the effectiveness of risk management,but also affect the performance of the portfolio.Therefore,measure the market risk of the portfolio accurately is important.Volatility modeling and correlation modeling is two key aspects of the market risk measures.Currently,GARCH model is the most commonly used volatility models.Although GARCH class model can describe clustering and fat tail of volatility,but they can not capture the multifractal characteristics of financial assets.Markov Switching Multifractal Model(MSM)can simultaneous characterize the features described above,this article will use MSM model to measure volatility,so to improve the accuracy of risk measurement.Taking into account the extreme financial market events have occurred,further combine MSM Model and Extreme Value Theory(EVT).In addition,the correlation between the financial markets will affect the portfolio's market risk.The correlation is generally non-linear and non-symmetrical between assets,and changeable.So this article adopt the Copula model to model correlation structure between assets.To avoid set exogenous variables of evolution equations subjectively,this article use Generalized Autoregressive Score models(GAS)as time-varying evolution equation of Copula.Then use correlation model to measure portfolio risk and study the portfolio selection problem.First,the paper introduces the method of market risk measurement and Copula function.Secondly,introduces the process of building correlation model and select CSI 300 Index and the Hang Seng Index daily data as samples.Subsequently,the paper selects daily data of CSI300 Index and the Hang Seng Index as samples to construct correlation models.When constructing correlation model,we consider three aspects:the marginal model,evolution equation of the time varying parameter for Copula and Copula function.AS for the marginal model,we select GJR-GARCH,GJR-GARCH-EVT,MSM,MSM-EVT model to describe the distribution of financial asset.AS for the evolution equation of the time varying parameter for Copula,we select the GAS model and ARMA(1,10)process of Patton,then combine with Rotated Gumbel Copula and t-Copula model to construct sixteen correlation models.Base on these correlation models,we use VaR and ES to measure the market risk of the portfolio,and through backtesting test to compare these different models.Finally,we select the best risk measurement models to study portfolio selection problem,examine whether a good risk measurement model can achieve better performance.The empirical results show that the marginal model,evolution equation of the time varying parameter for Copula and the choose of Copula will affect the accuracy of risk measurement.Among the 16 correlation model,MSM-EVT-R-GAS model can accurately measure the risk,which the marginal distribution is MSM-EVT,time varying evolution equation of Copula is the GAS model,Copula function is Rotated Gumbel Copula.Portfolio based on the MSM-EVT-R-GAS model perform better.Assess the risk accurately help investors build portfolio which can attain higher profits.