Risk Measurement And Optimal Steategy Selection Of Portfolio Based On SV Model And COPULA

Porfolio theory is one of the important research contents in Finance. It aims to attain the portfolios of maximum of the investment's return with the given the risk of portfolio or of the minimum of investment's risk with the given investment's return. In order to break through the limitations of risk measurement and the normal distribution hypothesis in the traditional Markowitz's mean-variance model, we must apply new risk measurement and find out a more suitable jiont distribution. It is a crucial problem for calculating portfolio risk and getting the optimal investment strategy.VaR and CVaR are put forth recently as new risk measurement, especially CVaR. It becomes a latest research content in financial management; Copula function supplies a convenient way to getting jiont distribution of portfolio. This method overcomes many disadvantages of mutinormal distribution hypothesis.The problem of portfolio optimization based on Mean-CVaR is the main content of the paper. We apply Copula function, SV model, CVaR and Monte Carlo simulation analysis, and then solve the non-normal, nonlinear related problems of portfolio. It supplies a new method of the optimal investment strategy and the measurement of portfolio risk. The object of study is the portfolio of Shanghai Composite Index, Shenzhen Component Index and Hang Seng Index. Through the comparative studies of GARCH model and SV model, we find that SV-t model performs better on depicting the distribution of the return of risky asset. Then we emphases on the Copula selection through the goodness test of some Copula functions. We find that t-Copula is more suitable on depicting the dependence structure of portfolio. Finally, SV-t model and Copula function are used on the measurment of portfolio risk and the problem of optimal portfolio strategy selection based on mean-CVaR. We get the joint distribution and the result of portfolio strategy selection which conforms to reality better. Empirical results show that t-Copula-SV-t model is superior to the traditional model in portfolio risk measurement both and portfolio strategy selection.