Risk Management Of Portfolio

The measurement of market risk is a primary concern for regulators and forinternal risk control. After being proposed in 1993, Value-at-Risk(VaR)approachhas become the standard for risk management industry. But VaR has varioustheoretical deficiencies. Conditional VaR(CVaR)is an alternative risk measure tothe quantile which overcomes the theoretical deficiencies of VaR.VaR and CVaR are tools which be widely used in application to financialrisk management and regarded as extreme quantile method. This paper presentsthe theory of extreme value and character of tail of distribution. The applicationmethod of peak over threshold is used to calculate VaR and CVaR to estimatetail-related risk.In tradition portfolio theory, linear correlation is used in describing the depen-dency of di?erent capital, but it is not enough. In statistics, copula is a commontool to construct the multivariate joint distribution and to analyze the multi-variate dependency structure. Based on the character of copula, a multivariatedistribution function which can re?ect the actual distribution and the dependenceof financial asset returns is developed.Finally, on the assumption of investor's exponential utility function, usingthe Mixed Gumble copula, empirical research is done on the performance ofthe portfolio selection in order to research the e?ect of measuring the actualdistribution and dependence on portfolio selection.