Portfolio Risk Analysis Based On Copula-SV-GED Model

70 years since the last century, after the collapse of the Bretton Woods system, some impact on major financial crisis occur frequently, which cause a heavy upheaval of the world's economic environment. The risk of the investment environment people face has been complicated increasingly, which demands a better risk management. As value at risk (VaR) could, at a certain confidence level, make the most possible losses quantitative during a given period, which the traditional risk measurement indicators can not do. In the various ways of risk management, VaR has become a commonly used tool to measure the financial risks. However, the traditional VaR calculation is based on the normal distribution of individual asset returns and the linear relationship of portfolio risky assets returns. The large numbers of empirical studies have shown that a single financial asset has a "spike tail", but also non-linear relationship exists between the assets.In this article, it first introduces the background of VaR models, its calculation methods, strengths and weaknesses in detail and discusses the basic theory and the nature of Copula functions, Copula function parameter estimation methods and evaluation criteria.Secondly, the article build the Copula-SV model which simulate the distribution of return on single asset by the SV model, and describe the correlation between assets by Copula function. Through the empirical analysis, Gumbel-Copula-SV-GED model can achieve better results ,which combine SV-GED model from three forms of SV model with Gumbel Copula from three kinds of Copula functions. This model has some significance for the practical application.Finally, full-text gave a summary and outlook.