Risk Analysis And Empirical Research Of Portfolio Based On Copula

In the financial sector, risk is everywhere. For investors, though they cannot eliminate risk, but can reduce it by spreading investments, and control the risk within their affordable range under the returns in certain circumstances. If the estimation for risk is too conservative, may make the investors miss good investment opportunities; if investors underestimate the investment risk, may bring them greater losses. Therefore, it is important to accurately measure risk. The key to accurately measure portfolio risk is dependent on the analysis and characterization of the relationship between the investment portfolio. Before the Copula theory introduced in the financial sector, the analysis of the correlation between financial assets always was based on the premise of the linear hypothesis, and researchers use the linear correlation coefficient to capture the dependency relationships between financial assets. In fact, the relationship between financial assets is non-linear, so the risk value estimated with linear relationship is not accurate. Under the guidance of the risk of inaccurate values, investment decisions of investors may not achieve the purpose of reducing the risk. To address this problem, based on the perspective of Copula theory, jointing the GARCH model, which can fully characterize ROE procedures phenomena, such as spike, thick tail and volatility, this paper researches the risk value of the portfolio empirically.Firstly, this paper reviews the related research at home and abroad systemly, starting from theory to introducte risk measurement tool in detail. Such as, the estimation methods of Va R, the types of GARCH model and the distribution of residuals, the species and characteristics of the different Copulas model capturing the dependence between variables. In particular, the dependence structure of the tail is introduced in more detail. This paper selects the closing price of Tencent and Vanke from June 28, 2005 to March 18, 2015 as the research object, and analyzes the sample data by using Eviews7.0 and Matlab2009 b.The main conclusions as follows: First, through using the GARCH model to portray peak, thick tail and volatility clustering of Tencent and Vanke shares daily return, we can find that two stock returns have leverage. Tencent stock negative returns impact the volatility greater than the positive. The leverage of Vanke with Tencent is just the opposite. Second, through Copula model to analyze the dependence patterns of the two stocks, find that two stocks with symmetrical tail dependence, and the rank correlation is low. The result shows that it is feasible to build the portfolio by these two stocks. Third, on the premise of the portfolio feasible, this paper uses Monte Carlo simulation method to estimate the risk of the portfolio. And by the test we find that it is most effective to estimate Va R by t-Copula-EGARCH-t model. Fourth, base on the effective model, this paper tries to determine the investment proportion of Tencent and Vanke two stocks. Through the empirical research finds that the optimal investment proportion of the two stocks is 0.3: 0.7.