The Research On Portfolio Based On Vine Copula Model

With the rapid development of financial liberalization,the connection between countries or regions has become closer,the relationship between financial assets has become increasingly complex.Institutions and investors are facing increasingly investment risks.For institutions,the use of portfolio can effectively spread and control risks.With the increased number of assets in one portfolio,the traditional research methods have “dimension curse”.The Vine copula model can more accurately describe the dependency structure of the portfolio with multiple assets,therefore can reasonably construct portfolio and diminish risks.This article takes 5 industry indexes as examples,within the use of ARMA-GARCH model and skewed student’s T-distribution model to describe the marginal distribution of each index.Adopt three types of Vine Copula model: R-Vine,C-Vine,D-Vine to calculate the relationship between each index,using AIC,BIC,and LogLik methods to determine the fitness of each model.The empirical result shows that R-Vine has the best fitness among them.In addition,predict the VaR of equal weight portfolio by using Monte Carlo Simulation and Sliding Time Window methods.The empirical results indicate all of the models passed the Kupiec LR test,therefore the Vine copula models are effective in predict high dimension portfolio’s VaR.Vine Copula model is a powerful method of the correlation analysis of multiple assets,and has greater realistic significance for institutions to structure portfolio with multiple assets and to manage risks.