Multi-asset Portfolio Risk Measurement Based On Vine Copula Model

In this paper,"Vine" decomposition method, a multi-joint distribution function into Pair Copula function and the density function of a single variable between two variables of product, constructed Pair Copula-GARCH model obtained multivariate joint distribution function so that it can predict the risk of multi-asset portfolios. Wherein each of the properties Pair copula function can be selected according to the distribution of a binary variable t or normal Copula Copula function of this binary symmetrical or asymmetrical Clayton and Gumbel Copula Copula this function to fit, which makes assets between dependent differences can be accurately captured Pair copula structure to.Portfolio risk in three different sections of the paper stocks daily return data composed of empirical analysis. By GARCH model yields a single estimate the distribution of assets, and then with Pair copula be combined to estimate the parameters of the model are obtained joint distribution portfolio. Finally, the historical data to estimate value method to calculate the VaR predicted a loss portfolio values to predict a period of the risks.Advantages with Vine structural decomposition Copula function is a correlation matrix generated by this process is always positive definite no restrictions on the parameters. Furthermore, it has overcome the direct advantage of a multi-joint Copula function to construct multivariate distributions, there is a correlation between the variables, which led to inaccurate results and complexity of the defects. Multivariate Copula function constructed directly compared with rattan construction methods multivariate joint distribution decomposition does not require the assumption of independence between variables. Therefore, this new approach to building high-dimensional structural model of the relevant opened up a new path.