Portfolio Selection Based On Multivariate Copula Models

With the continuous development of global financial integration,the correlations among financial assets have become increasingly close and complex which can affect the performance of portfolio selection.Therefore,it is crucial for investors to depict the correlations among financial assets effectively.Considering that the characteristic of dependence among financial assets is nonlinear and time-varying,we establish a time-varying multivariate Copula model to investigate the time-varying correlations among financial assets.It has certain significance for investors to construct the portfolio selection model.At the same time,taking the different risk preferences of investors into consideration,the risk aversion parameter is introduced into the portfolio selection model,which can improve the practicability of portfolio selection model.Specifically,the research mainly contains the following two aspects:(1)In order to overcome the shortcomings of the normality assumption and linear correlation assumption of the traditional portfolio selection model,we propose a D-Vine Copula-CAViaR method to estimate the conditional joint distribution of multiple financial assets.At the same time,the portfolio selection model based on generalized Omega ratio is constructed,and the solution scheme is given according to the simulated annealing algorithm.The portfolio selection of multiple financial assets is realized.Firstly,the marginal distributions of individual financial assets are simulated by CAViaR model.Then,the D-Vine Copula method is used to estimate the joint distribution of multiple financial assets,and according to the estimated joint distribution of multiple conditions,the returns of financial assets are simulated.Next,considering the different risk preferences of investors,the risk aversion parameters are introduced into the portfolio selection model.Finally,an empirical study of five international financial assets is carried out.The results show that better investment performance can be obtained by using D-Vine Copula-CAViaR model to make portfolio selection.(2)In order to depict the time-varying dependent structure effectively among financial assets and investigate its influencing factors,we propose a time-varying multivariate Copula-MIDAS-GARCH model.Firstly,the influence of low-frequency variables and high-frequency variables on the time-varying dependence structure of financial assets is studied by using the proposed time-varying multivariate Copula-MIDAS-GARCH model.Then,a VaR-based multivariate portfolio selection model is constructed to compare the portfolio performance among different models.Finally,the model is applied to the time-varying dependence structure characterization and portfolio selection of three financial assets in the Chinese capital market.The empirical results show that the time-varying multivariate dependence structure can predict the financial crisis to a certain extent,and the proposed model can also obtain better portfolio performance.In this dissertation,a portfolio selection model based on multivariate Copula is constructed,which can effectively solve the problem of nonlinear time-varying correlation structure characterization and portfolio optimization among multiple financial assets.The proposed model can enrich the research contents of portfolio selection simultaneously.The empirical results show that this model can also provide decision-making reference for regulators in the financial industry and provide investors with advice on the rational allocation of assets.