基于高维R-vine Copula的金融市场投资组合优化研究

为优化国际金融市场的投资组合,本文以全球具有代表性的七大股票市场重要股票指数作为金融市场的典型代表:首先运用较为灵活的APARCH模型来刻画股票指数收益序列的"典型事实"特征,其次针对投资组合优化模型中变量之间复杂相依关系,采用最大生成树MST (maximum spanning tree,MST)算法选择的R-vine Copula来刻画七个股票市场的相依结构,进而测度R-vine Copula相依结构下组合风险CVaR,最后基于R-vine Copula相依结构条件下建立Mean-CVaR投资组合模型,并实证对比了Mean-VaR,Mean-CVaR和基于R-vine Copula相依结构下的MeanCVaR模型的拟合效果.实证结果表明:考虑资产之间的相依结构能起到优化投资组合的效果,在降低投资组合风险的同时增加了回报率;基于R-vine Copula相依结构下的Mean-CVaR模型投资组合优化效果明显优于Mean-CVaR模型,而Mean-VaR模型较其它两种模型表现相对较差.

Research on financial market portfolio optimization based on high-dimensional R-vine Copula

In order to optimize the investment portfolio of the international financial market, this paper selects the important stock index of the seven major stock markets in the world. First, we use the more flexible APARCH model to describe the "stylized facts" of the stock index return sequence. In the portfolio optimization model, the R-vine Copula, which is selected by the maximum spanning tree (MST) algorithm is used to describe the interdependent structure of the seven stock markets, and then measure portfolio risk under R-vine Copula dependent structure CVaR. Finally, the Mean-CVaR portfolio model was established under the R-vine Copula dependent structure condition. And compare Mean-VaR, Mean-CVaR and Mean-CVaR model based on R-vine Copula dependency structure. The empirical results shows that the interdependence between assets can be used to optimize the portfolio effect, reducing the risk of portfolio risk while increasing the rate of return. The Mean-CVaR model based on the R-vine Copula dependent structure is better than the Mean-CVaR model, while the Mean-VaR model has a relatively poor performance.