基于高维混频D-Vine Copula的投资组合决策  

Portfolio Selection via High-Dimensional D-Vine Copula-MIDAS Method

作　　者：王未卿[1] 李雨晴 汪刘凯 李梦婷 付泽益 WANG Weiqing;LI Yuqing;WANG Liukai;LI Mengting;FU Zeyi(School of Economics&Management,University of Science and Technology Beijing,Beijing 100089;School of Finance,Nanjing University of Finance&Economics,Nanjing 210023)

机构地区：[1]北京科技大学经济管理学院,北京100089 [2]南京财经大学金融学院,南京210023

出　　处：《系统科学与数学》2025年第2期376-397,共22页Journal of Systems Science and Mathematical Sciences

基　　金：国家自然科学基金资助项目(72301025,72072010);中央高校基本业务经费(FRF-TP-22-060A1,FRF-BR-23-08B);北京市社科基金(23GLB022)资助课题。

摘　　要：考虑到大数据背景下混频数据信息对于高维资产投资组合的影响,文章在Vine Copula框架下引入混频信息,将混频数据采样模型MIDAS代入高维D-Vine Copula中,提出基于高维混频D-Vine Copula的CVaR组合投资决策模型,从而同时应对Copula框架下投资组合决策中“维数灾难”与“混频数据信息利用不充分”的挑战.首先,基于高维混频D-Vine Copula模型估计资产多元条件联合分布;其次,根据已估计的联合分布,模拟资产收益变动规律;最后,通过最小化CVaR得到资产组合的最优投资权重,从而建立最小CVaR组合投资决策模型并给出求解方案.文章选取中国新能源市场7支股票进行实证研究,结果表明:文章所提出的模型能够充分揭示与模拟金融资产收益变动规律,获得更低的投资风险.Considering the impact of mixed information for high-dimensional portfolios,this paper introduces the idea of mixed information extraction under the Vine Copula framework,substitutes the mixed data sampling model MIDAS into high-dimensional D-Vine Copula,and proposes a CVaR portfolio selection model based on high-dimensional D-Vine Copula-MIDAS,so as to simultaneously address the challenges of“dimension disaster”and“insufficient mixed information extraction”under the framework of Copula.Firstly,estimate the multivariate conditional joint distribution of assets based on the high-dimensional D-Vine Copula-MIDAS model;Secondly,simulate the dynamic features of assets returns based on the estimated joint distribution.Finally,the optimal investment weight of the assets is obtained by minimizing CVaR,thereby establishing a minimum CVaR portfolio selection model.This paper selects 7 stocks on the Chinese new energy market for empirical studies,and the results show that the CVaR portfolio selection model based on high-dimensional D-Vine Copula-MIDAS can fully reveal and simulate the dynamic features of financial assets returns and obtain lower investment risks.

关 键 词：组合投资 高维混频D-Vine Copula 最小CVaR模型 个股相依性 新能源市场 

分 类 号：F832.51[经济管理—金融学]