Abstract: This paper examines the selection of effective factors out of various dimensions of variables that characterize the common stocks of a market, which are then put together to form an additional rationality of maximizing the integrated preference strength(IPS)of effective factors for investment portfolios; in a subsequent manner, an IPS-Mean-CVaR model for multiple-factor portfolio selection is constructed and tested. With 10-year-data analysis of Husheng 300 stocks from 2006 to 2015, here we report three conclusions: (1)the IPS portfolio, constructed with the integrated strength of preference of effective factors outperforms that of every single effective factors; (2)the entropy-TOPSIS technique demonstrates better utility of factors integration than the commonly applied factors scoring technique; and (3)an IPS-Mean-CVaR model outperforms the classical Mean-CVaR model, and expands the capability of data analysis and its applicability for portfolio optimization.

Key words: Multi-factor portfolio selection, Entropy-TOPSIS, Mean-CVaR, Strength of preference

摘要： 本文从股票多维特征因子中选择有效因子,融合形成最大化有效因子综合偏好强度(IPS)的附加理性,构建并验证IPS-均值-CVaR投资组合优化模型。基于沪深300股2006~2015年数据分析显示:(1)有效因子IPS投资组合优越于单因子投资组合;(2)IPS方法相较于因子打分法,具有更优的多维数据整合功效;(3)IPS-均值-CVaR投资组合优化模型相对于均值-CVaR模型具有更优越的资产选择能力,也拓展了投资组合模型的多维数据处理能力和适用性。

关键词: 多因子投资组合, 熵权TOPSIS, 均值-CVaR, 偏好强度