基于凸二次规划的基金投资风格分析模型

The Fund Investment Style Analysis Model Based on Convex Quadratic Programming

简要分析了基金风格划分常用的基于收益时间序列的回归法(RBSA法)和基于持仓数据的分析法(PBSA法),重点讨论了RBSA方法,根据Sharpe的回归分析模型,解释RBSA方法不仅能合理构造基金业绩评价基准,还能客观展现基金经理的主动选择能力。作相关假设把回归模型转化为一个凸二次规划问题,利用最优化理论转化为求解局部极小值问题,并且介绍了一种降维算法,能有效快速地求解此类问题。

This paper briefly presents Portfolio-Based Style Analysis and Return-Based Style Analysis for fund investment and focuses on the latter. According to the regression model of Sharpe, merits of RBSA method have been explained, which can not only construct reasonable benchmark to evaluate fund performance, but also can show fund managers’ abilities objectively. The regression model has been turned into a convex quadratic programming based on some assumptions, so the relevant optimization theory could be used. A dimension reduction algorithm has been introduced to solve such problems quickly and effectively.