Study of Dynamic Multifactor Model Application In China A-Shares

Abstract

This article analyzes the factor investment landscape in China A-shares and explores a feasible solution to construct an adaptive multifactor model for stock selection aiming at stable outperformance over the benchmark CSI 300 Index. A diversified factor database, with more than 60 factors across five factor groups, is constructed for factor behavioral study and model preparation. After analyzing and extracting common time-series and cross-sectional factor predictive power characteristics, a dynamic model with monthly factor selection and tilting based on factor predictive power momentum, persistency, and crowdedness measures is proposed and backtested. From January 2013–March 2020, the proposed model has an information ratio of 1.1152 net of transaction costs—a strong outperformance versus the static models and the simple dynamic model using solely factor momentum. This research offers directional insights into multifactor model applications in the A-shares market.