Optimization of a Four-Factor Model for China Using Machine Learning Algorithm and Intelligent Computational Method

Due to the special characteristics of the Chinese A-share market, the traditional three-factor model for US stocks cannot be well applied to the Chinese market. We adopt the China-appropriate four-factor model proposed by Liu, Zhou and Zhu (2019) to systematically apply machine learning to enhance the stock return prediction module in fundamental quantitative investment. The empirical results show that LZZ's four-factor model is indeed better than the FF-3 model. At the same time, the machine learning algorithm is able to effectively identify complex patterns between factor-excess returns complex patterns, and its investment strategy can achieve better investment performance than traditional linear algorithms.