Quantitative Stock Selection Based on Artificial Intelligence

Abstract

Artificial intelligence has emerged as a prominent catalyst for technological advancements in recent years. As a cross-disciplinary domain encompassing computational mathematics, statistics, and informatics, artificial intelligence holds significant potential for applications in financial trading and quantitative analysis. This research paper focuses on the utilization of machine learning techniques to analyze seven major fundamental factors of Chinese listed companies from September 2013 to September 2022. Specifically, five machine learning algorithms, including linear regression, Lasso regression, ridge regression, random forest, and decision tree models, are employed to identify the top 30 stocks that offer the most promising expected returns for an equal-weight holding strategy. The performance of this strategy is then compared with that of the CSI 300 index, a widely recognized benchmark. The findings demonstrate that, within the same time period, the aforementioned algorithms outperform the CSI 300 index in terms of returns and drawdowns. Notably, the ridge regression model exhibits the most favorable performance, boasting an annualized return of 14.45% and a maximum drawdown of 0.95% among all selected models. This study, by employing a diverse range of linear and non-linear machine learning algorithms for modeling purposes, contributes to the advancement of the quantitative investment field and provides valuable theoretical insights for the formulation of novel trading strategies.