Empirical Asset Pricing via Machine Learning - Evidence from the Chinese stock market

Abstract: This thesis builds upon existing research on the application of machine learning in asset pricing in the US and European stock markets, by incorporating unique predictive indicators specific to the Chinese stock market, to explore whether machine learning can also be successfully applied in the Chinese stock market. Empirical results show that machine learning models outperform OLS significantly in predicting A-share returns, and this conclusion also applies to different portfolios we have constructed. In the analysis of feature importance, we found that the retail investors' dominating presence in the Chinese stock market makes macroeconomic variables and variables containing direct trading information, such as technical indicators, trading volume, and turnover, more influential. This is in contrast to the US market and reflects the characteristics of the Chinese stock market.