Machine-Learning in the Chinese Factor Zoo

Abstract

We add to the emerging literature on empirical asset pricing in the Chinese stock market by building and analyzing a comprehensive set of factors with 1,160 signals for return prediction. Using various machine learning algorithms, we investigate which signals dominate in the Chinese market, a market characterized by a large proportion of retail investors with speculative motives, state-owned firms, and short-sales restrictions. Contrary to studies for the U.S. market, liquidity and fundamental factors emerge as the most important predictors, while price trend signals are less significant. We find that retail investors' dominating presence positively affects short-term predictability, particularly for small stocks. Another feature that distinguishes the Chinese from the U.S. market is the high predictability of large stocks and state-owned enterprises over longer horizons. Our portfolio analysis shows that this overall increased predictability leads to significantly higher out-of-sample performance than in other markets, which remains economically significant after transaction costs.