Quant BuffetRelax, Not Over Thinking

Timing the Factor Zoo via Deep Learning: Evidence from China

Log in to collect

Academic paper

Timing the Factor Zoo via Deep Learning: Evidence from China

AuthorsTian Ma; Cunfei Liao; Fuwei Jiang

Institute
  • Minzu University of China
  • ?School of Economics, Minzu University of China
  • Nanjing University of Science and Technology
  • ?Nanjing University of Science and Technology - School of Economics and Management
  • Central University of Finance and Economics
  • ?Central University of Finance and Economics (CUFE)

Strategy in a nutshell

This strategy implements factor timing in the Chinese stock market using 146 characteristic-based factors. It applies PCA for dimensionality reduction and employs deep learning (feed-forward neural networks) to forecast factor returns, forming conditional portfolios that outperform OLS-based and buy-and-hold strategies, delivering higher returns, lower risk, and superior Sharpe ratios.

Economic rationale

The strategy works because deep learning captures nonlinear interactions among factors, improving predictability and robustness in an emerging market. Timing factors generates higher returns, lower volatility, and economic gains compared to traditional linear or unconditional approaches.

Backtest performance

Annualised return59%
Volatility42%
Sharpe ratio1.41