Building and Testing a Factor-Based Chinese Stock Selection Strategy

Summary

This Chinese-language post introduces a custom-coded stock selection strategy on the BigQuant platform. It says the strategy combines five selection rules, mainly seeking stocks that have fallen sharply and begun to stabilize, while also attempting to identify potential upward moves. The post contrasts rule-based strategies, whose selection logic can be inspected and adjusted, with platform-generated AI strategies, which are presented as easier to start with but harder to interpret and tune.

The suggested workflow is to define or adapt factors, combine them into selection rules, backtest each rule, inspect daily holdings and returns, and revise the logic after reviewing both winning and losing positions. It recommends changing the backtest period to check whether results persist. The author reports annualized and cumulative returns for a shared strategy, and says it holds positions for two days while concentrating heavily in one stock. However, the supplied text does not include the strategy source or enough backtest detail to assess costs, survivorship or look-ahead bias, or out-of-sample robustness. The reported returns should therefore be treated as unverified claims.

Key ideas

• The strategy combines five stock selection rules centered on oversold stocks that appear to stabilize.
• Custom-coded rules are described as more transparent and easier to adjust than platform-generated AI selections.
• The proposed development process tests individual factors and rules, reviews holdings and returns, and iterates on the logic.
• The post recommends checking performance over different backtest periods, but does not supply enough detail to validate robustness.
• The described implementation holds positions for two days and concentrates heavily in one stock.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.