Machine-Learning Ranking for a Bottom-Reversal A-Share Strategy
Summary
This strategy applies a stock-ranking model to a filtered pool of Chinese A-shares, aiming to find stocks that may rebound from depressed levels and hold up better when the broader market weakens. The target universe is the small and medium-sized board. Its feature set combines two technical indicators with turnover, fund-flow, and price-volume factors. Three technical indicators first define a bottom-reversal candidate pool; the next-day close-to-close return is then used as the model target. The described learner is StockRanker, trained on data from June 2015 through December 2019 and evaluated on a later period from January 2020 through January 2024. Each day, the strategy buys the highest-ranked stock and holds it for two days.
The document states the intended defensive behavior but supplies no actual performance statistics, benchmark comparisons, or drawdown figures. It also leaves feature definitions, portfolio and transaction-cost assumptions, and safeguards against look-ahead bias unspecified. The claimed resilience should therefore be treated as a hypothesis requiring independent validation.
Key ideas
- The model ranks filtered A-shares for a potential bottom reversal.
- Features combine technical indicators, turnover, fund flows, and price-volume information.
- The next-day close-to-close return serves as the prediction target.
- The design uses StockRanker, a historical training period, and a later test period.
- The trading rule buys the top-ranked stock and holds it for two days.
- The document provides no performance statistics or implementation details needed to assess robustness.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.