A Chinese Stock Ranking Workflow with Engineered Factors
Summary
The document presents a Chinese-stock ranking workflow embedded in a report about a Papermill execution error that arose when submitting a strategy for simulated trading. It constructs price, volume, valuation, and technical features, creates a forward-return label with extreme values clipped into quantile bounds, trains a tree-based ranking model on historical data, and applies the model to later observations. The trading logic buys highly ranked stocks, weights holdings by rank, and aims to hold them for several days while replacing lower-ranked positions.
The supplied material does not include the actual error message or a diagnosis, and it does not report strategy performance. Its code and settings describe one implementation, but do not establish that the features are predictive or that the simulation deployment issue is resolved. The date ranges and model settings are examples from this specific strategy, not evidence of general effectiveness.
Key ideas
- The workflow engineers price, volume, valuation, and technical features for stock ranking.
- It labels examples using forward returns and clips extreme outcomes before creating ranking bins.
- A tree-based ranking model is trained on historical Chinese-stock data and used to score later data.
- The portfolio buys top-ranked stocks and replaces lower-ranked holdings as it rebalances.
- The report provides no error trace, performance results, or proof that the deployment problem is fixed.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.