Building and Submitting A-Share Intraday Factors for a Quant Challenge

Summary

This guide describes participation in a Chinese A-share quantitative challenge with factor discovery and end-to-end modeling tracks. Both target future 30-minute VWAP returns using minute bars and order book snapshots. For the factor track, submissions return a factor value for each stock and sampling time; larger values indicate higher predicted returns. The guide emphasizes point-in-time data use, valid sampling times, controlling missing and invalid values, and avoiding future information. It also outlines rolling history, data-source access, SQL and Python templates, and platform evaluation using information coefficients, long-short Sharpe, stress stability, and turnover.

Submission instructions cover a callable entry point, reproducible model artifacts, runtime and environment limits, and separate public validation and private out-of-sample evaluation. The document is a competition operations guide, not evidence that any particular factor or model works. It supplies no factor performance results; its data ranges, deadlines, resource limits, and evaluation rules are specific to this event and may change.

Key ideas

• The challenge predicts future 30-minute VWAP returns from A-share minute data and order book snapshots.
• Factor outputs must align to valid sampling times and stocks, contain finite values, and avoid look-ahead data.
• Rolling calculations may use earlier history, while returned predictions must stay within the evaluation interval.
• Evaluation considers information coefficients, long-short performance, stress stability, and turnover.
• Public validation results and private out-of-sample evaluation serve different purposes.

Tags

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