XGBoost Ensemble Signals for Intraday Chinese Stock Trading

Summary

The document describes an intraday strategy that combines classification and regression models built with XGBoost to rank stocks by predicted daily price change. Its features include overnight returns and opening-auction behavior, such as staged price moves, trading value, order imbalance, limit-up events, and bid-ask quote measures. Several features are reported to have positive or negative relationships with intraday returns, while the absolute change in a quote measure is said to vary in influence over time.

The strategy buys equal-weighted positions at the open in the top 2% of daily signals and exits at the close. The reported out-of-sample test covers January through October 2019 and includes a stated two-sided transaction cost assumption; it reports a 57.24% win rate, 130.2% annualized return, 4.31 Sharpe ratio, and 18.9% maximum drawdown. These are results from a limited historical test, and the document provides no detail here on data construction, validation design, or robustness across other periods and markets.

Key ideas

• The method combines classification and regression models to rank expected intraday stock returns.
• Opening-auction activity, overnight returns, and order-book measures are among the reported predictive features.
• The strategy buys equal-weighted positions in the highest-ranked 2% at the open and sells at the close.
• The reported out-of-sample results cover only January through October 2019 and depend on the stated transaction cost assumption.

Tags

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