Comparing Rolling Linear Regression and XGBoost Trading Models

Summary

This account describes a BigQuant assignment that applied linear regression and XGBoost within a rolling-training strategy, then compared their backtest returns. The author reports annual returns of 30% for linear regression and 52% for XGBoost, but provides no sample period, benchmark, transaction costs, risk measures, or validation details, so these figures alone do not establish robustness or comparability.

The workflow reused a strategy framework, swapped in each model, added factors, and retrieved backtest results to plot a comparison. The author recommends building a complete framework first, filling in components, and debugging each stage with progress checks. They also argue that factor quality can matter more than model choice for a usable strategy, while noting plans to screen factors and tune hyperparameters. This is an individual learning reflection rather than a controlled study; its claim about factor importance is a personal conclusion, not demonstrated by the reported comparison.

Key ideas

• The author compared linear regression and XGBoost within a rolling-training strategy.
• Reported annual returns were 30% for linear regression and 52% for XGBoost, without supporting risk or validation details.
• The workflow reused a strategy framework, added factors, and plotted results obtained from backtest data.
• The author recommends building and debugging a strategy in stages.
• The author's view that factor quality matters more than model choice is a personal judgment requiring further testing.

Tags

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