This repository contains code and results accompanying an independent empirical study of momentum-based trading strategies evaluated under rigorous cross-validation frameworks.

• Dataset: 25M+ daily U.S. equity observations
• Time period: Multiple decades
• Methodology:
  • Purged Cross-Validation
  • Combinatorially Purged Cross-Validation (CPCV)
  • Probability of Backtest Overfitting (PBO)
• Focus:
  • Out-of-sample robustness
  • Transaction costs
  • Turnover and drawdowns

The accompanying research paper is publicly available as a preprint on SSRN:

An Empirical Evaluation of Cross-Sectional Equity Signals Under Backtest Overfitting Diagnostics
SSRN Preprint (DOI): https://dx.doi.org/10.2139/ssrn.6078546

Please cite the SSRN version if referencing this work.

src/ Core modeling, backtesting, and validation code
figures/ Figures used in the accompanying paper
results/ Generated result files (CSV and parquet outputs)
data/ Instructions for obtaining raw data
legacy/ Archived earlier experimental work

Download the raw dataset from Kaggle following the instructions in: data/README.md

pip install -r requirements.txt

python src/makedataset.py

python src/run_cv.py

This repository is intended for research and educational purposes only. It does not constitute financial or investment advice.