Equity Alpha Research Lab

Equity Alpha Research Lab is a synthetic-data public demonstration of a systematic equity research workflow.

It focuses on factor construction, XGBoost sweep research, cost-aware portfolio simulation, walk-forward validation, bias controls, and research hygiene. It is not a production alpha disclosure.

What This Repository Demonstrates

• Synthetic OHLCV data generation
• Momentum, volatility, reversal, and liquidity-style toy factors
• Composite score construction
• Walk-forward train/test slicing
• Embargo-aware validation split
• XGBoost parameter sweep
• Leaderboard and feature importance output
• Out-of-sample predictions
• Prediction-to-portfolio demo
• Cost-aware toy backtest
• Turnover and drawdown reporting
• Reproducible manifest

What Is Intentionally Excluded

• Proprietary factor formulas
• Private stock universe logic
• Private parameters
• Live platform configuration
• Real trading logs
• Paid vendor data

Quick Start

python -m pip install -e ".[dev]"
python -m pytest -q
python examples/run_equity_alpha_demo.py --config configs/sample_config.yaml
python examples/run_xgboost_sweep_demo.py --config configs/xgboost_sweep_config.yaml

Outputs

The XGBoost sweep demo writes:

runs/<run_id>/
  ml_dataset.csv
  xgb_leaderboard.csv
  oos_predictions.csv
  feature_importance.csv
  prediction_portfolio.csv
  prediction_weights.csv
  manifest.json