Skip to content

CHANGELOG.md

Latest commit

 

History

History
37 lines (29 loc) · 1.27 KB

CHANGELOG.md

File metadata and controls

37 lines (29 loc) · 1.27 KB
Raw

Changelog

0.1.0 - Public Research Baseline

This is the first public baseline release of ml-quant-trading.

Highlights

  • End-to-end synthetic factor-to-backtest pipeline.
  • 213 factor dimensions:
    • 204 hand-crafted legacy factors.
    • 9 curated Alpha101-style factors.
  • Mask-aware PyTorch tensor primitives for cross-sectional panels.
  • Limit-up, limit-down, and halt bias correction.
  • MLP and Transformer model baselines.
  • Sign-aware losses: AdjMSE, IC, and RankIC.
  • Cross-sectional Markowitz portfolio optimization.
  • Vectorized backtesting and metrics.
  • Public-data factor IC notebook.
  • Tensor factor benchmark script.
  • CI across Python 3.9, 3.10, and 3.11.
  • Citation metadata through CITATION.cff.

Good First Contributions

  • Run make benchmark and submit a benchmark issue.
  • Try the public-data notebook and report reproducibility issues.
  • Add one public-data case study.
  • Improve factor family documentation.
  • Add tests for neutralization and backtest edge cases.

Notes

The repository is for research and engineering experimentation. It does not provide financial advice or live trading recommendations. Proprietary market data used in the paper cannot be redistributed, so the public reproduction paths focus on synthetic data and public-data examples.