Thanks for your interest in contributing to AutoForge!
- Report bugs — open an issue with steps to reproduce
- Suggest features — open an issue describing the use case
- Add indicators — extend
prepare.pywith new technical indicators - Improve fill logic — make
backtest.pymore realistic for specific platforms - Add examples — contribute toy strategies that demonstrate different patterns
- Write tests — help improve test coverage
- Fix docs — typos, clarifications, better examples
git clone https://github.com/saikodi/AutoForge.git
cd AutoForge
pip install -e ".[dev]"- Keep it simple. AutoForge is intentionally minimal. Don't add complexity unless it earns its place.
- No proprietary strategies. Example strategies should use well-known, publicly documented concepts (SMA crossover, RSI, etc.).
- Match the existing style. Look at how the current code is structured and follow the same patterns.
- Test your changes. Run existing examples to make sure nothing breaks.
- One thing per PR. Small, focused pull requests are easier to review.
- Python 3.10+
- No type annotations on internal code unless they clarify intent
- Docstrings on public functions
- Keep dependencies minimal (numpy + pandas core)
- Framework-ification — AutoForge is a few files, not a framework
- Heavy dependencies — keep the dep tree minimal
- Proprietary strategy logic or specific parameter values
- Features that only work on one platform (keep it platform-agnostic)
Open an issue. We're happy to help.