AI Hedge Fund is a proof-of-concept project that explores using AI to help make trading decisions. It’s meant for learning and research, not real trading. If you want a structured way to test “AI alpha models” and compare results in a controlled backtest, this repo is a starting point.
It doesn’t place real trades. The project includes a hedge-fund-like workflow where you build a fund mandate (strategies, staff, risk, cadence) and then run an interactive app or a non-interactive fund cycle from a saved file.
What sets it apart is the “hedge fund team” framing as a reusable workflow you can iterate on. Instead of hardcoding one strategy, you define the desk via mandates and then run it repeatedly, which makes experimentation easier.
The system asks for required API keys the first time it needs them, and it stores them locally. You can also use exported shell environment variables to override saved keys, which helps when you’re testing different providers or accounts.
+2 more
