AI Hedge Fund is an open-source multi-agent investing experiment with more than 64,000 stars on GitHub that simulates a full hedge fund team — legendary-investor personas included — debating your portfolio.
The project’s premise is delightful and instructive: instead of one model answering a stock question, a team of AI agents role-plays famous investment styles — value, growth, contrarian and quantitative perspectives modeled on well-known investors — alongside fundamental, technical, sentiment and risk-management specialists. Each agent analyzes the same tickers from its own philosophy, they surface their reasoning, and a portfolio manager agent weighs the arguments into final decisions, which can be backtested against historical data. It is explicitly an educational proof of concept rather than financial advice, and it has become one of the most-starred demonstrations of multi-agent design, trending 82 times in the archive. The code is MIT-licensed and actively maintained, with pushes in October 2026.
Part of the project’s appeal is how legible it makes multi-agent design: each persona’s prompts, data needs and decision style are explicit, so readers can see exactly why the team reaches a conclusion and where agents disagree. That transparency has made it a favorite reference for developers prototyping agent teams for research, trading simulations and other adversarial, judgment-heavy domains.
AI Hedge Fund is free and open source under the MIT license. You supply your own LLM API keys and any market-data access the agents use.
Developers learning multi-agent systems and investors curious how LLM teams reason about markets. Treat it as an educational sandbox for agent design — not as a source of financial advice.