A new open-source project, 'AI Hedge Fund,' has been released on GitHub, simulating a team of AI agents modeled after famous investors like Warren Buffett, Cathie Wood, and Nassim Taleb. The system, built by developer virattt, is designed as a proof of concept for educational and research purposes only, explicitly not intended for real trading or investment.
The project employs 17 distinct agents, each embodying a specific investment philosophy. For example, a 'Warren Buffett Agent' seeks wonderful companies at fair prices, while a 'Michael Burry Agent' hunts for deep value contrarian plays. Additional agents handle valuation, sentiment, fundamentals, and technical analysis, with a risk manager and portfolio manager coordinating final decisions.
How to run the simulation
Users can run the AI hedge fund locally via command line or a web application. Installation requires cloning the repository, setting up API keys for LLM providers like OpenAI or Anthropic, and obtaining financial data access. The system supports tickers such as AAPL, MSFT, and NVDA, and can backtest over specified date ranges.
The project is evolving toward a persistent, always-on fund that can be backtested, paper-traded, and optionally run live with pluggable 'alpha models.' However, the current version does not execute any real trades. The creator emphasizes that the software provides no investment advice and assumes no liability for financial losses.
Educational tool with clear disclaimers
The repository includes multiple disclaimers stating the project is for learning only. Users are advised to consult a financial advisor for actual investment decisions. The MIT-licensed code encourages contributions via small, focused pull requests.
While the concept of an AI hedge fund team has attracted attention in developer and finance communities, the project remains a simulation. It offers a unique way to explore how different investment strategies might perform in a controlled environment.