A new open-source project simulating 12 legendary investors like Warren Buffett and Peter Lynch as AI agents has exploded on GitHub, amassing over 51,000 stars.
Back
New AI hedge fund with 12 legendary investors gets 51k GitHub stars
An open-source project that transforms the philosophies of 12 iconic investors into a multi-agent AI system for stock analysis has captured the attention of the developer community, gaining over 51,700 stars on GitHub. The project, AI Hedge Fund, allows users to get stock insights from AI personas of Warren Buffett, Charlie Munger, and Peter Lynch, among others, and has been forked more than 9,000 times since its release.
"The core idea is to encode investment philosophies into agents, giving retail investors a 'master model'," said Virat Singh, the independent developer behind the project. The system uses a team of 18 total agents—12 based on famous investors and six specialist agents for tasks like valuation and risk management—to debate and decide on a final trading signal.
The project’s technical architecture uses a combination of popular frameworks, with a front end built on React and TypeScript and a Python and FastAPI backend. It uses LangGraph to orchestrate the multi-agent workflows, allowing each agent to pass information through a shared data dictionary. The system can connect to 13 large language models, including those from OpenAI, Anthropic, and Groq, and can also be run locally with open-source models via Ollama.
For investors, the project offers a novel way to stress-test ideas, not as a single recommendation, but as a debate between conflicting strategies. The inclusion of figures with opposing views, such as value investor Ben Graham and growth-focused Cathie Wood, is a key feature. The final output is a synthesis of these different approaches, managed by a dedicated Portfolio Manager agent.
Architecture and Backtesting
The AI Hedge Fund system is designed with a three-layer architecture. The front end features a React Flow-based visual editor, which allows users to drag and drop agent nodes to build custom investment committees. This visual workflow provides a more intuitive way to design and understand the logic of a trading strategy compared to code-only approaches.
The backend relies on LangGraph to manage the state and flow of information between the agents. All agents share a common AgentState data dictionary, ensuring consistency as a stock is analyzed from multiple perspectives. Data is fed into the system through various APIs, with support for professional financial data sources. A key feature for quantitative analysis is the built-in backtesting module. A user can run a strategy against historical data for tickers like AAPL, MSFT, and NVDA to see performance metrics before committing capital.
The Broader Trend
AI Hedge Fund is part of a growing trend of "agent-ifying" the strategies of well-known investors. Similar projects are emerging, aiming to democratize access to sophisticated investment analysis. However, the developer notes that the project has not been tested with real funds and does not guarantee returns.
The project has sparked discussion about its practical use. One user questioned how to act when the AI "masters" have conflicting opinions. Singh's project addresses this by having a final portfolio manager agent make the call, but the value, as some users noted, may be in hearing the debate itself. While the system can replicate an investment philosophy, it cannot replicate the results. For now, it serves as a powerful educational tool and a framework for building more advanced agent-based financial analysis systems.
This article is for informational purposes only and does not constitute investment advice.
