This AI runs a virtual trading firm with debating agents
TradingAgents simulates a full investment team with AI analysts, researchers, and a portfolio manager debating trades. 33K stars and growing fast.
What if you could have an entire investment team — analysts, researchers, a risk manager, and a portfolio manager — all powered by AI, debating every trade before it happens? That's exactly what TradingAgents does, and the open-source project just crossed 32,700 GitHub stars.
Built by Tauric Research, the framework turns AI language models into specialized financial agents that work together like a real trading firm.
A Full AI Investment Team
Instead of one AI making all the decisions, TradingAgents splits the work across multiple specialized agents — each with a specific role, just like a real trading desk:
The Analyst Team
• Fundamentals Analyst — Evaluates company financials and calculates what a stock is actually worth
• Sentiment Analyst — Monitors social media and public opinion to gauge market mood
• News Analyst — Watches global news and economic indicators for market-moving events
• Technical Analyst — Studies price charts and patterns using indicators like MACD and RSI (common tools traders use to predict price movements)
The Decision Pipeline
• Bullish & Bearish Researchers — Two researchers who debate each other, arguing the case for and against each trade
• Trader Agent — Takes all the research and makes timing and sizing decisions
• Risk Manager — Checks if the trade is too risky given current portfolio exposure
• Portfolio Manager — Final approval or rejection, then executes on a simulated exchange
Pick Your AI Brain
TradingAgents doesn't lock you into one AI provider. You can power your virtual trading firm with:
- OpenAI (GPT models)
- Google (Gemini)
- Anthropic (Claude)
- xAI (Grok)
- Open-source models via Ollama (run locally on your computer)
You can even mix models — use a powerful model for the portfolio manager's final decisions and a cheaper one for the analysts doing initial research.
Getting Started
The project runs locally and uses a CLI (command-line interface) with an interactive menu. Here's how to set it up:
git clone https://github.com/TauricResearch/TradingAgents.git
cd TradingAgents
pip install -r requirements.txt
python -m cli.mainYou'll need an API key from at least one AI provider and a free Alpha Vantage API key for market data.
Important: This Is a Research Tool, Not Financial Advice
The creators are clear: TradingAgents is a research framework for studying how AI agents can collaborate on financial analysis. It uses simulated exchanges, not real money. It's designed to help you understand market analysis patterns and explore how multi-agent AI systems make decisions — not to replace human financial advisors.
That said, if you're a finance student, a data-curious investor, or just someone fascinated by how AI can model complex decision-making, this is one of the most impressive open-source projects in the space. The built-in debate mechanism (where AI researchers argue opposite sides of each trade) is a genuinely clever approach to reducing AI bias.
Check out the full project on GitHub.
Related Content — Get Started with Easy Claude Code | Free Learning Guides | More AI News
Stay updated on AI news
Simple explanations of the latest AI developments