TradingAgents: Multi-Agents LLM Financial Trading Framework

Abstraction: Multi-agent LLM framework simulating specialized trading firm roles

Key points:

• Seven specialized agent roles: Fundamentals Analyst, Sentiment Analyst, News Analyst, Technical Analyst, Researcher, Trader, Risk Manager
• Bull and Bear researcher agents debate market conditions before trader agents act; all use ReAct prompting framework
• Evaluated on AAPL, GOOGL, AMZN over June–November 2024 using historical prices, news, social media, insider transactions
• Outperformed all baselines: e.g. 26.62% cumulative return on AAPL vs. -5.23% buy-and-hold
• Significantly higher Sharpe Ratios and lower maximum drawdown vs. rule-based strategies (MACD, KDJ, SMA)
• Communication uses structured reports and diagrams for efficiency; natural language reserved for debate phases

Connections: Trading Agents · AI Agents · Multi Agent Systems · Large Language Models

Source: https://tradingagents-ai.github.io/