TradingAgents

TradingAgents is an open-source multi-agent LLM framework for financial trading with 112,000+ stars on GitHub, modeling a real trading firm as a team of specialized AI agents.

Overview

TradingAgents, introduced in an accompanying research paper, recreates the structure of a professional trading company inside a multi-agent system. Fundamental analysts, sentiment analysts, news analysts and technical analysts each study the market from their own angle; bullish and bearish researchers debate the evidence; a trader agent synthesizes the discussion into decisions; and risk management and portfolio roles review the result before anything is finalized. The framework is built for research: it lets academics and practitioners test how LLM agents reason about markets, where multi-perspective debate helps, and where it fails. It connects to live market data and major LLM providers, and the repository remains actively maintained, with updates in October 2026. Star counts above 100,000 make it one of the most-watched finance-meets-AI projects on GitHub.

Key Features

• Trading-firm structure – analyst, researcher, trader, risk and portfolio roles mirror a real desk.
• Structured debate – bull and bear researchers argue positions before a decision is made.
• Multiple analyst types – fundamentals, sentiment, news and technical analysis in parallel.
• Market data integration – works with real financial data sources for backtests and simulations.
• Model flexible – run it with different LLM providers and compare agent behavior.
• Research-grade design – built to reproduce and extend the published TradingAgents paper.

💳 Pricing

TradingAgents is free and open source under the Apache License 2.0. You pay only for your own LLM API calls and any market-data subscriptions you connect.

Who Is It For

Researchers, quant-curious developers and finance professionals exploring multi-agent LLM systems for market analysis and trading simulation. It is a research framework, not financial advice.