TradingAgents
A multi-agent LLM framework for financial trading that simulates real-world trading firms with specialized AI agents for market analysis and decision-making.
What is TradingAgents?
TradingAgents is a multi-agent LLM framework for financial trading that simulates the collaborative decision-making process of a real-world trading firm. It uses specialized AI agents—including analysts, researchers, traders, and risk managers—to evaluate market conditions, debate strategies, and execute simulated trades. The framework is designed for research and experimentation in algorithmic trading and market analysis.
Researchers, quantitative developers, and financial AI enthusiasts interested in exploring multi-agent systems for algorithmic trading, market simulation, and LLM applications in finance.
Developers choose TradingAgents for its realistic simulation of trading firm dynamics, flexible multi-LLM provider support, and modular agent architecture built on LangGraph. It provides a comprehensive, open-source platform for experimenting with collaborative AI-driven trading strategies without relying on proprietary systems.
Overview
TradingAgents: Multi-Agents LLM Financial Trading Framework
Use Cases
Best For
- Researching multi-agent LLM systems in financial markets
- Simulating collaborative trading firm decision-making processes
- Backtesting AI-driven trading strategies with historical data
- Experimenting with different LLM providers (OpenAI, Gemini, Claude, etc.) for trading analysis
- Building modular, scalable frameworks for algorithmic trading research
- Studying risk management and portfolio optimization with AI agents
Not Ideal For
- Production systems requiring direct broker integration and real-time trade execution
- Teams with tight budgets unable to afford multiple LLM API calls per analysis
- Projects needing a simple, out-of-the-box trading bot with minimal configuration
- Applications focused solely on automated execution without collaborative agent debate
Pros & Cons
Pros
Integrates with OpenAI, Google, Anthropic, xAI, OpenRouter, and local models via Ollama, offering flexibility in model choice for cost and performance, as shown in the configurable Python API.
Deploys specialized agents—analysts, researchers, traders, risk managers—that engage in structured debates, mirroring professional collaboration, detailed in the framework architecture diagrams.
Includes dedicated agents for evaluating portfolio risk, market volatility, and liquidity before trade approval, ensuring robust decision-making, as highlighted in the risk management section.
Supports configurable date ranges and simulated order execution for strategy testing on past data, a key feature mentioned for research purposes.
Cons
Each analysis involves multiple LLM agent calls, leading to significant API expenses, especially with premium models like GPT-5.x, and the README notes dependency on external providers without built-in cost controls.
Designed solely for simulation and backtesting, with no support for connecting to live exchanges or brokers, limiting practical use to research rather than production trading.
Requires managing multiple API keys, virtual environments, and LangGraph architecture, which can be overwhelming, as evidenced by the detailed installation and Docker setup steps.
Frequently Asked Questions
Related Projects
HuggingFace Transformers🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
langchainThe agent engineering platform.
vllmA high-throughput and memory-efficient inference and serving engine for LLMs
unslothUnsloth is a local UI for training and running Gemma 4, Qwen3.6, DeepSeek, Kimi, GLM and other models.
Found a gem we're missing?
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.