What is TradingAgents?
TradingAgents is an open-source multi-agent financial trading framework that leverages large language models (LLMs) to simulate the collaborative decision-making process of a real-world trading firm. The framework decomposes complex trading tasks into specialized roles, each powered by an LLM agent: fundamental analysts evaluate company financials, sentiment experts gauge market mood from news and social media, technical analysts study price patterns, a trader executes strategies, and a risk management team oversees exposure. These agents engage in dynamic discussions to collaboratively assess market conditions and determine the optimal trading strategy, mimicking the synergy of a professional trading desk.
Key features include modular agent architecture, support for multiple LLM providers (OpenAI, Azure OpenAI, AWS Bedrock, and local models via Ollama), and easy installation via pip or Docker. The framework is designed for research purposes, allowing users to experiment with different LLM backends, temperature settings, and trading periods. Use cases range from academic research in multi-agent systems and financial AI to backtesting trading strategies and exploring AI-driven market analysis. Technical details: agents communicate through structured prompts, and the system handles data ingestion from various sources. Trading performance depends on factors like the chosen LLM, model temperature, data quality, and market conditions. The project is fully open-sourced on GitHub, with documentation and community contributions encouraged. It is not intended as financial advice but as a research tool to advance multi-agent LLM applications in finance.
Use Cases
Key Features
multi agent architecture LLM powered agents fundamental analysis sentiment analysis technical analysis risk management modular design multiple LLM support pip installation Docker supportWho is it for?
financial researchers, AI researchers, quantitative analysts, trading strategy developers, academic researchers, fintech developers
