TauricResearch

TradingAgents

AI#LLM#Multi-Agent#Finance#Trading#LangGraph

117

// summary

TradingAgents is a multi-agent based LLM financial trading framework designed to simulate the operational processes of real trading firms. The framework deploys specialized agents, including fundamental, sentiment, news, and technical analysis, to collaboratively evaluate market conditions and formulate trading strategies. The system is built using LangGraph, supports various mainstream LLM providers, and offers an interactive command-line interface as well as a Python development API.

// technical analysis

TradingAgents is a multi-agent financial trading framework built on LangGraph, designed to simulate the operational workflows of real trading firms. By decomposing complex trading tasks into specialized roles such as fundamental, sentiment, news, and technical analysis, and introducing a researcher debate mechanism, the system achieves a deep assessment of market conditions. This architectural design not only improves decision-making robustness but also ensures the rigor of trading decisions through hierarchical control by risk management and portfolio managers. The framework supports various mainstream LLM providers, offering researchers a highly modular and extensible platform for financial AI experimentation.

// key highlights

Adopts a multi-agent collaborative architecture to simulate the complete financial trading process from market analysis to risk control.

Features a multi-dimensional analysis team covering fundamentals, sentiment, news, and technical indicators to provide comprehensive market insights.

Introduces a long-short debate mechanism to balance potential returns with inherent risks through structured discussion.

Supports a wide range of LLM providers, including OpenAI, Google, Anthropic, xAI, OpenRouter, and local Ollama models.

Built on LangGraph, offering high flexibility and modularity for developers to customize configurations and extensions.

Provides an interactive CLI tool to support real-time tracking of agent analysis progress and trading decision processes.

// use cases

Conduct in-depth market analysis and trading decisions through multi-agent collaboration

Evaluate financial assets using fundamental, technical indicators, and sentiment analysis

Implement automated trading strategy execution through risk management and portfolio manager agents

// getting started

Developers can clone the repository, create a Python 3.13 environment using conda, and then run pip install . to install dependencies. After configuring the .env file with the necessary API keys, you can launch the interactive CLI via the tradingagents command or import the TradingAgentsGraph class in your Python code for custom trading strategy development.