TradingAgents/docs/README.md

5.2 KiB

TradingAgents Documentation

Welcome to the TradingAgents documentation. This guide will help you understand, use, and extend the TradingAgents multi-agent financial trading framework.

Overview

TradingAgents is a multi-agent trading framework that mirrors the dynamics of real-world trading firms. By deploying specialized LLM-powered agents - from fundamental analysts, sentiment experts, and technical analysts, to traders and risk management teams - the platform collaboratively evaluates market conditions and informs trading decisions.

Documentation Structure

Getting Started

Architecture

Understand the system design and how components interact:

API Reference

Detailed API documentation for developers:

Module Reference

Documentation for core modules:

  • Backtest Module - Historical strategy replay, slippage/commission models, results analysis, report generation
  • Alerts Module - Multi-channel notifications (Slack, SMS, webhooks)
  • Execution Module - Broker integrations (Alpaca, IBKR, Paper), order management, risk controls
  • Memory Module - Layered memory system, trade history, risk profiles
  • Portfolio Module - Portfolio state, performance metrics, CGT calculator
  • Simulation Module - Scenario runner, strategy comparator, economic conditions
  • Strategy Module - Signal conversion, strategy execution

Guides

Step-by-step tutorials for common tasks:

Testing

Learn about the testing infrastructure:

Development

Contributing and development guidelines:

Key Concepts

Multi-Agent Architecture

TradingAgents decomposes complex trading tasks into specialized roles:

  • Analyst Team: Fundamentals, Sentiment, News, and Technical analysts
  • Researcher Team: Bull and Bear researchers who debate insights
  • Trader Agent: Makes trading decisions based on comprehensive analysis
  • Risk Management: Evaluates portfolio risk and validates strategies
  • Portfolio Manager: Final approval and execution oversight

LangGraph Framework

Built on LangGraph for:

  • State management across agent workflows
  • Tool orchestration for data access
  • Conditional routing based on agent outputs
  • Memory persistence for context retention

Data Vendor Abstraction

Flexible data sourcing through configurable vendors:

  • yfinance: Stock prices and technical indicators
  • Alpha Vantage: Fundamental data and news
  • Google News: Alternative news sources
  • Local: Offline backtesting data

Support

License

TradingAgents is released under the MIT License. See the LICENSE file for details.

Disclaimer

TradingAgents is designed for research and educational purposes. It is not intended as financial, investment, or trading advice. Trading performance may vary based on many factors including model selection, data quality, and market conditions. See Tauric AI Disclaimer for full details.