A Chinese-focused multi-agent LLM framework for financial trading analysis. Features A-share/HK/US support, FastAPI+Vue3 architecture, and full workflow from data sync to report generation.
Overview#
TradingAgents-CN is a multi-agent LLM framework designed for financial trading analysis, specifically optimized for the Chinese market. It provides a full-stack solution from data acquisition and intelligent analysis to simulated trading.
Core Features#
- Multi-Agent Collaboration: Simulates an investment team (Market Analyst, Fundamental Analyst, News Analyst, Trader, Risk Manager) for comprehensive analysis.
- Intelligent News Analysis: AI-driven news filtering, relevance analysis, quality scoring, and sentiment recognition.
- Market Support: Deep integration with A-shares, as well as Hong Kong and US stock markets. Utilizes data sources like Tushare, AkShare, and BaoStock.
- LLM Integration: Supports multiple providers including OpenAI, DeepSeek, Tongyi Qianwen, and Gemini, with dynamic supplier management and smart model selection.
Architecture#
The project features a decoupled architecture:
- Backend: FastAPI + Uvicorn for asynchronous high performance.
- Frontend: Vue 3 + Vite + Element Plus for an enterprise-grade UI.
- Data Layer: MongoDB (persistence) + Redis (caching/sessions).
- Communication: RESTful API + WebSocket for real-time notifications.
Deployment Options#
- Standalone (Green): Ready-to-use for Windows users.
- Docker: Recommended for production; supports amd64/arm64 architectures.
- Source Code: For developers requiring customization.
Highlights#
- Comprehensive user permission management and audit logs.
- Visual configuration center for models and data sources.
- Export professional analysis reports in Markdown, Word, or PDF formats.
- Simulated trading system for strategy validation.