TradingAgents-CN (Chinese-enhanced TradingAgents)
TradingAgents-CN is a Chinese-localized and feature-enhanced fork of the open-source multi-agent trading framework TradingAgents. It aims to help Chinese-speaking researchers, developers, and learners systematically explore how multi-agent systems and large language models (LLMs) can be applied to stock analysis and strategy experiments. The project emphasizes education, research, safety and compliance, and explicitly states it does not provide live trading instructions.
Key highlights
- Architecture: modernized backend (FastAPI + Uvicorn) and frontend (Vue 3 + Vite + Element Plus) for a SPA experience, replacing the previous Streamlit-based UI.
- Data & markets: built-in support for multiple markets (A-share, Hong Kong, US) via common Chinese data sources and adapters (e.g., Tushare/AkShare/BaoStock). Designed to unify data access for analysts and agents.
- Multi-LLM & provider support: dynamic supplier management to plug in multiple LLM providers, native OpenAI support and integrations for other providers; includes intelligent model selection and persistence of model choices.
- Multi-agent workflows: orchestrates specialized agents (news analyst, technical analyst, fundamental analyst, portfolio manager, etc.) to produce structured multi-perspective analysis and reports.
- Deployment & ops: Docker multi-architecture images (amd64 + arm64), GitHub Actions CI, Redis + MongoDB stack for performance and cache, and one-click Docker Compose deployment options.
- Product features: user/role management, configuration center for model and data sources, SSE/WebSocket real-time notifications, batch analysis, simulated trading environment, customizable report export (Markdown/Word/PDF), and logs/audit for reproducibility.
Who is this for
- Learners and researchers who want a hands-on environment to study multi-agent LLM workflows for financial analysis.
- Developers building AI-assisted research tools and prototypes for equity analysis and strategy backtesting (in simulated environments).
- Community contributors who want to localize docs, test integrations, and improve AI-based analysis modules.
Important notes & limits
- Research & learning orientation: The project is explicitly positioned for education and research; it does not provide financial advice or direct instructions for live trading.
- Licensing: repository uses a hybrid licensing model. Most code is Apache-2.0, while the
app/andfrontend/directories are indicated as proprietary in the repo (commercial use of those parts may require separate licensing). Users should read LICENSE and contact the maintainer for commercial authorization when needed. - Extensibility: designed for easy addition of LLM providers, data adapters, and custom analysis agents; includes contribution guidelines and testing calls for volunteers.
Quick start & resources
- Repo (source): https://github.com/hsliuping/TradingAgents-CN
- Docs & tutorials: repository includes docs, quick-start guides, and links to video tutorials and deployment guides (Docker, local install, green build for Windows).
- Community & contact: Issues on GitHub, project WeChat/QQ groups, and maintainer email for inquiries and test volunteer recruitment.
TradingAgents-CN provides a practical bridge for Chinese-speaking practitioners to experiment with multi-agent LLM systems applied to equity research, while emphasizing reproducibility, modularity, and deployment readiness for research environments.
