A-share multi-agent investment research framework designed to deeply specialize the original TradingAgents for the Chinese stock market. It provides a comprehensive system for automated investment research, analysis, and decision-making, targeting engineers, researchers, and power users interested in quantitative A-share analysis. The framework aims to deliver actionable insights by integrating A-share specific data, rules, and analytical roles.

How It Works

This project is a fork of TauricResearch/TradingAgents, re-engineered for A-shares. It employs a multi-agent architecture featuring seven specialized analysts (Market, Social, News, Fundamentals, Policy, Hot Money Tracker, Lockup Watcher) that feed into a bull vs. bear debate, a risk assessment debate, and a final portfolio management decision. A dual LLM design is utilized: quick_think_llm for agents performing specific tasks and deep_think_llm for managers synthesizing global information for strategic decisions. The system incorporates A-share specific trading constraints like T+1 settlement, price limits, and minimum lot sizes.

Quick Start & Requirements

• Installation: Clone the repository, navigate to the directory, and run pip install -e ..
• Prerequisites: Python version 3.10 or higher. Requires API keys for LLM providers such as MiniMax (recommended), DeepSeek, Zhipu GLM, Qwen, OpenAI, Anthropic, or Kimi. No external API keys or subscriptions are needed for data sources.
• Running: Can be executed via a Python script (example provided), a command-line interface (tradingagents), or a built-in Streamlit Web UI (tradingagents-web).
• Documentation: Links to LLM provider application sites are included in the README for API key acquisition.

Highlighted Details

• A-share Data Sources: Integrates free, direct-connect data from mootdx, Tencent Finance, Eastmoney, Sina Finance, 10jqka, Cailianshe, and Baidu Stock, avoiding paid services like Tushare or overseas APIs.
• Specialized Analysts: Extends the original framework with three A-share specific analyst roles: Policy Analyst, Hot Money Tracker, and Lockup Watcher, crucial for understanding the A-share market dynamics.
• A-share Trading Rules: Fully incorporates A-share trading constraints, including T+1 settlement, daily price limits, minimum transaction lot sizes, and trading hours.
• Dual LLM Architecture: Leverages distinct LLMs for different stages of the analysis pipeline, optimizing for speed and depth of reasoning.
• Output: Generates reports in Chinese, while internal debates are conducted in English to maintain reasoning quality. The A-share market's benchmark is the CSI 300 index.

Maintenance & Community

The project encourages community engagement through GitHub Issues for feature requests. While it acknowledges the upstream project, specific details on active maintainers, sponsorships, or dedicated community channels (like Discord/Slack) are not prominently featured in the README.

Licensing & Compatibility

The project is licensed under the Apache License 2.0, inheriting the license from its upstream fork. This license permits commercial use and integration into closed-source projects, subject to the terms of the Apache 2.0 license.

Limitations & Caveats

This project is explicitly stated to be for learning, research, and technical demonstration purposes only and does not constitute investment advice. All analysis reports and trading signals are AI-generated and may contain errors or biases. The authors disclaim responsibility for any investment losses incurred from using this tool.