StockAgent is an AI-driven quantitative analysis platform designed for the A-share market, integrating Large Language Models (LLMs), multi-factor stock selection, quantitative backtesting, and news aggregation. It aims to provide investors with intelligent market analysis and strategy validation tools, automating report generation and offering deep insights through a combination of AI and traditional financial metrics.

How It Works

The platform employs a distributed microservice architecture, featuring Web, Data Sync, Inference, and Backtest nodes communicating via gRPC and HTTP/WebSocket. Data Sync nodes handle market data (Tushare) and news aggregation, while Inference nodes leverage LLMs (GPT-4, DeepSeek, Qwen) for intelligent analysis and report generation. The Backtest node executes quantitative strategies, supporting A-share specific rules. A vector database (Milvus) is utilized for semantic search and event clustering, enhancing AI-driven insights.

Quick Start & Requirements

• Primary Install: Docker Compose is recommended for deployment.
• Prerequisites: Python 3.11+, Node.js 18+, MongoDB 7.0, Redis 7.0, Milvus 2.4+ (optional for AI features), Docker & Docker Compose.
• Configuration: Requires TUSHARE_TOKEN and LLM_API_KEY.
• Links: Deployment details are in AgentServer/deploy/README.md.

Highlighted Details

• AI-Powered Analysis: Integrates multiple LLMs for generating comprehensive stock analysis reports, technical diagnostics, fundamental assessments, and capital flow interpretations.
• Multi-Factor Stock Selection: Features over 17 built-in factors (momentum, value, quality, growth, etc.) with customizable weights and high-performance backtesting capabilities.
• News Aggregation & Clustering: Automatically collects news from 10+ sources, clustering similar events using LLMs for hot topic tracking.
• Automated Reporting: Generates daily morning/afternoon reports and provides historical report querying.
• Vector Database Integration: Milvus enables semantic search for news and event analysis.

Maintenance & Community

The project roadmap indicates core infrastructure and features are implemented, but advanced functionalities like real-time WebSocket data, strategy visualization, mobile adaptation, and live trading integration are still pending. Contribution guidelines are provided, but specific community channels (Discord/Slack) or notable contributors are not detailed.

Licensing & Compatibility

The project is released under the MIT License, which is permissive and generally compatible with commercial use and closed-source linking.

Limitations & Caveats

The Data Sync Node is currently a single instance and not horizontally scalable. Some advanced AI features, such as semantic search and event clustering, are dependent on the optional Milvus vector database. The project appears to be actively developed, with several key features still marked as incomplete on the roadmap, suggesting it may not be production-ready for all use cases, particularly those requiring live trading or advanced real-time capabilities.