A-Scope-Research

# A-Scope Research - Financial Intelligence Team Based on MCP (Model Context Protocol), an intelligent analysis system for China's A-share market, providing comprehensive analysis support for investment decisions through collaboration and debate among multiple professional intelligent agents. ## 🌟 System Features ### 🤖 Professional Intelligent Agent Team - **Technical Analyst**: Focuses on technical indicators, chart patterns, and trend analysis - **Fundamental Analyst**: Concerned with financial data, company value, and industry analysis - **Quantitative Analyst**: Employs mathematical models, statistical methods, and risk quantification - **Market Sentiment Analyst**: Analyzes investor sentiment, market psychology, and public opinion - **Risk Manager**: Assesses investment risk and formulates risk control strategies ### 🔄 Intelligent Collaboration Process 1. **Parallel Analysis**: Each intelligent agent independently analyzes target stocks from a professional perspective 2. **Team Debate**: Multi-round discussions, exchanging views and questioning assumptions 3. **Final Decision**: Investment advice based on consensus from the debate 4. **Risk Assessment**: Comprehensive risk analysis and control recommendations ### 🛠️ Technical Architecture - **MCP Protocol Integration**: Real-time access to A-share market data and analysis tools - **Multi-Model Support**: Compatible with mainstream large models such as OpenAI, Zhipu GLM, Tongyi Qianwen, etc. - **Real-time Web Interface**: Intelligent agent dialogue interface similar to WeChat group chat - **Flexible Configuration**: Supports YAML configuration file customization of models and parameters ## 🚀 Quick Start ### 1. Environment Preparation ```bash # Clone the project git clone <repository-url> cd A-Scope-Research # Create a virtual environment (recommended) python -m venv .venv # Activate the virtual environment # Windows .venv\Scripts\activate # Linux/Mac source .venv/bin/activate # Install dependencies pip install -r requirements.txt ``` ### 2. Configuration Settings #### Edit `config.yaml` ```yaml agents: technical_analyst: name: "Technical Analyst" model: "gpt-3.5-turbo" # or other supported models api_key: "your_api_key_here" # replace with your API key base_url: "https://api.openai.com/v1" # or other API addresses temperature: 0.7 max_tokens: 2000 # ... other intelligent agent configurations ``` #### Supported Model Service Providers | Service Provider | base_url | Model Examples | |--------|----------|----------| | OpenAI | `https://api.openai.com/v1` | `gpt-3.5-turbo`, `gpt-4` | | Zhipu GLM | `https://open.bigmodel.cn/api/paas/v4/` | `glm-4`, `glm-3-turbo` | | Tongyi Qianwen | `https://dashscope.aliyuncs.com/compatible-mode/v1` | `qwen-turbo`, `qwen-plus` | | Moonshot | `https://api.moonshot.cn/v1` | `moonshot-v1-8k`, `moonshot-v1-32k` | | DeepSeek | `https://api.deepseek.com/v1` | `deepseek-chat`, `deepseek-coder` | ### 3. Operation Mode #### 🌐 Web Interface Mode (Recommended) ```bash python main.py --mode web ``` Access `http://localhost:8501` to open the Web interface #### 💻 Command Line Mode ```bash # Analyze a specified stock python main.py --mode cli --stock 000001 ``` #### 🎭 Demo Mode ```bash # Interactive demo python main.py --mode demo ``` ## 📊 Web Interface Functionality ### Main Functions - **Real-time Team Status**: Displays the initialization status and tool count of each intelligent agent - **Stock Analysis Input**: Supports 6-digit A-share code input - **Analysis Result Display**: Professional analysis reports from each intelligent agent - **Debate Process Visualization**: Real-time debate interface similar to group chat - **Final Decision Summary**: Investment advice and risk assessment - **Result Export**: Complete analysis report in JSON format ### Interface Features - 🎨 **Intelligent Agent Avatars**: Each intelligent agent has a unique emoji avatar - 🎨 **Color Differentiation**: Different roles use different color themes - ⏰ **Timestamp Display**: Complete analysis timeline - 🔧 **Tool Invocation Display**: Real-time display of MCP tool invocation process - 📱 **Responsive Design**: Adapts to different screen sizes ## 🔧 Project Structure ``` A-Scope-Research/ ├── main.py # Main entry file ├── config.yaml # Intelligent agent configuration file ├── mcp.json # MCP server configuration ├── requirements.txt # Python dependency packages ├── README.md # Project documentation ├── src/ │ ├── __init__.py │ ├── agents/ # Intelligent agent module │ │ ├── __init__.py │ │ ├── base_agent.py # Basic intelligent agent class │ │ └── team_manager.py # Team manager │ ├── prompts/ # Intelligent agent prompts │ │ ├── __init__.py │ │ ├── technical_analyst.py # Technical analyst │ │ ├── fundamental_analyst.py # Fundamental analyst │ │ ├── quantitative_analyst.py # Quantitative analyst │ │ ├── sentiment_analyst.py # Sentiment analyst │ │ └── risk_manager.py # Risk manager │ └── ui/ # User interface │ ├── __init__.py │ └── streamlit_app.py # Streamlit Web application └── .venv/ # Virtual environment (automatically generated) ``` ## 🎯 Usage Scenarios ### Individual Investors - Obtain multi-angle professional analysis opinions - Understand the pros and cons of different investment strategies - Make more rational investment decisions ### Investment Institutions - Assist in the decision-making process of research teams - Provide standardized analysis framework - Reduce analysis bias from a single perspective ### Financial Education - Learn the application of different analysis methods - Understand the complexity of investment decisions - Observe the thought process of professional analysts ## 🔍 Analysis Example ### Input ``` Stock Code: 000001 (Ping An Bank) ``` ### Output Example #### Technical Analyst's View > 📈 From a technical perspective, 000001 is currently in an upward channel, with effective MA20 support and a clear MACD golden cross signal, suggesting buying on dips... #### Fundamental Analyst's View > 📋 Ping An Bank's Q3 financial report shows a stable ROE above 13%, well-controlled non-performing loan ratio, relatively reasonable valuation, and long-term investment value... #### Quantitative Analyst's View > 🔢 Based on historical data backtesting, the risk-adjusted return at the current price level is 1.8, with volatility at the historical median, suggesting standard position allocation... #### Market Sentiment Analyst's View > 😊 The recent banking sector sentiment has warmed up, with significant institutional fund inflows, but attention should be paid to policy expectations and their impact on sentiment... #### Risk Manager's View > 🛡️ The main risk comes from interest rate policy changes and credit cycle fluctuations, suggesting setting an 8% stop-loss level and not exceeding 15% of the portfolio... ## ⚙️ Advanced Configuration ### Custom Intelligent Agents You can customize the behavior and professional strengths of intelligent agents by modifying files in the `src/prompts/` directory. ### MCP Tool Extension Add new MCP servers in `mcp.json` to extend data sources and analysis tools. ### Model Parameter Tuning Adjust the `temperature` and `max_tokens` parameters of each intelligent agent in `config.yaml` to optimize output quality. ## 🐛 Troubleshooting ### Common Issues 1. **Model Connection Failure** - Check if the API key is correct - Verify the base_url format - Confirm the network connection is normal 2. **MCP Tool Connection Failure** - Check the accessibility of the MCP server URL - Verify firewall settings - Confirm reasonable timeout settings 3. **Dependency Package Installation Failure** - Use a domestic mirror source: `pip install -i https://pypi.tuna.tsinghua.edu.cn/simple -r requirements.txt` - Upgrade pip version: `pip install --upgrade pip` ### Debug Mode Enable detailed log output: ```bash # Set environment variable export LOG_LEVEL=DEBUG python main.py --mode cli --stock 000001 ``` ## 🤝 Contribution Guide Welcome to submit issues and pull requests to improve the project! ### Development Environment Setup ```bash # Install development dependencies pip install -r requirements.txt pip install pytest black flake8 mypy # Code formatting black src/ # Code inspection flake8 src/ # Type checking mypy src/ # Run tests pytest ``` ## 📄 License This project adopts the MIT license, see [LICENSE](LICENSE) file for details. ## 🙏 Acknowledgments - [LangChain](https://github.com/langchain-ai/langchain) - Large model application framework - [MCP](https://github.com/modelcontextprotocol) - Model context protocol - [Streamlit](https://streamlit.io/) - Web application framework ## 📞 Contact Information If you have questions or suggestions, please contact us through: - Submit [GitHub Issue](https://github.com/guangxiangdebizi/A-Scope-Research/issues) - Send email to: guangxiangdebizi@gmail.com --- **Disclaimer**: This system is for learning and research purposes only and does not constitute investment advice. Investment carries risks, and decisions should be made with caution.