TradingAgents-CN Multi-Agent Trading Framework

Between November 13 and 14, 2025, TradingAgents-CN implemented significant updates focused on data integrity, screening performance, and system reliability. Key improvements include:

• Data Quality: Fixed issues with volume, turnover, and trade date displays.
• Screening Performance: Optimized field types and enabled database-level optimized screening, resulting in a 10x+ performance increase.
• Synchronization Mechanism: Resolved trade_date missing issues, implemented failure rollback mechanisms, and addressed API rate-limiting cascades.
• Deployment Simplification: The application now automatically creates views and indexes upon startup, removing the need for manual script execution.
• Bug Fixes: Resolved issues related to logger imports, MongoDB connections, and circular imports.

TradingAgents-CN integrates multiple financial data sources with a specific focus on the Chinese A-share market to provide agents with comprehensive, accurate, and real-time market information.

As of version 0.1.6, the system uses a hybrid strategy for data retrieval:

• Tushare: Primary source for historical data.
• AKShare: Used for real-time data.
• Unified Interface: The system provides transparent switching between these sources, maintaining backward compatibility so existing API calls do not need to change.

The October 27, 2025 update focused on improving system stability, user experience, and cross-platform compatibility. Key improvements include:

• Compliance Optimization: Refinement of compliance-related terminology and expressions.
• Error Handling: Improved error notification mechanisms.
• Logging System: Fixes and enhancements to the logging infrastructure.
• Architecture Support: Added support for ARM architectures.

These changes were implemented across 12 commits to ensure a more robust and compliant environment for trading agents.

Between November 1st and November 4th, 2025, TradingAgents-CN introduced significant updates focused on cross-platform deployment and data integrity. Key improvements include:

• Windows Portable (Green) Version: A portable package that allows for one-click startup without a formal installation process.
• Windows Installer: A standardized installation flow that supports setting the application to launch automatically on system startup.
• Fundamental Analysis Enhancements: Added total market capitalization data to improve valuation metrics.
• Port Conflict Detection: Automatic detection and cleanup of processes occupying required ports.
• LLM Configuration Fixes: Resolved issues where API Key updates were not taking effect.
• Multi-platform Packaging: Support for Windows, Linux, and macOS packaging.

As of the November 11, 2025 update, the US stock data source architecture and caching system have been overhauled to improve flexibility and performance.

Key Improvements:

• Multi-Source Support: The system now supports multiple data providers including yfinance, Alpha Vantage, and Finnhub.
• Dynamic Configuration: Data source configurations and priorities are managed via database collections (datasource_groupings and system_configs).
• Three-Layer Caching: A default integrated caching strategy using Redis, MongoDB, and File is enabled to accelerate data access.
• Enhanced Data Management: Improved database import/export functionality, including fixes for collection naming, date format conversions, and parameter passing.

The Message Data System provides a unified architecture for managing social media and internal financial messages. Key capabilities include:

• Crawler System: Supports social media (Weibo, Douyin) and internal messages (research reports, analyst notes) via a unified scheduler.
• Intelligent Processing: Automatic sentiment analysis (positive/negative/neutral), importance assessment (high/medium/low), and keyword extraction.
• High Performance: Dual-collection storage with 24 optimized indexes for millisecond-level queries and full-text search support for Chinese content.
• API & Production Ready: 30+ RESTful endpoints, standard response formats, error handling, and performance monitoring.

The scripts/ directory is organized by functional categories to assist with setup, validation, maintenance, development, deployment, containerization, and Git workflows.

The Prompt Template System v1.0.1 is an enhanced framework designed to manage and execute prompt templates for 13 specialized trading agents. It extends the base v1.0 functionality by adding persistent storage, user management, analysis preferences, and historical tracking.

Core Components

• 13 Agents: Categorized into Analysts (4), Researchers (2), Debaters (3), Managers (2), and Traders (1).
• 31 Templates: Includes 2-3 preset templates per agent, supporting user customization and version management.
• Analysis Preferences: Users can select from three modes: Aggressive (high risk/reward), Neutral (balanced), or Conservative (low risk/stable).
• Key Features: Database storage (5 collections), User management, Template management, History tracking, Versioning, 27 Web API endpoints, and Frontend UI integration.

The Tushare integration provides a standardized and reliable way to access fundamental financial data for quantitative analysis and investment research.

Supported Data Types

• Income Statement (利润表): Detailed revenue and expense data.
• Balance Sheet (资产负债表): Assets, liabilities, and equity information.
• Cash Flow Statement (现金流量表): Inflow and outflow of cash.
• Financial Indicators (财务指标): Calculated ratios and key performance metrics.

Key Features

• Standardization: Unified data formats and field mapping across different reports.
• Reliability: Built-in error handling and retry mechanisms for API calls.
• Efficiency: Optimized API calling patterns and batch processing capabilities.
• Flexibility: Supports various query methods and parameter configurations for multi-period trend analysis.

TradingAgents-CN is an AI-powered stock analysis learning platform that uses a multi-agent collaboration mechanism to help users understand stock analysis methods. It is specifically optimized for the Chinese A-share market and is an open-source project designed as a learning tool rather than a source of investment advice.

Key Features:

• Multi-agent collaborative analysis
• Optimized for China A-shares
• Open-source and self-deployable
• Disclaimer: It is a learning tool and does NOT provide investment advice, guaranteed returns, or professional consulting.