SuperAPI AI Directory
Partner with us中文AI newsFavorites
DirectoryNews and tutorials

TradingAgents-CN - Chinese Multi-Agent Financial Trading Decision Framework

Author: AI小集 · Original source: AI工具集

What is TradingAgents-CN

TradingAgents-CN is a Chinese financial trading decision framework based on multi-agent large models. It was developed on the basis of TauricResearch/TradingAgents and provides Chinese users with a complete documentation system and localization support. The framework simulates the professional division of labor and collaborative decision-making process of a real trading company, evaluating market conditions through the collaboration of multiple specialized AI agents, including a multi-agent collaboration architecture such as an analyst team, researcher team, trader agent, risk management, and management. It supports multiple large language models, such as Alibaba Bailian, Google AI, OpenAI, and Anthropic, and can be operated intuitively through a Web interface. TradingAgents-CN supports Chinese financial markets such as A-shares and Hong Kong stocks, integrates Chinese financial data such as Tushare and AkShare, and provides Chinese users with convenient and intelligent financial trading decision support.

TradingAgents-CN

Main Features of TradingAgents-CN

  • Multi-Agent Collaboration Architecture
    • Analyst Team: Includes fundamental analysts, technical analysts, news analysts, and social media analysts, responsible for analyzing the market from different dimensions.
    • Researcher Team: Composed of bullish researchers and bearish researchers, conducting structured debates and providing in-depth market insights.
    • Trader Agent: Integrates all information and makes the final trading decision.
    • Risk Management: Conducts multi-level risk assessment and management of trades to ensure trading safety.
    • Management: Coordinates the work of each team to ensure the quality and efficiency of decisions.
  • Multi-LLM Model Support
    • Domestic LLM Integration: Fully supports Alibaba Bailian (qwen-turbo, qwen-plus-latest, qwen-max) and Google AI (gemini-2.0-flash, gemini-1.5-pro, gemini-1.5-flash).
    • International LLM Support: Supports models such as OpenAI (GPT-4o, GPT-4o-mini, GPT-3.5-turbo) and Anthropic (Claude-3-Opus, Claude-3-Sonnet, Claude-3-Haiku).
  • Intuitive Operation: A modern Web interface based on Streamlit. Users do not need command-line operations and can use it through a browser.
  • Real-Time Progress Display: The analysis process is visualized, with progress displayed in real time to avoid user waiting anxiety.
  • Intelligent Configuration: Supports 5 levels of research depth selection, from quick analysis (2-4 minutes) to comprehensive analysis (15-25 minutes).
  • Result Display: Structurally displays investment recommendations, target prices, confidence levels, risk assessments, etc.
  • Chinese Interface: A fully Chinese-localized user interface and analysis results, lowering the usage threshold for Chinese users.
  • Comprehensive Data Integration
    • A-share Data Support: Provides A-share real-time quotes and historical data through the Tongdaxin API.
    • US Stock Data Support: Supports real-time quotes from data sources such as FinnHub and Yahoo Finance.
    • News Data Integration: Integrates Google News, financial news, etc., providing real-time news data.
    • Social Data Support: Supports sentiment analysis of social media such as Reddit and Twitter.
    • Database Support: Supports MongoDB data persistence and Redis high-speed caching, improving data access efficiency.
  • Parallel Processing: Multi-agent parallel analysis, improving analysis efficiency.
  • Intelligent Caching: Multi-layer caching strategy, reducing API call costs.
  • Real-time Analysis: Supports real-time market data analysis, responding quickly to market changes.
  • Flexible Configuration: Highly customizable agent behavior and model selection, meeting the needs of different users.
  • Cost Control: Real-time statistics on Token usage, helping users understand costs. Provides cost estimates under different configurations, helping users optimize usage costs.
  • Chinese Documentation System: Provides over 50,000 words of detailed Chinese documentation, covering a complete path from beginner to expert.
  • Example Tutorials: Includes 100+ practical code examples, helping users get started quickly and learn in depth.

Technical Principles of TradingAgents-CN

  • Role Specialization and Division of Labor: Assign clear, well-defined roles and specific goals to LLM agents, decomposing complex trading tasks into smaller, manageable subtasks, so that each agent focuses on its area of expertise.
  • Multi-Agent Collaboration Mechanism: Adopts a multi-agent system architecture, simulating the organizational structure and collaboration processes of real-world trading companies. Through information sharing, debate, and feedback among agents, it achieves comprehensive analysis of market information and optimization of decision-making.
  • Model Selection and Application: Selects appropriate LLM models for processing based on the needs of different tasks. For example, using fast-thinking models (such as gpt-4o-mini) for data retrieval and information summarization, and using deep-thinking models (such as gpt-4o) for complex reasoning and decision support.
  • LLM and Natural Language Processing: Based on the powerful natural language processing capabilities of LLMs, it conducts in-depth understanding and analysis of text data, extracting key information and knowledge. Agents can engage in natural language dialogue and debate, and explain the decision-making process to users.
  • Fusion of Structured and Unstructured Data: Combines structured data (such as financial statements, trading data, etc.) with unstructured data (such as news articles, social media sentiment, etc.). Through the analysis and processing of LLMs, it uncovers correlations and potential value among data, providing richer information support for trading decisions.
  • Data Pipeline Design: Through efficient data pipeline design, it implements processes such as data acquisition, cleaning, and feature engineering. For example, after raw data is obtained from the FinnHub API, it undergoes cleaning and feature engineering processing, and is ultimately merged into the agent knowledge sharing pool.

Project Address of TradingAgents-CN

  • Github Repository：https://github.com/hsliuping/TradingAgents-CN

Application Scenarios of TradingAgents-CN

  • Individual Stock Analysis: Conducts in-depth analysis of information such as the fundamentals, sentiment, news, and technical indicators of individual stocks, and formulates buy, sell, or hold trading strategies.
  • Portfolio Risk Assessment: Conducts comprehensive risk assessment of a portfolio or individual assets, identifies potential risk factors, and formulates corresponding risk control measures.
  • Market Risk Warning: Real-time monitoring of market risks, timely issuance of warning signals, helping investors avoid or reduce investment losses.
  • Market trend analysis: Providing market research tools for financial institutions and researchers, conducting in-depth analysis of market trends, industry dynamics, and macroeconomic factors.
  • Investment strategy research: Through debate and collaboration among multiple agents, exploring and researching new investment strategies to provide theoretical support for investment decisions.