20 KiB

Raw Blame History

TradingAgents Configuration and Prompt Modification Guide

📖 Overview

This document provides a comprehensive guide for new users to modify configurations and customize prompts in the TradingAgents project. Through this guide, you will learn:

How to modify system configuration parameters
How to configure multi-market support (US stocks and China A-shares)
How to setup database integration (MongoDB and Redis)
How to configure multiple LLM providers (DashScope, OpenAI, Google, Anthropic)
How to customize prompts for various agents
How to add new features and configurations

🌟 New Features Overview

TongDaXin API Integration: Real-time A-share data access
Market Selection: Interactive CLI market selection
Exchange Support: Shanghai, Shenzhen, ChiNext, STAR Market
Intelligent Caching: Optimized data retrieval and storage

🤖 DashScope (Alibaba Cloud) Integration

Qwen Model Series: qwen-turbo, qwen-plus, qwen-max, qwen-max-longcontext
Embedding Service: DashScope embeddings for memory system
Intelligent Fallback: Automatic fallback to OpenAI when unavailable

🗄️ Database Integration

MongoDB: Persistent data storage and analytics
Redis: High-performance caching
Adaptive Cache: Intelligent cache management with automatic fallback

🔧 Configuration File Locations and Descriptions

1. Main Configuration Files

📁 `tradingagents/default_config.py`

Purpose: Core configuration file defining all default parameters

DEFAULT_CONFIG = {
    # Directory configuration
    "project_dir": "Project root directory path",
    "results_dir": "Results output directory",
    "data_dir": "Data storage directory", 
    "data_cache_dir": "Cache directory",
    
    # LLM model configuration
    "llm_provider": "dashscope",        # LLM provider: "dashscope", "openai", "google", "anthropic"
    "deep_think_llm": "qwen-plus",      # Deep thinking model
    "quick_think_llm": "qwen-turbo",    # Quick thinking model
    "backend_url": "https://dashscope.aliyuncs.com/api/v1",  # API backend URL
    
    # Debate and discussion settings
    "max_debate_rounds": 1,             # Maximum debate rounds
    "max_risk_discuss_rounds": 1,       # Maximum risk discussion rounds
    "max_recur_limit": 100,             # Maximum recursion limit
    
    # Tool settings
    "online_tools": True,               # Whether to use online tools
}

Modification Method:

Directly edit the tradingagents/default_config.py file
Modify the corresponding configuration values
Restart the application for changes to take effect

📁 `main.py`

Purpose: Runtime configuration override, allows temporary parameter adjustments without modifying default config

# Create custom configuration
config = DEFAULT_CONFIG.copy()
config["llm_provider"] = "google"                    # Use Google models
config["backend_url"] = "https://generativelanguage.googleapis.com/v1"
config["deep_think_llm"] = "gemini-2.0-flash"      # Deep thinking model
config["quick_think_llm"] = "gemini-2.0-flash"     # Quick thinking model
config["max_debate_rounds"] = 2                      # Increase debate rounds
config["online_tools"] = True                        # Enable online tools

Modification Method:

Edit the config section in main.py
Add or modify configuration items to override
Save and run

2. Dynamic Configuration Management

📁 `tradingagents/dataflows/config.py`

Purpose: Provides dynamic configuration get/set functionality

# Get current configuration
config = get_config()

# Dynamically modify configuration
set_config({
    "llm_provider": "anthropic",
    "max_debate_rounds": 3
})

🌟 New Features Configuration

1. Environment Variables Configuration (`.env`)

📁 `.env` File Setup

Purpose: Configure API keys and database settings

Required API Keys:

For US Stock Analysis:

# Choose one LLM provider
OPENAI_API_KEY=your_openai_api_key_here
# OR
GOOGLE_API_KEY=your_google_api_key_here
# OR
ANTHROPIC_API_KEY=your_anthropic_api_key_here

# FinnHub - Required for financial data
FINNHUB_API_KEY=your_finnhub_api_key_here

For China A-Share Analysis:

# DashScope - Required for Chinese stock analysis
DASHSCOPE_API_KEY=your_dashscope_api_key_here

# FinnHub - Required for financial data
FINNHUB_API_KEY=your_finnhub_api_key_here

For DashScope LLM Provider:

# DashScope - Required for Qwen models
DASHSCOPE_API_KEY=your_dashscope_api_key_here

# FinnHub - Required for financial data
FINNHUB_API_KEY=your_finnhub_api_key_here

Optional API Keys:

# OpenAI - Optional fallback
OPENAI_API_KEY=your_openai_api_key_here

# Google AI - For Gemini models
GOOGLE_API_KEY=your_google_api_key_here

# Anthropic - For Claude models
ANTHROPIC_API_KEY=your_anthropic_api_key_here

Database Configuration (Optional):

# MongoDB - For persistent data storage
MONGODB_ENABLED=false
MONGODB_HOST=localhost
MONGODB_PORT=27018
MONGODB_USERNAME=admin
MONGODB_PASSWORD=your_mongodb_password
MONGODB_DATABASE=tradingagents

# Redis - For high-performance caching
REDIS_ENABLED=false
REDIS_HOST=localhost
REDIS_PORT=6380
REDIS_PASSWORD=your_redis_password
REDIS_DB=0

2. Market Selection Configuration

📁 CLI Market Selection

Purpose: Configure supported markets and data sources

Supported Markets:

US Stock Market
- Format: 1-5 letter symbols (e.g., AAPL, SPY)
- Data Source: Yahoo Finance
- Validation: ^[A-Z]{1,5}$
China A-Share Market
- Format: 6-digit codes (e.g., 000001, 600036)
- Data Source: TongDaXin API
- Validation: ^\d{6}$
- Exchanges: Shanghai (60xxxx), Shenzhen (00xxxx), ChiNext (30xxxx), STAR (68xxxx)

Configuration in Code:

# Market-specific configuration
market_config = {
    "us_stock": {
        "data_source": "yahoo_finance",
        "pattern": r'^[A-Z]{1,5}$'
    },
    "china_a_share": {
        "data_source": "tongdaxin",
        "pattern": r'^\d{6}$'
    }
}

3. Database Integration Configuration

📁 MongoDB Configuration

Purpose: Persistent data storage and analytics

Setup Steps:

Start MongoDB:

docker run -d -p 27017:27017 --name mongodb mongo

Enable in .env:
```
MONGODB_ENABLED=true
```

Configuration Options:

mongodb_config = {
    "host": "localhost",
    "port": 27018,
    "database": "tradingagents",
    "username": "admin",
    "password": "your_password"
}

📁 Redis Configuration

Purpose: High-performance caching

Setup Steps:

Start Redis:

docker run -d -p 6379:6379 --name redis redis

Enable in .env:
```
REDIS_ENABLED=true
```

Configuration Options:

redis_config = {
    "host": "localhost",
    "port": 6380,
    "password": "your_password",
    "db": 0
}

4. LLM Provider Configuration

📁 DashScope (Alibaba Cloud) Configuration

Purpose: Chinese-optimized LLM provider

Supported Models:

qwen-turbo: Fast response, suitable for quick tasks
qwen-plus: Balanced performance and cost (Recommended)
qwen-max: Best performance for complex analysis
qwen-max-longcontext: Ultra-long context support

Configuration Example:

dashscope_config = {
    "llm_provider": "dashscope",
    "deep_think_llm": "qwen-plus",
    "quick_think_llm": "qwen-turbo",
    "backend_url": "https://dashscope.aliyuncs.com/api/v1"
}

API Key Setup:

Visit: https://dashscope.aliyun.com/
Register Alibaba Cloud account
Enable DashScope service
Get API key
Set in .env: DASHSCOPE_API_KEY=your_key

📁 Multi-LLM Fallback Configuration

Purpose: Intelligent fallback between LLM providers

Fallback Priority:

Primary: DashScope (if configured)
Secondary: OpenAI (if configured)
Tertiary: Google AI (if configured)
Fallback: Anthropic (if configured)

Configuration:

fallback_config = {
    "primary_provider": "dashscope",
    "fallback_providers": ["openai", "google", "anthropic"],
    "auto_fallback": True,
    "retry_attempts": 3
}

🤖 Agent Prompt Modification Guide

1. Analyst Prompts

📁 Market Analyst (`tradingagents/agents/analysts/market_analyst.py`)

Location: system_message variable at lines 24-50

Current Prompt:

system_message = (
    """You are a trading assistant tasked with analyzing financial markets. 
    Your role is to select the **most relevant indicators** for a given market 
    condition or trading strategy from the following list..."""
)

Modification Example:

system_message = (
    """You are a professional market analyst specializing in financial market analysis.
    Your task is to select the most relevant indicators from the following list,
    providing analysis for specific market conditions or trading strategies.
    Goal: Choose up to 8 indicators that provide complementary insights without redundancy..."""
)

📁 Fundamentals Analyst (`tradingagents/agents/analysts/fundamentals_analyst.py`)

Location: system_message variable at lines 23-26

Key Modification Points:

Analysis depth requirements
Report format requirements
Focus financial metrics

📁 News Analyst (`tradingagents/agents/analysts/news_analyst.py`)

Location: system_message variable at lines 20-23

Key Modification Points:

News source preferences
Analysis time range
Types of news to focus on

📁 Social Media Analyst (`tradingagents/agents/analysts/social_media_analyst.py`)

Location: system_message variable at lines 19-22

Key Modification Points:

Sentiment analysis depth
Social media platform preferences
Sentiment weight settings

2. Researcher Prompts

📁 Bull Researcher (`tradingagents/agents/researchers/bull_researcher.py`)

Location: prompt variable at lines 25-43

Current Prompt Structure:

prompt = f"""You are a Bull Analyst advocating for investing in the stock.

Key points to focus on:
- Growth Potential: Highlight market opportunities, revenue projections, and scalability
- Competitive Advantages: Emphasize unique products, strong branding, or market dominance
- Positive Indicators: Use financial health, industry trends, and recent positive news as evidence
- Bear Counterpoints: Critically analyze bear arguments with specific data and sound reasoning
"""

Modification Suggestions:

Adjust analysis focus
Modify argumentation strategy
Customize rebuttal logic

📁 Bear Researcher (`tradingagents/agents/researchers/bear_researcher.py`)

Key Modification Points:

Risk identification focus
Pessimistic scenario analysis
Strategy for countering bull arguments

3. Trader Prompts

📁 Trader (`tradingagents/agents/trader/trader.py`)

Location: System message in messages array at lines 30-36

Current Prompt:

{
    "role": "system",
    "content": f"""You are a trading agent analyzing market data to make 
    investment decisions. Based on your analysis, provide a specific 
    recommendation to buy, sell, or hold. End with a firm decision and 
    always conclude your response with 'FINAL TRANSACTION PROPOSAL: 
    **BUY/HOLD/SELL**' to confirm your recommendation.""",
}

Modification Example:

{
    "role": "system", 
    "content": f"""You are a professional trading agent responsible for analyzing 
    market data and making investment decisions.
    
    Decision Requirements:
    1. Provide detailed analysis reasoning
    2. Consider risk management
    3. Must end with 'FINAL TRANSACTION PROPOSAL: **BUY/HOLD/SELL**'
    
    Historical Lessons: {past_memory_str}""",
}

4. Risk Management Prompts

📁 Conservative Debater (`tradingagents/agents/risk_mgmt/conservative_debator.py`)

📁 Aggressive Debater (`tradingagents/agents/risk_mgmt/aggresive_debator.py`)

📁 Neutral Debater (`tradingagents/agents/risk_mgmt/neutral_debator.py`)

Key Modification Points:

Risk tolerance settings
Debate style adjustments
Decision weight allocation

5. Reflection System Prompts

📁 Reflection System (`tradingagents/graph/reflection.py`)

Location: _get_reflection_prompt method at lines 15-47

Current Prompt Structure:

return """
You are an expert financial analyst tasked with reviewing trading 
decisions/analysis and providing a comprehensive, step-by-step analysis.

1. Reasoning: Analyze whether each trading decision was correct
2. Improvement: Propose revisions for incorrect decisions  
3. Summary: Summarize lessons learned from successes and failures
4. Query: Extract key insights into concise sentences
"""

🎯 Prompt Modification Best Practices

1. Pre-modification Preparation

Backup Original Files:

cp tradingagents/agents/trader/trader.py tradingagents/agents/trader/trader.py.backup

Understand Agent Roles: Ensure modifications align with expected agent functionality
Prepare Test Environment: Validate modifications in test environment

2. Prompt Modification Techniques

🔍 Structured Prompts

system_message = f"""
Role Definition: You are a {role_name}

Main Tasks:
1. {task_1}
2. {task_2}
3. {task_3}

Analysis Requirements:
- Depth: {analysis_depth}
- Format: {output_format}
- Focus: {focus_areas}

Output Format:
{output_template}

Constraints:
- {constraint_1}
- {constraint_2}
"""

⚙️ Parameterized Prompts

def create_analyst_prompt(
    role="Market Analyst",
    analysis_depth="Detailed",
    time_horizon="1 week",
    risk_tolerance="Moderate",
    output_language="English"
):
    return f"""
You are a professional {role}, please analyze based on the following parameters:

Analysis Depth: {analysis_depth}
Time Horizon: {time_horizon}  
Risk Preference: {risk_tolerance}
Output Language: {output_language}

Please provide corresponding market analysis and investment recommendations based on these parameters.
"""

3. Common Modification Scenarios

📈 Adjusting Analysis Focus

# Original: General market analysis
system_message = "Analyze overall market trends..."

# Modified: Focus on specific industry
system_message = "Analyze technology stock market trends, focusing on AI, semiconductor, and cloud computing industries..."

🎯 Modifying Decision Style

# Original: Conservative
"provide conservative investment recommendations..."

# Modified: Aggressive  
"provide aggressive growth-oriented investment recommendations with higher risk tolerance..."

🔧 New Configuration Items

1. Cache Configuration (`tradingagents/dataflows/cache_manager.py`)

# Add new cache configuration in cache_manager.py
self.cache_config = {
    'us_stock_data': {
        'ttl_hours': 2,      # US stock data cached for 2 hours
        'description': 'US stock historical data'
    },
    'china_stock_data': {
        'ttl_hours': 1,      # A-share data cached for 1 hour  
        'description': 'A-share historical data'
    },
    # Add new cache type
    'crypto_data': {
        'ttl_hours': 0.5,    # Crypto data cached for 30 minutes
        'description': 'Cryptocurrency data'
    }
}

2. API Configuration

# Add new API configuration in default_config.py
DEFAULT_CONFIG = {
    # Existing configuration...
    
    # New API configuration
    "api_keys": {
        "finnhub": "your_finnhub_api_key",
        "alpha_vantage": "your_alpha_vantage_key",
        "polygon": "your_polygon_key"
    },
    
    # API limit configuration
    "api_limits": {
        "finnhub_calls_per_minute": 60,
        "alpha_vantage_calls_per_minute": 5,
        "polygon_calls_per_minute": 100
    }
}

🚀 Quick Start Examples

1. Switch to Google Models

# Edit main.py
config = DEFAULT_CONFIG.copy()
config["llm_provider"] = "google"
config["backend_url"] = "https://generativelanguage.googleapis.com/v1"
config["deep_think_llm"] = "gemini-2.0-flash"
config["quick_think_llm"] = "gemini-2.0-flash"

🚀 Supported Google Models

Fast Thinking Models (Quick Analysis):

gemini-2.0-flash-lite - Cost efficiency and low latency
gemini-2.0-flash - Next generation features, speed, and thinking ⭐ Recommended
gemini-2.5-flash-preview-05-20 - Adaptive thinking, cost efficiency

Deep Thinking Models (Complex Analysis):

gemini-2.0-flash-lite - Cost efficiency and low latency
gemini-2.0-flash - Next generation features, speed, and thinking ⭐ Current Default
gemini-2.5-flash-preview-05-20 - Adaptive thinking, cost efficiency
gemini-2.5-pro-preview-06-05 - Professional-grade performance

🔑 Google API Key Setup

Method 1: Environment Variable (Recommended)

export GOOGLE_API_KEY="your_google_api_key_here"

Method 2: In Code

import os
os.environ["GOOGLE_API_KEY"] = "your_google_api_key_here"

Method 3: .env File

# Create .env file in project root
GOOGLE_API_KEY=your_google_api_key_here

📋 Model Selection Examples

High Performance Setup:

config["deep_think_llm"] = "gemini-2.5-pro-preview-06-05"  # Best reasoning
config["quick_think_llm"] = "gemini-2.0-flash"  # Fast response

Cost-Optimized Setup:

config["deep_think_llm"] = "gemini-2.0-flash-lite"  # Economical
config["quick_think_llm"] = "gemini-2.0-flash-lite"  # Economical

Balanced Setup (Current Default):

config["deep_think_llm"] = "gemini-2.0-flash"  # Good performance
config["quick_think_llm"] = "gemini-2.0-flash"  # Good speed

2. Add Risk Control

# Edit tradingagents/agents/trader/trader.py
messages = [
    {
        "role": "system",
        "content": f"""You are a professional trading agent with strict risk control awareness.

        Trading Principles:
        1. Risk first, returns second
        2. Strict stop-loss, protect capital
        3. Diversified investment, reduce risk
        4. Data-driven, rational decisions

        Decision Process:
        1. Analyze market trends and technical indicators
        2. Assess fundamental and news impact
        3. Calculate risk-reward ratio
        4. Set stop-loss and take-profit points
        5. Make final trading decision

        Output Requirements:
        - Must include risk assessment
        - Must set stop-loss points
        - Must end with 'FINAL TRANSACTION PROPOSAL: **BUY/HOLD/SELL**'
        
        Historical Experience: {past_memory_str}""",
    },
    context,
]

📝 Important Notes

Backup Important: Always backup original files before modification
Test Validation: Validate modifications in test environment
Version Control: Use Git to manage configuration changes
Documentation Updates: Update related documentation promptly
Team Collaboration: Sync configuration changes with team members

🔗 Quick File Index

Function	File Path	Description
Main Config	`tradingagents/default_config.py`	System default configuration
Runtime Config	`main.py`	Runtime configuration override
Dynamic Config	`tradingagents/dataflows/config.py`	Configuration management interface
Market Analyst	`tradingagents/agents/analysts/market_analyst.py`	Technical analysis prompts
Fundamentals Analyst	`tradingagents/agents/analysts/fundamentals_analyst.py`	Fundamental analysis prompts
News Analyst	`tradingagents/agents/analysts/news_analyst.py`	News analysis prompts
Social Media Analyst	`tradingagents/agents/analysts/social_media_analyst.py`	Sentiment analysis prompts
Bull Researcher	`tradingagents/agents/researchers/bull_researcher.py`	Bull analysis prompts
Bear Researcher	`tradingagents/agents/researchers/bear_researcher.py`	Bear analysis prompts
Trader	`tradingagents/agents/trader/trader.py`	Trading decision prompts
Reflection System	`tradingagents/graph/reflection.py`	Reflection analysis prompts
Cache Config	`tradingagents/dataflows/cache_manager.py`	Cache management configuration

Through this guide, you should be able to easily modify the TradingAgents project's configuration and prompts to meet your specific needs.

20 KiB Raw Blame History

TradingAgents Configuration and Prompt Modification Guide

📖 Overview

🌟 New Features Overview

🇨🇳 China A-Share Market Support

🤖 DashScope (Alibaba Cloud) Integration

🗄️ Database Integration

🔧 Configuration File Locations and Descriptions

1. Main Configuration Files

📁 tradingagents/default_config.py

📁 main.py

2. Dynamic Configuration Management

📁 tradingagents/dataflows/config.py

🌟 New Features Configuration

1. Environment Variables Configuration (.env)

📁 .env File Setup

2. Market Selection Configuration

📁 CLI Market Selection

3. Database Integration Configuration

📁 MongoDB Configuration

📁 Redis Configuration

4. LLM Provider Configuration

📁 DashScope (Alibaba Cloud) Configuration

📁 Multi-LLM Fallback Configuration

🤖 Agent Prompt Modification Guide

1. Analyst Prompts

📁 Market Analyst (tradingagents/agents/analysts/market_analyst.py)

📁 Fundamentals Analyst (tradingagents/agents/analysts/fundamentals_analyst.py)

📁 News Analyst (tradingagents/agents/analysts/news_analyst.py)

📁 Social Media Analyst (tradingagents/agents/analysts/social_media_analyst.py)

2. Researcher Prompts

📁 Bull Researcher (tradingagents/agents/researchers/bull_researcher.py)

📁 Bear Researcher (tradingagents/agents/researchers/bear_researcher.py)

3. Trader Prompts

📁 Trader (tradingagents/agents/trader/trader.py)

4. Risk Management Prompts

📁 Conservative Debater (tradingagents/agents/risk_mgmt/conservative_debator.py)

📁 Aggressive Debater (tradingagents/agents/risk_mgmt/aggresive_debator.py)

📁 Neutral Debater (tradingagents/agents/risk_mgmt/neutral_debator.py)

5. Reflection System Prompts

📁 Reflection System (tradingagents/graph/reflection.py)

🎯 Prompt Modification Best Practices

1. Pre-modification Preparation

2. Prompt Modification Techniques

🔍 Structured Prompts

⚙️ Parameterized Prompts

3. Common Modification Scenarios

📈 Adjusting Analysis Focus

🎯 Modifying Decision Style

🔧 New Configuration Items

1. Cache Configuration (tradingagents/dataflows/cache_manager.py)

2. API Configuration

🚀 Quick Start Examples

1. Switch to Google Models

🚀 Supported Google Models

🔑 Google API Key Setup

📋 Model Selection Examples

2. Add Risk Control

📝 Important Notes

🔗 Quick File Index

20 KiB

Raw Blame History

📁 `tradingagents/default_config.py`

📁 `main.py`

📁 `tradingagents/dataflows/config.py`

1. Environment Variables Configuration (`.env`)

📁 `.env` File Setup

📁 Market Analyst (`tradingagents/agents/analysts/market_analyst.py`)

📁 Fundamentals Analyst (`tradingagents/agents/analysts/fundamentals_analyst.py`)

📁 News Analyst (`tradingagents/agents/analysts/news_analyst.py`)

📁 Social Media Analyst (`tradingagents/agents/analysts/social_media_analyst.py`)

📁 Bull Researcher (`tradingagents/agents/researchers/bull_researcher.py`)

📁 Bear Researcher (`tradingagents/agents/researchers/bear_researcher.py`)

📁 Trader (`tradingagents/agents/trader/trader.py`)

📁 Conservative Debater (`tradingagents/agents/risk_mgmt/conservative_debator.py`)

📁 Aggressive Debater (`tradingagents/agents/risk_mgmt/aggresive_debator.py`)

📁 Neutral Debater (`tradingagents/agents/risk_mgmt/neutral_debator.py`)

📁 Reflection System (`tradingagents/graph/reflection.py`)

1. Cache Configuration (`tradingagents/dataflows/cache_manager.py`)