AutoHedge 🚀

Build your autonomous hedge fund in minutes. AutoHedge harnesses the power of swarm intelligence and AI agents to automate market analysis, risk management, and trade execution.

🌟 Features

• Multi-Agent Architecture: Leverages specialized AI agents for different aspects of trading

  • Director Agent for strategy and thesis generation
  • Quant Agent for technical analysis
  • Risk Management Agent for position sizing and risk assessment
  • Execution Agent for trade implementation

• Real-Time Market Analysis: Integrates with market data providers for live analysis

• Risk-First Approach: Built-in risk management and position sizing

• Structured Output: JSON-formatted trade recommendations and analysis

• Comprehensive Logging: Detailed logging system for trade tracking and debugging

• Extensible Framework: Easy to customize and extend with new capabilities

📋 Requirements

• Python 3.8+
• swarms package
• tickr-agent
• Additional dependencies listed in requirements.txt

🚀 Quick Start

Installation

pip install -U autohedge

Environment Variables

OPENAI_API_KEY=""
WORKSPACE_DIR="agent_workspace"

Basic Usage

# Example usage
from autohedge import AutoFund

# Define the stocks to analyze
stocks = ["NVDA"]

# Initialize the trading system with the specified stocks
trading_system = AutoFund(stocks)

# Define the task for the trading cycle
task = "Let's analyze nvidia to see if we should buy it, we have 50k$ in allocation"

# Run the trading cycle and print the results
print(trading_system.run(task=task))

🏗️ Architecture

AutoHedge uses a multi-agent architecture where each agent specializes in a specific aspect of the trading process:

graph TD
    A[Director Agent] --> B[Quant Agent]
    B --> C[Risk Manager]
    C --> D[Execution Agent]
    D --> E[Trade Output]

Loading

Agent Roles

1. Director Agent

   • Generates trading theses
   • Coordinates overall strategy
   • Analyzes market conditions

2. Quant Agent

   • Performs technical analysis
   • Evaluates statistical patterns
   • Calculates probability scores

3. Risk Manager

   • Assesses trade risks
   • Determines position sizing
   • Sets risk parameters

4. Execution Agent

   • Generates trade orders
   • Sets entry/exit points
   • Manages order execution

📊 Output Format

AutoHedge generates structured output using Pydantic models:

class AutoHedgeOutput(BaseModel):
    id: str                         # Unique identifier
    name: Optional[str]             # Strategy name
    description: Optional[str]      # Strategy description
    stocks: Optional[List[str]]     # List of stocks
    task: Optional[str]             # Analysis task
    thesis: Optional[str]           # Trading thesis
    risk_assessment: Optional[str]  # Risk analysis
    order: Optional[Dict]           # Trade order details
    timestamp: str                  # Timestamp
    current_stock: str              # Current stock being analyzed

🔧 Configuration

AutoHedge can be configured through environment variables or initialization parameters:

trading_system = AutoFund(
    name="CustomStrategy",
    description="My Trading Strategy",
    stocks=["NVDA", "AAPL"],
    output_dir="custom_outputs"
)

📝 Logging

AutoHedge uses the loguru library for comprehensive logging:

logger.add(
    "trading_system_{time}.log",
    rotation="500 MB",
    retention="10 days",
    level="INFO",
    format="{time:YYYY-MM-DD at HH:mm:ss} | {level} | {message}"
)

🔍 Advanced Usage

Custom Agent Configuration

from autohedge import TradingDirector, QuantAnalyst, RiskManager

# Custom director configuration
director = TradingDirector(
    stocks=["NVDA", "AAPL"],
    output_dir="custom_outputs"
)

# Custom analysis
analysis = director.generate_thesis(
    task="Generate comprehensive analysis",
    stock="NVDA"
)

Risk Management

from autohedge import RiskManager

risk_manager = RiskManager()
assessment = risk_manager.assess_risk(
    stock="NVDA",
    thesis=thesis,
    quant_analysis=analysis
)

Diagrams

🏗️ System Architecture

High-Level Component Overview

flowchart TB
    subgraph Client
        A[AutoHedge Client] --> B[Trading System]
    end
    
    subgraph Agents["Multi-Agent System"]
        B --> C{Director Agent}
        C --> D[Quant Agent]
        C --> E[Risk Agent]
        C --> F[Execution Agent]
        
        D --> G[Technical Analysis]
        D --> H[Statistical Analysis]
        
        E --> I[Risk Assessment]
        E --> J[Position Sizing]
        
        F --> K[Order Generation]
        F --> L[Trade Execution]
    end
    
    subgraph Output
        K --> M[JSON Output]
        L --> N[Trade Logs]
    end

Loading

Trading Cycle Sequence

sequenceDiagram
    participant C as Client
    participant D as Director
    participant Q as Quant
    participant R as Risk
    participant E as Execution
    
    C->>D: Initialize Trading Cycle
    activate D
    D->>D: Generate Thesis
    D->>Q: Request Analysis
    activate Q
    Q-->>D: Return Analysis
    deactivate Q
    D->>R: Request Risk Assessment
    activate R
    R-->>D: Return Risk Profile
    deactivate R
    D->>E: Generate Order
    activate E
    E-->>D: Return Order Details
    deactivate E
    D-->>C: Return Complete Analysis
    deactivate D

Loading

Trade State Machine

stateDiagram-v2
    [*] --> Initialization
    Initialization --> ThesisGeneration
    
    ThesisGeneration --> QuantAnalysis
    QuantAnalysis --> RiskAssessment
    
    RiskAssessment --> OrderGeneration: Risk Approved
    RiskAssessment --> ThesisGeneration: Risk Rejected
    
    OrderGeneration --> OrderExecution
    OrderExecution --> Monitoring
    
    Monitoring --> ThesisGeneration: New Cycle
    Monitoring --> [*]: Complete

Loading

Data Flow

flowchart LR
    subgraph Input
        A[Market Data] --> B[Technical Indicators]
        A --> C[Fundamental Data]
    end
    
    subgraph Processing
        B --> D[Quant Analysis]
        C --> D
        D --> E[Risk Analysis]
        E --> F[Order Generation]
    end
    
    subgraph Output
        F --> G[Trade Orders]
        F --> H[Risk Reports]
        F --> I[Performance Metrics]
    end

Loading

Class Structure

classDiagram
    class AutoFund {
        +String name
        +String description
        +List stocks
        +Path output_dir
        +run()
    }
    
    class TradingDirector {
        +Agent director_agent
        +TickrAgent tickr
        +generate_thesis()
    }
    
    class QuantAnalyst {
        +Agent quant_agent
        +analyze()
    }
    
    class RiskManager {
        +Agent risk_agent
        +assess_risk()
    }
    
    class ExecutionAgent {
        +Agent execution_agent
        +generate_order()
    }
    
    AutoFund --> TradingDirector
    AutoFund --> QuantAnalyst
    AutoFund --> RiskManager
    AutoFund --> ExecutionAgent

Loading

API Documentation

To use the API, git clone the repo:

1. Installation

pip3 install -r requirements.txt

2. Launch API Server

python api.py

Server will start at http://localhost:8000

API Endpoints

Authentication

All endpoints except /users (POST) require the X-API-Key header.

User Management

Create User

POST /users
Content-Type: application/json

{
    "username": "trader1",
    "email": "[email protected]",
    "fund_name": "Alpha Fund",
    "fund_description": "AI Trading Strategy"
}

Returns API key in response.

Get User Profile

GET /users/me
X-API-Key: your-api-key

Trading Operations

Create Trade

POST /trades
X-API-Key: your-api-key
Content-Type: application/json

{
    "stocks": ["NVDA", "AAPL"],
    "task": "Analyze tech stocks for $1M allocation",
    "allocation": 1000000.0,
    "strategy_type": "momentum",
    "risk_level": 7
}

List Trades

GET /trades?limit=10&skip=0&status=completed
X-API-Key: your-api-key

Get Specific Trade

GET /trades/{trade_id}
X-API-Key: your-api-key

Delete Trade

DELETE /trades/{trade_id}
X-API-Key: your-api-key

Analytics

Get Historical Analytics

GET /analytics/history?days=30
X-API-Key: your-api-key

Quick Test Script

import requests

BASE_URL = "http://localhost:8000"

# Create user and get API key
def get_api_key():
    response = requests.post(
        f"{BASE_URL}/users",
        json={
            "username": "test_trader",
            "email": "[email protected]",
            "fund_name": "Test Fund",
            "fund_description": "Test Strategy"
        }
    )
    return response.json()["api_key"]

# Use the API
api_key = get_api_key()
headers = {"X-API-Key": api_key}

# Create a trade
trade = requests.post(
    f"{BASE_URL}/trades",
    headers=headers,
    json={
        "stocks": ["NVDA"],
        "task": "Test trade",
        "allocation": 1000000.0
    }
)
print(trade.json())

Running in Production

1. Use a production ASGI server:

uvicorn main:app --host 0.0.0.0 --port 8000 --workers 4

2. Set environment variables:

export AUTOHEDGE_ENV=production
export AUTOHEDGE_LOG_LEVEL=INFO

Error Codes

• 401: Invalid API key
• 403: Unauthorized access
• 404: Resource not found
• 422: Validation error
• 500: Server error

Best Practices

1. Store API keys securely
2. Use appropriate error handling
3. Implement rate limiting in production
4. Monitor API usage
5. Regularly backup trade data

Interactive Documentation

• Swagger UI: http://localhost:8000/docs
• ReDoc: http://localhost:8000/redoc

🤝 Contributing

We welcome contributions! Please see our Contributing Guidelines for details.

1. Fork the repository
2. Create your feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

📜 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• Swarms for the AI agent framework
• Tickr Agent for market data integration

📞 Support

• Issue Tracker: GitHub Issues
• Discord: Join our community

---

Created with ❤️ by The Swarm Corporation