LLM Agent Trader

AI-powered stock trading backtesting system that integrates Large Language Models for intelligent trading decision analysis.

System Architecture

flowchart TD
    %% User Interface Layer
    A[Frontend - Next.js] --> B[API Gateway - FastAPI Backend]
    
    %% Main Function Modules
    B --> C[LLM Streaming Backtest Engine]
    B --> D[Backtest Analysis API]
    B --> E[Daily Feedback API]
    
    %% Data Layer
    F[Stock Data Service<br/>YFinance] --> C
    G[SQLite Database<br/>Backtest Logs] --> D
    G --> E
    
    %% LLM Strategy Engine
    C --> H[LLM Smart Strategy]
    H --> I[Azure OpenAI<br/>GPT-4]
    H --> J[Technical Analysis Engine]
    H --> K[Risk Management Module]
    
    %% Backtest Execution Flow
    C --> L[Trading Signal Generation]
    L --> M[Performance Calculation]
    M --> N[Result Recording]
    N --> G
    
    %% Style Definitions
    classDef frontend fill:#e1f5fe
    classDef backend fill:#f3e5f5
    classDef llm fill:#fff3e0
    classDef data fill:#e8f5e8
    
    class A frontend
    class B,C,D,E backend
    class H,I,J,K llm
    class F,G data

Loading

Quick Start

Prerequisites

• macOS/Linux: Native support for make commands
• Windows: May require additional setup (WSL, Git Bash, or make utility installation)

Install Dependencies

make install

Environment Setup

Copy and configure your .env file:

cp .env.example .env

Switch LLM Provider: Edit .env and comment/uncomment the API keys:

# Use Azure OpenAI (default)
AZURE_OPENAI_API_KEY=your_key
# GOOGLE_API_KEY=your_key  # comment out

# Use Google Gemini instead
# AZURE_OPENAI_API_KEY=your_key  # comment out  
GOOGLE_API_KEY=your_key

Development Mode

make run

🎉 Success! After setup, open your browser and navigate to: http://localhost:3000 to access the web application

Other Commands

make stop     # Stop all services
make test     # Run tests
make clean    # Clean cache files
make format   # Format code

Windows Users

If you encounter issues with make commands on Windows, consider:

• WSL (Windows Subsystem for Linux): Recommended approach
• Git Bash: Included with Git for Windows
• Make for Windows: Install GNU Make utility