Natural Language Trading Strategy Backtester

An intelligent web application that translates trading strategies written in plain English into a quantitative backtest, providing comprehensive performance metrics and an interactive equity curve.

📖 Table of Contents

• About The Project
• Key Features
• How It Works
• Technology Stack
• Getting Started
  • Prerequisites
  • Installation
  • Usage
• Project Structure
• Future Improvements
• License

🌟 About The Project

This project bridges the gap between complex trading ideas and quantitative analysis. Instead of writing intricate code to test a strategy, users can simply describe it in plain English.

The application leverages a Large Language Model (Google Gemini) to parse the user's intent, constructs a dynamic strategy, and runs a rigorous backtest against historical market data using the Backtrader framework.

The goal is to empower traders and analysts to rapidly iterate on their ideas without the friction of programming, providing instant feedback on a strategy's historical performance.

✨ Key Features

• Natural Language Processing: Describe your strategy in English (e.g., "Buy when the 20-day SMA crosses above the 50-day EMA").
• Dynamic Strategy Generation: The backend dynamically configures indicators and trading rules from LLM output.
• Comprehensive Performance Metrics: Includes Sharpe Ratio, Max Drawdown, Alpha, Beta, Win Rate, and more.
• Interactive Equity Curve: Visualize your portfolio’s value over time with Chart.js.
• Multi-Asset Universe: Backtests run on a predefined basket of 25 popular stocks and ETFs.
• Risk Management Built-In: Trades sized by fixed portfolio risk and ATR-based volatility.
• Responsive Web Interface: Modern UI for desktop and mobile.

🛠️ How It Works

1. Frontend (UI): User types a strategy and clicks Run Backtest.
2. Backend (Flask): Handles the /backtest request.
3. LLM Parser (llm_parser.py): Sends the natural language input to Google Gemini with a structured prompt.
4. JSON Strategy Config: Gemini returns JSON defining indicators and buy/sell rules.
5. Backtesting Engine (backtest.py): Runs strategy in Backtrader with historical data from yfinance.
6. Performance Results: Gathers metrics, chart data, and summary.
7. Frontend Update: Displays results dynamically with charts and statistics.

💻 Technology Stack

• Backend: Python, Flask
• Backtesting Engine: Backtrader
• Data Source: yfinance
• LLM: Google Gemini
• Frontend:
  • HTML5, CSS3 (CSS Grid)
  • JavaScript (ES6+)
  • Charting: Chart.js

🚀 Getting Started

Prerequisites

• Python 3.8+
• pip (Python package manager)
• Google API Key for Gemini (get one from Google AI Studio)

Installation

# Clone the repository
git clone https://github.com/JTSmullen/MCPStrategyBacktester.git
cd MCPStrategyBacktester

# (Optional) Create a virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

Set up your API key in a .env file:

GOOGLE_API_KEY="YOUR_API_KEY_HERE"

Run the Application

flask run

Visit http://127.0.0.1:5000 in your browser.

📈 Usage

1. Navigate to the web app.
2. Enter a strategy such as:
   • "Buy when the 14-day RSI is below 30, sell when above 70."
   • "Go long when the price closes above the 20-day Bollinger Band, sell when below."
   • "Buy when MACD crosses above signal line, sell when below."
   • "Go long when the 20-day SMA crosses above the 50-day EMA AND RSI(14) is below 60 AND the MACD line (12,26,9) is above the signal line. Exit the long position RSI(14) rises above 75 OR if the close price crosses below the lower Bollinger Band (20,2). "
3. Click Run Backtest.
4. Review metrics, summary stats, and interactive chart.

📂 Project Structure

.
├── assets/
│   └── ExampleImg.png
├── static/
│   ├── app.js
│   └── style.css
├── templates/
│   └── index.html
├── app.py
├── backtest.py
├── llm_parser.py
├── requirements.txt
└── README.md

🔮 Future Improvements

• Allow users to specify custom tickers and date ranges.
• Support advanced rules: stop-loss, take-profit, and trailing stops.
• Implement short-selling strategies.
• Save & compare backtest runs.
• Provide better error messages if the LLM cannot parse a strategy.

📄 License

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