Trading Algorithm Backtester Pro

A modern, elegant desktop application for backtesting predictive Python functions on historical prices with advanced account simulation and confidence-based trading strategies.

Features

• Modern GUI: Elegant dark-themed interface built with PySide6.
• Real-time Visualization: Dynamic charts using pyqtgraph that update as the backtest progresses.
• Multi-Timeframe Support: Backtest across 1m, 5m, 15m, 30m, 1h, and 1d timeframes simultaneously.
• Integrated Code Editor: Syntax-highlighted editor with validation and history tracking.
• Multithreaded Engine: High-performance backtesting that keeps the UI responsive.
• Data Management: Efficient loading of large CSV datasets with Feather format conversion for speed.
• Customizable Logic: User-defined success thresholds and auto-abort parameters.
• Futures Account Simulation: Complete trading account simulation with:
  • Configurable starting capital and leverage (1x-50x)
  • Dynamic position sizing based on confidence levels
  • Maximum position value limits with optional forced liquidation
  • Real-time P&L and account value tracking
  • Auto-abort when account falls below contract cost
• Confidence-Based Trading: Predictions include confidence levels that drive position sizing:
  • 90%+ confidence: 100% of available capital
  • 80-90%: 75% of capital
  • 70-80%: 50% of capital
  • 60-70%: 25% of capital
  • Below 60%: No position
• Prediction & Confidence Tuning: Adjust prediction values and confidence levels with real-time sliders for strategy optimization.
• Feedback-Adjusted Prediction Model: The default prediction_v1.py uses self-correcting linear regression — the engine passes its own last 20 predictions back into the function so it can compute and compensate for systematic bias.
• Extended Historical Context: Uses up to 100 historical Close prices for more accurate predictions.
• Detailed Run Statistics: Tracks trades entered, failed predictions, and total timespan covered per backtest run.

Installation

1. Clone the repository:

git clone https://github.com/trading-algorithm-backtester-simulator.git
cd trading-algorithm-backtester-simulator

1. Install dependencies:

pip install pyside6 pyqtgraph pandas numpy pygments feather-format

1. Ensure a historical price CSV-formatted OHLCV data file exists in data.

Usage

1. Run the application:

python main.py

1. Configure account parameters:

   • Set starting capital (default: $10,000)
   • Adjust leverage (1x-50x, default: 12x)
   • Optionally set maximum position value

2. Select the desired timeframes in the sidebar.

3. Write or edit your predictive function in the editor. The function must follow the signature:

   def predict(close_prices, recent_predicted_prices):
         # close_prices: list of up to 100 recent Close values (most recent last)
         # recent_predicted_prices: list of the last 20 values returned by this function
         # Must return a tuple: (predicted_price, confidence_percentage)
         # Example:
         predicted_price = 1950.50
         confidence = 75  # 0-100
         return predicted_price, confidence

4. Adjust thresholds and offsets as needed:

   • Up/Down thresholds: Define success criteria for predictions
   • Auto-abort threshold: Stop if prediction error exceeds this value
   • Prediction offset: Add/subtract a fixed amount to all predictions
   • Confidence offset: Adjust confidence levels up or down

5. Click Update / Validate to check for syntax errors.

6. Click Start Backtest to begin. Watch real-time updates of:

   • Price predictions vs actual values
   • Account value and invested capital
   • Success rates across timeframes

7. Use the Stop button to halt a running backtest.

UI Controls

• Start Backtest: Enabled only after successful code validation.
• Stop: Enabled only while a backtest is in progress.
• Update / Validate: Compiles the current code and saves it to history if changed.
• Backtest History: Click any revision to revert the editor to that version.
• Clear All History: Resets the history list to the original v1 function (with confirmation).
• Prediction Offset Slider: Adjust predictions by -10.0 to +10.0 points.
• Confidence Offset Slider: Adjust confidence levels by -80 to +80 percentage points.
• Sell Above Max Checkbox: Automatically reduce positions that exceed maximum position value.

Project Structure

• main.py: Main application entry point and UI logic with account simulation.
• data_engine.py: Data loading and resampling engine.
• highlighter.py: Syntax highlighter for the code editor.
• prediction_v1.py: Feedback-adjusted linear regression prediction model (default starting template).
• get_last_30_closes.py: Utility script to fetch recent closing prices for any timeframe.
• data/: Directory for historical datasets (CSV and Feather formats).
• PLAN.md: Development roadmap and feature planning.
• CHANGELOG.md: Detailed version history.