Cryptocurrency Linear Regression Trading System

A comprehensive cryptocurrency trading system that uses linear regression models to predict price movements and execute trades based on technical indicators and market data from Coinbase.

🚀 Live Demo: Generate real crypto price predictions using machine learning!

Quick Start

git clone https://github.com/Deftextra/crypto_trader.git
cd crypto_trader
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
python main.py --mode predict --symbol BTC-USD

Features

• Data Collection: Fetches real-time and historical cryptocurrency data from Coinbase Pro API
• Feature Engineering: Creates 100+ technical indicators and features from raw price data
• Machine Learning: Implements multiple linear regression models with feature selection
• Trading Strategy: Executes trades based on model predictions with risk management
• Backtesting: Comprehensive backtesting framework with performance analytics
• Visualization: Creates detailed performance reports and charts
• Risk Management: Position sizing, stop-loss, take-profit, and drawdown controls

Project Structure

crypto_trader/
├── data/
│   ├── __init__.py
│   └── collector.py          # Data collection from Coinbase API
├── utils/
│   ├── __init__.py
│   └── features.py           # Feature engineering and technical indicators
├── models/
│   ├── __init__.py
│   ├── linear_model.py       # Linear regression models
│   └── saved/                # Directory for saved models
├── strategies/
│   ├── __init__.py
│   └── linear_strategy.py    # Trading strategy implementation
├── backtesting/
│   ├── __init__.py
│   └── backtest.py           # Backtesting framework
├── output/                   # Generated visualizations and reports
├── main.py                   # Main application entry point
├── requirements.txt          # Python dependencies
└── README.md                 # This file

Installation

1. Clone or create the project directory:

mkdir crypto_trader
cd crypto_trader

2. Create a virtual environment:

python3 -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

Usage

The system supports four main modes of operation:

1. Training Mode

Train a model on historical data and save it for later use:

python main.py --mode train --symbol BTC-USD --days 90

2. Backtesting Mode

Run a complete backtest on historical data:

python main.py --mode backtest --symbol BTC-USD --days 60 --save-viz

Options:

• --save-viz: Save performance visualizations
• --output-dir: Specify output directory for results

3. Prediction Mode

Generate price predictions for current market conditions:

python main.py --mode predict --symbol BTC-USD

Or use a previously saved model:

python main.py --mode predict --symbol BTC-USD --model-path models/saved/BTC-USD_20240115_143022.joblib

4. Model Comparison Mode

Compare different model configurations:

python main.py --mode compare --symbol BTC-USD --days 60

Supported Cryptocurrencies

The system works with any cryptocurrency pair available on Coinbase Pro. Popular pairs include:

• BTC-USD (Bitcoin)
• ETH-USD (Ethereum)
• ADA-USD (Cardano)
• DOT-USD (Polkadot)
• LINK-USD (Chainlink)
• LTC-USD (Litecoin)
• XLM-USD (Stellar)
• ALGO-USD (Algorand)

Configuration

The system uses default configuration parameters that can be customized by modifying the _get_default_config() method in main.py:

config = {
    # Data parameters
    'symbol': 'BTC-USD',
    'days': 60,
    'granularity': 3600,  # 1 hour intervals
    
    # Model parameters
    'model_type': 'ridge',        # 'linear', 'ridge', 'lasso', 'elastic'
    'feature_selection': 'kbest', # 'none', 'kbest', 'rfe'
    'n_features': 15,
    'target_hours': 1,            # Prediction horizon
    
    # Strategy parameters
    'prediction_threshold': 0.5,   # Minimum prediction confidence (%)
    'lookback_periods': 5,
    
    # Risk management
    'initial_capital': 10000,
    'max_position_size': 0.1,     # 10% of portfolio
    'stop_loss_pct': 0.02,        # 2% stop loss
    'take_profit_pct': 0.04,      # 4% take profit
    'max_drawdown_pct': 0.1,      # 10% max drawdown
    'min_prediction_confidence': 0.01,
    
    # Backtesting
    'commission': 0.001,          # 0.1% commission
    'slippage': 0.001             # 0.1% slippage
}

Features Generated

The system automatically generates 100+ features from raw OHLCV data:

Technical Indicators

• Moving Averages (SMA, EMA) with multiple periods
• Bollinger Bands
• RSI (Relative Strength Index)
• MACD (Moving Average Convergence Divergence)
• Stochastic Oscillator
• Average True Range (ATR)
• Williams %R

Price Features

• Price changes and returns
• High-low percentage ranges
• Volume-weighted average price (VWAP)
• Price position within daily range

Statistical Features

• Rolling means, standard deviations, min/max
• Volatility measures
• Skewness of price movements

Lagged Features

• Historical values of key indicators
• Multiple lookback periods

Time Features

• Hour of day, day of week, month
• Cyclical encoding of time features

Model Types

The system supports multiple linear regression variants:

1. Linear Regression: Basic linear regression
2. Ridge Regression: L2 regularization for feature stability
3. Lasso Regression: L1 regularization for feature selection
4. Elastic Net: Combined L1/L2 regularization

Risk Management

The system includes comprehensive risk management:

• Position Sizing: Based on portfolio value and prediction confidence
• Stop Loss: Automatic loss cutting at specified percentage
• Take Profit: Profit taking at target levels
• Maximum Drawdown: Trading halt if losses exceed threshold
• Confidence Threshold: Only trade on high-confidence predictions

Performance Metrics

Backtesting provides detailed performance analytics:

Returns

• Total return, annual return
• Risk-adjusted returns (Sharpe, Sortino, Calmar ratios)

Risk Metrics

• Maximum drawdown
• Volatility
• Value at Risk (VaR)
• Conditional VaR

Trading Metrics

• Total trades, win rate
• Average win/loss
• Profit factor
• Trade duration statistics

Visualization

The system generates comprehensive performance reports including:

• Portfolio value over time
• Drawdown charts
• Returns distribution
• Price charts with trading signals
• Rolling performance metrics

Example Output

==================================================
BACKTEST RESULTS
==================================================
Symbol: BTC-USD
Period: 60 days
Initial Capital: $10,000.00
Final Value: $10,847.32
Total Return: 8.47%
Annual Return: 51.82%
Volatility: 45.23%
Sharpe Ratio: 1.15
Max Drawdown: -12.34%
Total Trades: 23
Win Rate: 60.87%
Profit Factor: 1.67

Limitations and Disclaimers

⚠️ Important Disclaimers:

1. Educational Purpose: This system is for educational and research purposes only
2. No Financial Advice: Not intended as financial advice or recommendations
3. Past Performance: Past performance does not guarantee future results
4. Market Risk: Cryptocurrency trading involves substantial risk of loss
5. No Warranty: The system is provided "as-is" without warranties

API Rate Limits

The system respects Coinbase Pro API rate limits by:

• Adding delays between requests
• Using reasonable request frequencies
• Caching data when possible

Contributing

To contribute to this project:

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests if applicable
5. Submit a pull request

License

This project is provided for educational purposes. Use at your own risk.

Support

For issues or questions:

1. Check the logs in crypto_trader.log
2. Review the error messages
3. Ensure all dependencies are installed
4. Verify API connectivity

Changelog

Version 1.0.0

• Initial release
• Basic linear regression models
• Coinbase data integration
• Comprehensive backtesting
• Risk management system
• Performance visualization