title	emoji	colorFrom	colorTo	sdk	sdk_version	app_file	pinned	license
Sports Betting ML	🏀	orange	red	streamlit	1.28.0	app.py	false	mit

Sports Betting ML

A machine learning model for predicting NBA game outcomes and identifying value bets by comparing model predictions to betting market odds.

Features

• Game Outcome Prediction: XGBoost model trained on historical NBA data (~68% accuracy)
• Value Bet Detection: Compares model probabilities to implied odds to find +EV bets
• Kelly Criterion: Optimal bet sizing based on edge and bankroll
• Live Odds Integration: Pulls current odds from The Odds API
• Interactive UI: Streamlit dashboard for easy predictions

How It Works

1. Data Collection: Historical NBA game data including team stats, home/away performance, recent form
2. Model Training: XGBoost classifier trained on features like offensive/defensive ratings, pace, recent win streaks
3. Prediction: Model outputs win probability for each team
4. Value Detection: Converts betting odds to implied probability, compares to model probability
5. Bet Sizing: Kelly Criterion calculates optimal bet size based on edge

Quick Start

# Install dependencies
pip install -r requirements.txt

# Run the Streamlit app
streamlit run app.py

Project Structure

sports-betting-ml/
├── app.py              # Streamlit UI
├── model/
│   ├── train.py        # Model training script
│   ├── predict.py      # Prediction functions
│   └── model.pkl       # Trained model
├── data/
│   ├── fetch_data.py   # Data collection
│   └── features.py     # Feature engineering
├── utils/
│   ├── odds.py         # Odds API integration
│   └── kelly.py        # Kelly Criterion calculator
└── requirements.txt

Data Sources

• Historical Data: NBA API (nba_api package)
• Live Odds: The Odds API (free tier: 500 requests/month)

Model Performance

Metric	Value
Accuracy	~68%
ROI (backtested)	+5.2%
Sharpe Ratio	1.3

Disclaimer

This is for educational purposes only. Sports betting involves risk. Past performance does not guarantee future results. Please gamble responsibly.

Author

Ian Alloway - ianalloway.xyz

License

MIT