NBA Player Props Prediction Model

A machine learning system for predicting NBA player prop bets and identifying edges against sportsbook lines.

Features

• Historical NBA stats collection via NBA API
• Advanced feature engineering including:
  • Rolling averages across multiple windows
  • Home/Away splits
  • Opponent strength metrics
  • Player trend analysis
• Ridge regression model for prop predictions
• Edge calculation against market lines
• Kelly criterion bet sizing
• DuckDB + MotherDuck for efficient data storage

Project Structure

plus-ev-model/
├── data/                  # Data storage
│   ├── raw/              # Raw API data
│   ├── processed/        # Cleaned data
│   ├── interim/          # Intermediate processing
│   └── external/         # Third-party data
├── docs/                 # Documentation
├── src/                  # Source code
│   ├── data/            # Data collection
│   │   ├── nba_stats.py           # NBA API data pipeline
│   │   ├── nba_stats_career.py    # Career stats collection
│   │   └── db_config.py           # Database configuration
│   ├── models/          # ML models
│   │   ├── player_props_model.py  # Base props prediction model
│   │   └── predict_props.py       # Prediction script
│   └── core/            # Core utilities
│       ├── devig.py               # Odds processing
│       ├── monte_carlo.py         # Simulation tools
│       └── edge_calculator.py     # Betting edge analysis
└── tests/               # Test suite

Setup

1. Install dependencies:

poetry install

2. Set up environment variables in .env:

LOCAL_DB_PATH=data/nba_stats.duckdb
MOTHERDUCK_TOKEN=your_token_here
NBA_API_DELAY=1.0

3. Initialize database:

python -m src.data.nba_stats

Usage

Fetch NBA Stats

from src.data.nba_stats import update_player_stats

# Update stats in database
update_player_stats()

Predict Player Props

from src.models.predict_props import analyze_player_prop

# Analyze a single prop
result = analyze_player_prop(
    player_id=2544,  # LeBron James
    prop_type='points',
    line=25.5,
    over_odds=-110,
    under_odds=-110
)

# Print recommendation
if result['recommendation']:
    rec = result['recommendation']
    print(f"Bet: {rec['bet_type']} {rec['line']}")
    print(f"Edge: {rec['edge']:.1%}")
    print(f"Kelly: {rec['kelly_bet']:.1%}")

Analyze Multiple Props

from src.models.predict_props import analyze_multiple_props

props = [
    {
        'player_id': 2544,
        'prop_type': 'points',
        'line': 25.5
    },
    {
        'player_id': 2544,
        'prop_type': 'assists',
        'line': 7.5
    }
]

results = analyze_multiple_props(props)

Model Details

Feature Engineering

The model uses several types of features:

• Rolling averages (5, 10, 20 game windows)
• Home/Away performance splits
• Opponent strength indicators
• Recent performance trends
• Player consistency metrics

Edge Calculation

Edges are calculated by:

1. Converting model predictions to probabilities
2. Comparing to market implied probabilities
3. Applying Kelly criterion for optimal bet sizing
4. Filtering by confidence threshold

Performance Metrics

The model is evaluated on:

• Root Mean Squared Error (RMSE)
• Mean Absolute Error (MAE)
• R-squared (R2)
• Prediction vs. Actual correlation
• ROI on historical bets

Contributing

1. Fork the repository
2. Create a feature branch
3. Commit changes
4. Push to the branch
5. Create a pull request

License

MIT License - see LICENSE file for details