BBallBot

A machine learning system for basketball betting predictions using historical odds data.

Project Structure

• main.py - Main entry point for training and using models
• config.py - Configuration constants and file paths
• data_processor.py - Data loading and preprocessing functions
• model_trainer.py - Model training and optimization using Optuna
• betting.py - Betting logic and profit calculations
• utils.py - Utility functions for data manipulation
• odds_data/ - Directory containing historical odds data CSV files

Components

Data Processing

• Team indexing and win/loss matrix creation
• Feature preparation using NMF (Non-negative Matrix Factorization)
• Data splitting for training and testing

Model Training

• Neural network model using scikit-learn's MLPClassifier
• Hyperparameter optimization with Optuna
• Model evaluation using accuracy and profit metrics

Betting Logic

• Kelly Criterion for bet sizing
• Profit calculation and tracking
• Implied probability calculations from odds

Usage

Training a New Model

python main.py --year 2022 --train

This will:

1. Load odds data for the specified year
2. Train a model using Optuna for hyperparameter optimization
3. Save the best model and its parameters
4. Print accuracy and profit metrics

Loading a Trained Model

python main.py --year 2022

This will load a previously trained model for the specified year.

Requirements

• Python 3.8+
• NumPy
• Pandas
• scikit-learn
• Optuna
• joblib

Data Format

The odds data CSV files should have the following columns:

• Home Team
• Away Team
• Home Score
• Away Score
• Home Odds
• Away Odds