Summary

georgedouzas/sports-betting provides a Python-based toolkit for creating, testing, and deploying sports betting models. It targets AI agents, Python developers, and researchers by offering modular components for data acquisition and strategy backtesting, enabling systematic model development and prediction. The library aims to empower users to build and validate data-driven betting strategies.

How It Works

The core architecture comprises dataloaders and bettors. Dataloaders abstract data acquisition, combining statistics and odds sources to shape data for modeling, ensuring data provenance. Bettors encapsulate any scikit-learn estimator, facilitating backtesting of betting strategies against historical odds and predicting value bets for upcoming events. This approach allows for flexible integration with existing machine learning workflows and custom data pipelines.

Quick Start & Requirements

• Installation:
  • User: pip install sports_betting
  • AI Agent (MCP Server): pip install 'sports_betting[mcp]'
  • Development: Clone repository and run pdm install.
• Prerequisites: Requires Python. Development installation uses PDM. No specific hardware (GPU, CUDA) or OS dependencies are listed.
• Documentation: Quick start examples and API usage are detailed within the README.

Highlighted Details

• Data Breadth: Access to 918 league, division, and season combinations across 27 leagues (1994-2026), including historical odds from multiple bookmakers.
• Market Analysis: Demonstrates identifying profitable betting markets, such as the 'draw' market, with backtesting yielding positive results.
• Model Integration: Seamlessly wraps scikit-learn estimators, allowing for hyperparameter tuning and custom model pipelines.
• Specific Edge Discovery: Highlights a significant draw edge found predominantly in Italian leagues (+9.0% yield), though cautioned as potentially susceptible to search bias.
• Data Pipeline: Training and fixture data share column structures, enabling models trained on historical data to predict upcoming matches.

Maintenance & Community

No specific details regarding maintainers, community channels (e.g., Discord, Slack), or roadmap were found in the provided README content.

Licensing & Compatibility

The README does not specify a software license. Consequently, compatibility for commercial use or closed-source linking cannot be determined from the provided text.

Limitations & Caveats

The effectiveness of discovered strategies may be influenced by extensive model searching, necessitating careful validation. Users are advised to hold back a season for independent testing before deploying real capital. The library's data extraction process consumes paid odds feed resources, suggesting efficient data management practices. Certain profitable edges, like the Italian draw market, are flagged with suspicion due to the discovery methodology. Fixture data extraction may yield no results during league off-seasons.