Sports Edge

Sports Edge is my sports analytics and prediction project. It started with a simple NBA question: could I turn public sports data into repeatable, testable predictions?

The journey

The first version was a set of notebooks and small scripts. I used them to explore team ratings, game data, and basic models. The project then grew in stages:

1. NBA experiments became scheduled prediction jobs.
2. BigQuery became the warehouse for raw data, features, predictions, and results.
3. Supabase became the serving layer for the portfolio and dashboard.
4. GitHub Actions and Google Cloud Scheduler made refreshes repeatable.
5. The system expanded to NFL, MLB, MLB home-run markets, PGA, college basketball research, and World Cup forecasting.
6. A Next.js dashboard turned the outputs into a product that is easier to inspect.

The goal is not only to make a prediction. The goal is to show the full path from raw data to a monitored public result. Each sport has a different level of maturity. NBA is the most active surface. Some leagues are still research or candidate work.

What it is now

Sports Edge runs Python pipelines that collect data and odds, build features, run models, and write results to BigQuery. Selected results move to Supabase. The Next.js dashboard and my portfolio read from that serving layer.

Built with

• Python
• pandas, NumPy, scikit-learn, LightGBM, XGBoost, and PyTorch
• Google BigQuery
• Supabase and Postgres
• Next.js, React, TypeScript, and Tailwind CSS
• GitHub Actions, Google Cloud Scheduler, and Cloud Run

Next

The next phase is about clearer model versioning, stronger evaluation, and a better view of performance across leagues. The system is useful as a working product, but it is still an active research project.

Contact

• Email: [email protected]
• LinkedIn: Drew Boynton