↗
Build a better baseline
Use team ratings and game-level statistics to estimate the margin between two teams, rather than relying on rankings alone.
Inside the model ↗COLLEGE BASKETBALL SPREAD PREDICTIONS
Hoops, to the decimal.
Every game has an edge. We find it. College basketball spread predictions with the line, the pick, and the model’s margin—without promises or hype.
EVERY GAME HAS AN EDGE. WE FIND IT.EST. IN 2026 ●
01 / THE IDEA
More than a
gut feeling.
College basketball is unpredictable. The point isn't to pretend otherwise. It's to use data, test assumptions, and find out where a simple model gets things right, and where it falls apart.
⌁
Make the picks public
Put predictions in one place, then track them over time. A forecast is only useful if you can come back and grade it.
View predictions ↗◎
Learn from the misses
Review surprising outcomes, model errors, and the assumptions worth revisiting. The process matters as much as the picks.
Read the notebook ↗02 / THE PROCESS
From raw data
to a final score.
The project connects a data pipeline, feature engineering, a regression model, and a dashboard. The goal is a repeatable process that can make predictions before games tip off, then evaluate those predictions against what actually happened.
- ✓ Collect and prepare team and game data
- ✓ Train a model to estimate home-team margin
- ✓ Publish predictions and evaluate results
THE SEASON IS A DATASET
Every game tells
a story.
Follow the predictions, read the post-game breakdowns, and see what the model learns along the way.
See the picks ↗