How CourtSide Works
Everything you need to know about how CourtSide ranks games and tells you which one is worth watching right now.
What is the Watch Score?
The Watch Score is a 0–100 rating of how worth-watching a game is right now. A score of 90+ means drop everything and tune in. Below 40 usually means a blowout or a matchup with nothing on the line.
Live game scores update every 30 seconds. Pregame scores are predictions based on team quality and historical patterns. Game summaries update dynamically — during blowouts they refresh every 10 minutes, but as games tighten and the clock winds down, updates accelerate, reaching every 20–30 seconds during a close game in the final two minutes. The XCite Score drives the update frequency automatically.
How is a live game scored?
Six factors combine to produce the final score. Each carries a different weight:
| Factor | Weight | What it measures |
|---|---|---|
| Closeness | 28% | How tight is the margin? A game within 6 points in the final 5 minutes scores highest. |
| Time Remaining | 22% | Late-game moments matter more. A 4-point game with 2 minutes left beats the same margin at halftime. |
| Lead Changes | 15% | Back-and-forth action. Frequent lead changes in a half signal a true battle. |
| Upset Likelihood | 15% | Is a heavy underdog hanging around? When a big favorite is at risk, stakes spike. |
| Ranked Stakes | 12% | Are AP Top 25 teams involved? Ranked matchups draw more eyeballs for a reason. |
| Tournament Implications | 8% | Does this game affect a team's bubble status or conference title chase? |
How are pregame scores predicted?
Before tip-off, we use a machine learning model trained on 63,813 historical college basketball games to predict how exciting a matchup will be.
The single biggest predictor: how evenly matched the teams are. Tight efficiency gaps between offense and defense produce the most exciting games.
Projected final scores are calculated using each team's adjusted offensive efficiency, defensive efficiency, and tempo — the same methodology used by KenPom.
What are the factor chips (↑ / ↓)?
The small chips on pregame cards explain why a game got its score. Each chip is powered by SHAP values — a technique from machine learning that breaks down exactly how much each factor contributed.
- ↑ chip — this factor is pushing the excitement score higher (e.g., “Close Matchup ↑”)
- ↓ chip — this factor is dragging the score down (e.g., “Low Rankings ↓”)
The top 2 contributing factors are shown on each card.
Where does the data come from?
- ESPN — live scores, period, clock, national rankings, TV network, and betting odds
- BartTorvik — team efficiency ratings (adjusted offense, defense, tempo) and win-loss records
- Haslametrics — supplemental team strength ratings
Live data refreshes every 30 seconds. Advanced stats are ingested daily.
Why doesn't every game have a predicted score?
Advanced stats (efficiency ratings, tempo) are available for roughly 64% of Division I programs in our database. Most Power 6 and mid-major conferences are covered. Some low-major programs are missing — those games show a basic watch score without a predicted final or thrill rating.