STRANOVA is SPIN's computer rating system. It turns the gap between two teams into a win chance for each side. We publish that number before kickoff, we freeze it, and we post the final next to it afterward. This page is the running record. It updates after every round, and every STRANOVA card and social post links back here.
Round 1: Utah quarterfinals, Friday, Nov. 6
The numbers below were locked on Nov. 5 at 7:51 a.m. MT using model version 0.4. Results post Saturday morning.
| Class | Game | Pre-game win chance | Final | Result |
|---|---|---|---|---|
| 6A | American Fork at Skyridge | Skyridge 81%, American Fork 19% | Pending | Pending |
| 6A | Davis at Herriman | Herriman 57%, Davis 43% | Pending | Pending |
| 6A | Lone Peak at Mountain Ridge | Mountain Ridge 90%, Lone Peak 10% | Pending | Pending |
| 6A | Syracuse at Corner Canyon | Corner Canyon 99%, Syracuse 1% | Pending | Pending |
| 5A | Fremont at West | West 96%, Fremont 4% | Pending | Pending |
| 5A | Springville at Bountiful | Springville 52%, Bountiful 48% | Pending | Pending |
| 5A | Pleasant Grove at Orem | Orem 94%, Pleasant Grove 6% | Pending | Pending |
| 5A | Woods Cross at Alta | Alta 90%, Woods Cross 10% | Pending | Pending |
| 4A | Highland at Ridgeline | Ridgeline 93%, Highland 7% | Pending | Pending |
| 4A | Crimson Cliffs at Snow Canyon | Crimson Cliffs 83%, Snow Canyon 17% | Pending | Pending |
| 4A | Salem Hills at Hurricane | Hurricane 91%, Salem Hills 9% | Pending | Pending |
| 4A | Provo at Skyline | Skyline 76%, Provo 24% | Pending | Pending |
The running tally
| Measure | Where it stands |
|---|---|
| Favorites that won | 0 of 0 so far. STRANOVA's own forecast for Friday is about 10 of 12. |
| Calibration at 90% or higher | No games graded yet |
| Calibration at 75 to 89% | No games graded yet |
| Calibration at 60 to 74% | No games graded yet |
| Calibration at 50 to 59% | No games graded yet |
| Always pick the home team | No games graded yet |
| Always pick the higher seed | No games graded yet |
| Brier score and log loss | No games graded yet |
The rules we set before any result existed
- Favorites that won: how many favorites won, running total each round.
- Calibration: how often teams at 90% or higher, 75 to 89, 60 to 74 and 50 to 59 actually win.
- Projected scores: the cards also show a projected score. In testing, those projections missed by about 9 points a team on average. They are a rough picture of the game, not a prediction of the fourth quarter.
- Baselines beside it: always going with the home team and always going with the higher seed, so you can see what the model adds.
- Brier score and log loss, for the people who will check the math.
- Sample size: 12 games say very little. Judge it at 100.
A miss is called a miss. A 57% call that loses was still 43% right, so we will not say the model was wrong about it. If we find a data error after the freeze, we publish a correction with the recomputed number and leave the frozen card exactly as it ran.
The fingerprint
The record of the frozen Round 1 numbers has this fingerprint:
`d1903400887fd7317e688233db7c8b09b7128ebab131bd9a74334eb7ad5d01fe`
If a single number changed after kickoff, the fingerprint would not match.
STRANOVA rates teams and their results. It does not rate or grade any player, and it is not a betting product. Not for wagering.
