Each team's chance to win
The test: the model was tuned on 2021-22 to 2024-25 and then run, game by game, through last season, 2025-26, which played no part in tuning it. Before each game it knew the results so far and who sat out each team's previous game; it did not have that season's injury reports, which are only partly saved, so it knew less than the live model does. Over 1310 games it gave the favorite 67% on average, and the favorite won 68% of the time.
| We gave the favorite | The favorite won | Games |
|---|---|---|
| 55% | 55% | 458 |
| 65% | 66% | 374 |
| 75% | 78% | 281 |
| 85% | 86% | 167 |
| 92% | 90% | 30 |
Every chance shown before tip-off since November 20, 2026 is scored after the game; 6 so far, too few to show.
Chances shown during games
Every chance shown while a game was on, scored against who won. A late lead makes these more certain, so there are more of them near 100%.
| We gave the favorite | The favorite won | Games |
|---|---|---|
| 55% | 51% | 39 |
| 65% | 77% | 22 |
| 75% | 67% | 36 |
| 85% | 39% | 46 |
| 97% | 63% | 168 |
Is he playing today?
The chance shown on each player's page is scored after every game. See today's chances.
Too early to judge: we started scoring on November 20, 2026, and 13 chances have been checked so far.
Do the injury status shares hold up?
The test: we took every team's shares from 2021-22 to 2024-25 and asked what they would have said about last season, 2025-26, which they never saw. These are the games from October 22, 2025 to December 19, 2025, before the league changed its injury report in December 2025, so the same rules applied. Each player's chance is his team's share at his status on its first listing, pulled toward the league's share until the team has enough listings of its own.
Listed probable: 137 listings
We would have said 87% on average; 90% played. Using each team's own share did better than the league's share alone.
| We would have said | They played | Listings |
|---|---|---|
| 78% | Too few | 7 |
| 86% | 90% | 96 |
| 92% | 97% | 34 |
Listed questionable: 453 listings
We would have said 54% on average; 55% played. Using each team's own share did better than the league's share alone.
| We would have said | They played | Listings |
|---|---|---|
| 39% | 47% | 17 |
| 46% | 50% | 107 |
| 54% | 56% | 245 |
| 65% | 61% | 84 |
Listed doubtful: 49 listings
We would have said 5% on average; 6% played. Using each team's own share did better than the league's share alone.
| We would have said | They played | Listings |
|---|---|---|
| 5% | 4% | 46 |
| 12% | Too few | 3 |
Rows with fewer than 10 players say "Too few". The 2025-26 reports are still being saved from the league's archive, so this check grows until it covers every game before the December change.
Playoff chances
The test: last season, 2025-26, simulated 1,000 times from its November 20 standings, knowing only the games played by then, and every team's playoff chance set against whether it made the playoffs. 2025-26 played no part in tuning the simulations, which were tuned the same way on 2021-22 to 2024-25.
Over 30 teams we said 53% on average, and 53% made it.
| We gave | On average | Made it | Teams |
|---|---|---|---|
| Under 30% | 7% | 0% | 9 |
| 30% to 70% | 54% | 60% | 10 |
| Over 70% | 91% | 91% | 11 |
This season's own check comes when the regular season ends in April 2027. How the simulations work.
Stats updated 10:41 am PT. Scores update every minute during games.