GSGhost Screen Analytics

2026-27 Season Forecast

We play out the rest of the season a million times every morning: every game, every injury, the play-in, the playoffs, the lottery and the awards. Every number here is the share of those seasons in which it happened.

Loading…

Standings and odds

Seeds 1-6 go straight to the playoffs; 7-10 play in. Click a team to see its full range of outcomes and rotation. Columns sort when clicked.

Where we disagree with the model

The model only goes where its tests point. These are our own calls, set by hand, next to what the model says.

Surprise factor

Some teams are harder to call than others. Teams bringing back fewer of last season's minutes, and teams leaning on rookies, land further from their projections. The schedule column is how many wins each team's remaining schedule adds or costs, from rest, back-to-backs, travel and altitude. Breakout is the chance a team beats its projection by 8 or more wins. Collapse is the chance it falls 8 or more short.

Injuries in the forecast

From the current injury report. A listed player misses every game his team plays before his expected return date, then plays his normal share of games. Return dates are estimates and change often. The forecast picks up every change the next morning.

Title odds

Champion
Likeliest Finals

Awards

Each simulated season produces a full stat line for every player, games played included. A voter model built from 2010-11 to 2025-26 ballots picks the winner. Players need 65 games to qualify, so a missed month can cost an award. Award odds use 100,000 of the simulated seasons.

The 65-game watch

Chance each of the league's best players reaches 65 games, the minimum for MVP, Defensive Player of the Year, All-NBA and the other major awards.

Upcoming games

Click a game to see why the model leans the way it does.

Team by team

Our read on every team before the season: last year's results, who came and went, and what the model sees. Ratings are points per 100 possessions above an average player.

How the odds have moved

Player ratings

Points per 100 possessions a player adds to his team's margin, compared with an average player, split into offense and defense. It comes from every lineup since 2010-11 and is anchored by box score stats. These ratings drive the team strengths above.

How it works

Players. Each player's rating comes from every lineup he has played in over the last three seasons, the most recent weighted most. It is anchored to his box score stats, then adjusted for age. A rookie starts from what players taken in his draft slot have done in their first year.

Teams. A team's strength is its players' ratings weighted by expected minutes. We tested whether lowering a team's strength for injury-prone players helped. It didn't (test S2), so it isn't used. Injury luck is simulated instead: in each season some players stay healthy and some don't, around each player's own track record.

Schedule. Every game is adjusted for rest (the second night of a back-to-back costs about 1.9 points), a third game in four nights, time zones crossed, and the altitude in Denver and Utah (test S1).

Surprises. Each simulated season gives every team a luck draw. The size of that draw depends on how many minutes are returning, roster age and rookie minutes (test S3).

The last 15 games. Teams already out of the race played about 3 points worse than their strength over their last 15 games, and teams with their seed locked about 2.4 points worse (test S4). The simulation applies both based on where each team stands when its last 15 begin.

Seasons. In each of the 1,000,000 simulations, every player gets an availability draw (some seasons he stays healthy, some he doesn't) and every team gets a luck draw. Then every remaining game is played: home court is worth about 2.2 points, and a typical game swings about 13 points either way. Ties in the standings are broken by point differential, a simplification of the league's tiebreakers. The play-in and every playoff round are played out game by game, home court included. The lottery uses the league's odds for the #1 pick.

During the season. Results move a team's strength. After 20 games a team is judged half on its preseason rating and half on how it has actually played (adjusted for opponents and home court). After 60 games, mostly on results. Box scores update minutes, team changes and games played. The current injury report sets who is out and for how long: a listed player misses every game his team plays before his expected return date.

Test record

Every bar below was written down before the test was run and is in FORECAST_PREREG.txt. Each season was predicted using only the seasons before it. Failed tests stay on the page.

TestQuestionBarResultVerdict

What it can't see