NBA player projections turn a player's recent minutes and per-minute production into one projected stat line for his next game. Our NBA player projections page does this for 6 stats, points, rebounds, assists, made 3s, steals and blocks, and gives each number a floor and a ceiling.
This guide explains how a projection is built in 4 steps, why it rarely matches a season average, and how to read the band around it. Every example comes from the board for Friday, November 20, 2026, read on the afternoon of the 20th: the 10 free rows, led by Luka Doncic of the Los Angeles Lakers at 35.9 projected points against the Sacramento Kings. The season and last-5 figures come from each player's game log on the same day. These are pre-game projections, not results.
What is an NBA player projection?
An NBA player projection is the model's average expectation for 1 player in 1 game, across 6 stats. It is not a season average and not a forecast of a best night. If the same game were played 100 times, the projection is roughly where the middle of those 100 results would land.
Each row on the player projections board names the player, the team, the opponent and the date, then lists the 6 projected stats with a low and a high. Each row links to the player's page, and each game has its own preview with team form, such as the Lakers vs Kings game preview. The game-level numbers, the projected margin and total, are on our NBA game projections.
How is a stat line projected from minutes and rates?
A stat line is projected in 4 steps: expected minutes, a per-minute rate, an adjustment for the matchup, and a correction for past error. The core idea is minutes multiplied by production per minute. A player who scores 1 point a minute and plays 36 minutes projects to about 36 points before any adjustment.
Step 1: minutes. Recent minutes are the strongest signal. The model reads each player's average minutes over his last 3 and last 5 games and his season average, weighted toward the most recent games. Step 2: rates. It reads points, rebounds, assists, 3s, steals and blocks over the last 3, 5 and 10 games and the season, plus usage rate over the same windows, so a player whose role has grown gets credit for it.
Step 3: the matchup. The opponent's points, rebounds and assists allowed, split by guards, forwards and centers, and both teams' pace adjust the line. A fast game means more possessions and more of every stat. Step 4: the error check. Once a game is final every line is graded, and if a player's projections have run high or low over enough games, his next line is shifted by that average miss, capped at 20% of the projection.
The detailed build, including the regression and tree models behind steps 2 and 3, is in our guide to how NBA projection models work. This article stays with what the numbers on the board mean.
A worked example: Luka Doncic's 35.9 points
Luka Doncic projects to 35.9 points against Sacramento because his minutes are steady and his scoring rate has cooled. Over 14 games he averages 38.1 points in 36.4 minutes, a rate of 1.05 points a minute. Over his last 5 games he averages 34.4 points in 37.3 minutes, a rate of 0.92.
Multiply his last-5 minutes by his season rate and you get 39.0 points (37.3 x 1.05). Use his last-5 rate instead and you get 34.4. The projection of 35.9 sits between the 2, closer to the recent number, which is what weighting toward recent games does. Divided by 37.3 minutes, the projection works out to 0.96 points a minute.
| Luka Doncic | Minutes | Points | Per minute |
|---|---|---|---|
| Season (14 games) | 36.4 | 38.1 | 1.05 |
| Last 10 games | 36.8 | 38.9 | 1.06 |
| Last 5 games | 37.3 | 34.4 | 0.92 |
| Projection, Nov 20 vs SAC | 37.3 (last 5) | 35.9 | 0.96 |
The same pattern shows up in his other lines. He averages 7.6 assists this season and projects to 7.7. He averages 6.9 rebounds but 5.4 over his last 5 games, and projects to 6.1. Every game behind these averages is on the Luka Doncic game log.
Why are projections lower than season averages?
All 10 free projections for November 20 sit below the player's season scoring average, by 0.1 to 3.0 points. A projection blends several windows, adjusts for the matchup and pulls back early-season numbers built on only 13 to 15 games, so it rarely equals the season average.
| Player (team v opponent) | Projected | Season | Last 5 |
|---|---|---|---|
| 1. Luka Doncic (LAL v SAC) | 35.9 | 38.1 | 34.4 |
| 2. Shai Gilgeous-Alexander (OKC v NOP) | 30.6 | 31.1 | 33.6 |
| 3. Franz Wagner (ORL v BKN) | 24.1 | 24.4 | 30.8 |
| 4. Darius Acuff Jr. (SAC v LAL) | 23.7 | 26.2 | 28.0 |
| 5. Devin Booker (PHX v UTA) | 23.5 | 24.8 | 22.6 |
| 6. Donovan Mitchell (CLE v IND) | 23.4 | 24.8 | 26.4 |
| 7. Nikola Jokic (DEN v HOU) | 23.1 | 24.1 | 26.0 |
| 8. Jamal Murray (DEN v HOU) | 22.7 | 25.7 | 28.0 |
| 9. Anthony Davis (WAS v CHI) | 22.1 | 24.6 | 22.6 |
| 10. Keyonte George (UTA v PHX) | 21.5 | 21.6 | 22.6 |
The biggest gap belongs to Jamal Murray of the Denver Nuggets: 22.7 projected against a 25.7 season average, even though he has scored 28.0 a game over his last 5. The smallest is Keyonte George, 21.5 against 21.6. Franz Wagner shows how far the model leans on longer windows: 30.8 points over his last 5 games, 24.4 for the season, and a projection of 24.1 at the Brooklyn Nets.
Projections go above averages too when the inputs point that way. Anthony Davis projects to 10.5 rebounds against the Chicago Bulls, above his 10.1 season average, and Shai Gilgeous-Alexander projects to 5.1 assists against the New Orleans Pelicans, above his 4.3.
What do the floor and ceiling mean?
The floor and ceiling bound where 1 game's result should land most of the time, and on the November 20 board they sit about 11 to 14 points either side of a points projection. Doncic's 35.9 runs from 21.7 to 50.1, a band 28.4 points wide. Keyonte George's 21.5 runs from 10.5 to 32.5, 22.0 points wide.
| Player and stat | Projection | Floor | Ceiling |
|---|---|---|---|
| Doncic points | 35.9 | 21.7 | 50.1 |
| Gilgeous-Alexander points | 30.6 | 17.5 | 43.8 |
| Jokic rebounds | 10.3 | 5.3 | 15.2 |
| Jokic assists | 7.9 | 3.8 | 12.0 |
| Acuff Jr. made 3s | 3.2 | 0.6 | 5.8 |
| Doncic steals | 1.9 | 0 | 4 |
Wide bands are normal. A 1-game result swings with foul trouble, a blowout that sends starters to the bench after 3 quarters, or a cold shooting night. Steals and blocks are low-count stats, so their bands are wide next to the number: Doncic's 1.9 steals can land anywhere from 0 to 4. When a projection sits close to a number you care about, read the band, not just the middle.
How should you use NBA player projections?
Use NBA player projections in 3 ways: to compare players on 1 night, to plan fantasy lineups, and to spot role changes early. The projections board sorts by any of the 6 stats, so the top scorers, rebounders or passers for a slate take 1 click to find.
For fantasy, a projection is the single-night number that season-long fantasy basketball rankings cannot give you, and it pairs with the salaries on our NBA DFS page. For points, rebounds and assists combined, add the 3 lines: Jokic's 23.1, 10.3 and 7.9 make 41.3, the measure explained in our guide to PRA in basketball.
Check the news before you trust a line. A player ruled out is removed from the board rather than shown at 0, but a late scratch or a teammate's absence can change minutes after the board was built, so read the NBA injury report and the NBA starting lineups close to tip-off. To see how a whole game might play out from these numbers, run it through the NBA game simulator, or open the NBA projections hub for every board in 1 place.
Where the November 20 numbers come from
The projections are the 10 free rows of the Statz NBA player projections for Friday, November 20, 2026, read that afternoon; those 10 rows cover 7 of the night's games. The board covered 119 players across 7 slates from November 20 to 27. Season, last-10 and last-5 averages come from each player's game log on the same day, for example the Nikola Jokic game log. Projections rebuild daily, so the board you open later will show newer numbers.