We Tested NBA Referee Crews Across 20,238 Games. Here's What We Found.
By CW · October 19, 2026 · 2 min read
A lot of people believe that some NBA referee crews push games toward high scores and others toward low ones. We wanted to know if it's true before putting it in any of our numbers. So we tested it.
The short answer: in this data, no. We found nothing we could use.
What we tested We looked at 17 seasons (2003 through 2019) and 20,238 games. For every game we had the final score and the officiating crew, taken from public records compiled by the open‑source sportsdataverse project. Scores are measured against what the two teams normally score, not against the betting line. Crews change from game to game, so we scored each individual official and treated a crew as the officials working that night.
The question was simple: do games with the same crew tend to score higher or lower than you'd expect, and does that repeat?
That second part matters. Any crew will have a few high‑scoring games by luck. A real effect has to show up again in the next season.
How we tried not to fool ourselves We set the bar before looking at the result. A crew effect only counted as real if both of these held:
- An official’s results in one season had to predict its results in the next, with a correlation of at least 0.15. We measured 0.098 across 858 official-season pairs.
- Officials had to differ from each other by at least 0.75 points of total scoring once luck was removed. They differed by 0.72.
We shrank small samples toward the average, because a crew with 30 games and an extreme record is more likely lucky than special.
The result: it missed both bars. The difference between officials is real but tiny, about 0.7 points in a game whose total swings by about 18 points on its own, and it doesn't carry over from one season to the next. Our test says this is noise.
What this doesn't tell you We tested total points. We did not test foul counts, free throws, or individual player stats. Those might behave differently, and the public data we're allowed to use doesn't let us check them well. Our data stops after 2019, so it can’t speak to how crews have behaved in recent seasons.
We also can't say that no referee ever has any effect on any game. We can say that sorting games by crew doesn't give a usable edge on scoring.
What changes on BFN Nothing. Referee crews are not part of any pick or projection, and they won't be until something proves itself in a test like this one.
That is a rule we follow in general: an adjustment has to earn its place before it can change a number.
What we're doing next The NBA season opens on October 20. From opening night we log every game and check the officiating crew shortly before tip. As the sample grows, we'll run the same test again. If crews start to show a real, repeatable effect, we'll say so. If they don't, we'll say that too.
For entertainment only · 21+ · 1‑800‑MY‑RESET