Most NBA matchups have no usable history
Of 21,324 scorer and defender pairings in the tracked NBA data, 87% have fewer than 15 matchup possessions and 44.9% have fewer than five. The median pairing has been observed in exactly one game. Only 390 pairings, about 1.8%, clear three meetings and 25 possessions. For almost any specific matchup someone wants to look up, the honest answer is that there is not enough history to say anything.
5 min read · Data as of 7 June 2026 · Analysis 27 July 2026
The whole distribution, not the headline
Matchup datasets are usually described by their row count, which sounds impressive and tells you nothing about whether any individual matchup is knowable. The number that matters is how those rows are distributed across pairings, because that is what determines whether the pairing you care about has any evidence behind it.
| Matchup possessions | Pairings | Share | Avg meetings |
|---|---|---|---|
| Under 5 | 9,567 | 44.9% | 1.0 |
| 5 to 15 | 8,967 | 42.1% | 1.2 |
| 15 to 25 | 1,764 | 8.3% | 1.5 |
| 25 to 50 | 857 | 4.0% | 2.0 |
| 50 to 100 | 140 | 0.7% | 3.7 |
| 100 or more | 29 | 0.1% | 4.3 |
Read the top two rows together: 87% of pairings sit under 15 possessions, and both bands average close to a single meeting. The long tail is genuinely tiny. Twenty-nine pairings in the entire league have crossed 100 matchup possessions.
Why it is this thin
Nothing is broken. It is a consequence of how basketball is played. Two given players share the floor only in the games their teams meet, only in the stretches both are in, and only on the trips they are actually matched. Switching, help, and rotation split what remains across several defenders.
Conference opponents may meet three or four times a season, cross-conference opponents twice. Multiply a couple of meetings by a handful of matched possessions each and you arrive at the table above.
Where the publishing line sits
A pairing is published on this site only after three meetings and 25 matchup possessions. That leaves 390 of 21,324, about 1.8%. It is a low bar in absolute terms, and it still excludes 98% of the pairings someone might search for.
It would be straightforward to publish all 21,324 and let readers judge. We do not, because a page presenting a rate built on four possessions is not neutral. Presentation implies that a number is worth reading, and most of these are not.
What to ask of any matchup source
- What is the minimum sample required before a pairing is shown? If there is no answer, the extremes on that site are noise.
- How many pairings clear it? A source publishing tens of thousands of matchup pages is not applying one.
- Is the sample displayed next to the rate, or buried? Anything that hides it is asking to be misread.
- Does the sample count possessions or only games? Three meetings at four possessions each is not three meetings' worth of evidence.
How these figures were produced
- 21,324 pairings; 87% under 15 possessions; 44.9% under five
- Group the matchup table by scorer and defender, sum possessions, bucket. Full distribution reproduced in the table above.
- 390 pairings clear three meetings and 25 possessions
- Count of pairs with COUNT(DISTINCT game) >= 3 AND SUM(possessions) >= 25.
Source is NBA player-matchup tracking, 24,974 aggregated rows, last refreshed 7 June 2026. Figures describe possessions already played. Nothing here predicts a future result, and no odds, lines, or selections are published anywhere in this section.
Related research
Every extreme in matchup data is a small sample
Variation in NBA matchup rates falls fourfold as the sample grows, and 55% of thin pairings are scoreless. The eye-catching numbers only exist where the evidence is weakest.
ReadDo NBA matchup splits repeat? A split-half test
Splitting each NBA pairing's games in half and correlating the two halves gives r = 0.44. Corrected for length, reliability is about 0.61: real signal, and roughly 40% that will not repeat.
Read