Clutching Pearls
A very long discussion of a very small sample
I’m a bit of a defender of small samples in the NBA.
In another life, I worked as both a psych lab assistant and a data research project manager. I have worked with SPSS and done the Rs and Ps and the like. I’m not an expert in those fields1, but I can make my way around a white paper.
There is good science and bad science. That’s objectively true, based on the scientific method and reliability standards.
But there is a massive difference between data that I would use to prove in a whitepaper that a lineup featuring the same players is or is not beneficial in key variables and the data I would use to say “hey, this lineup is getting its ass kicked this season.”
Now, one of the fundamental problems is that something can absolutely not work in small samples and then work in large samples or vice versa. If you only use a lineup against the Nets and Wizards and then try it against the Thunder, that will turn very quickly.
You can also just… play better. This is the one that most fans never really consider. If a lineup doesn’t work, it should never play ever again instead of figuring out how to make it work. That improvement can also make something look better in larger samples.
Seeing something more often across more conditions is always beneficial, but also, you can also simply determine if something is working or not working in the NBA.
It’s why I so often use single-game plus-minus, which data researchers and analytics wonks absolutely despise, and for good reasons. But to me it’s beneficial as long as you don’t use it to describe how a player played instead of what worked or didn’t work in their minutes.
For example, Nikola Jokic was a -6 in Sunday’s loss to the Steph-less, Butler-less, Draymond-less, Porzing-less Golden State Warriors. Jokic also had 35 points, 12 rebounds, and 11 assists on 11-of-21 shooting.
I have consistently said that his box score numbers do not reflect how Jokic played. I have seen him loaf and rack up stats, and more often than that, he’s absolutely dominated a game with 18 points and 12 assists. But I do agree that you cannot look at that stat line and say that “Jokic played badly,” even by his impossible standards.
What I can say is that his impact was not meaningfully felt in that game the way they needed it. If Joker is going to put up that stat line, the team needs to win his minutes by double digits, not lose by six. That’s where the blame for the role players comes in.
“It must be the other guys because, I mean, look at that stat line!”
But what Joker has always been, from the moment he arrived, is a winning player. It’s what those of us covering the team marveled at before the stats became more and more absurd. “Man, they just win with him on the court.”
The Nuggets were -6 in the Jokic minutes, -5 in the non-Jokic minutes.
When Jokic was on the floor, they had a 119.8 defensive rating, without him it was 125, six points worse. Meanwhile, the offensive rating was only 110.8.
So if:
a. Jokic had an insane stat line and poured in 35 points on 21 shots and
b. the offense was still bad, then it was not his fault and has to be on the other players and
c. even if the defense was bad with him on the court but
d. it was worse with him off the floor…
Then it wasn't Joker’s fault, and everyone else was the problem.
Except this ignores the flow of the game, analysis of which requires breaking this into even smaller samples. Denver had a 139 defensive rating in the first quarter, and a 172 defensive rating in his 4th quarter. They put themselves behind the 8-ball with terrible, sluggish play2, and then melted down the stretch.
This isn’t to say the other Nuggets played well; they didn’t, because those same criticisms about Jokic apply to the others. They didn’t play well as a team.
But while I’ve seen games where Jokic has done absolutely everything and the team lost3, this was not one. No one played well on both ends of the floor.
The point of all this is to say that we can take meaningful analysis from small samples, something I do quite frequently.
Which is why it’s weird I don’t really believe in clutch time.
THE CLUTCH CONVERSATION
There is no other way to parse this; the Nuggets are absolute garbage in the clutch this year.
Denver is 14-15 in the clutch overall, with the second-worst defensive rating in clutch time and the 10th-highest clutch time turnover percentage.
What’s worse, with Nikola Jokic in the lineup, they are 6-134 this season, with a 137 defensive rating5 and a -19.3 net rating in 87.5 minutes. They don’t lose in the clutch.
They get blown off the floor, out the door, and into the nearest large body of water to sink slowly into the murky depths in the clutch with Jok.
To compare, the Nuggets had a 61.6% win percentage and an outrageous +12.1 net rating from 2018-19 through 2024-25.
So yeah, it’s been bad. I don’t need to tell you that; you watch the games, you know.
The question is whether it actually means anything.
I get why it feels like it means something. High-pressure situations are considered not only the most valuable but also the most determinative of important aspects of human nature and competition. You need to be able to execute when the stakes are highest, when the pressure is highest, when the going gets tough.
Certain teams and players have that clutch gene, and others don’t. And that matters in the playoffs, when things are toughest.
That’s the idea.
But whenever we’re faced with these ideas and tropes, I want to know, “Is that actually true?” It doesn’t mean they’re often false, I’d anecdotally bet it’s 50/50.6
So I did the work, at least for the last 10 years.
THE CLUTCH FACTS
I looked at clutch data for the last 10 years, comparing clutch performance in win percentage and net rating vs. playoff series advancement7. The results were absolutely fascinating.
Here’s the breakdown of clutch averages for various teams:
Teams that make the playoffs average a 57.2% Clutch Win Percentage and a +5.3 Clutch Net Rating.
Teams that advance to the Second Round average a 59.9% Clutch Win Percentage and a +7.6 Clutch Net Rating.
Teams that make the Conference Finals average a 59.9% Clutch Win Percentage (identical to 2nd round teams).
Conference Finalists actually average a worse Clutch Net Rating (+6.0) than Second Round teams (+7.6), proving elite teams can survive bleeding points late if they secure the win.
Teams that make the NBA Finals average a 61.5% Clutch Win Percentage and a +9.0 Clutch Net Rating.
Championship-winning teams average a 60.5% Clutch Win Percentage during the regular season.
Championship-winning teams average a +8.0 Clutch Net Rating during the regular season.
The average championship team ranks 6.7th in Clutch Win Percentage and 7.5th in Clutch Net Rating.
Now, all of that data says that the average champions and teams that win in the playoffs are good clutch teams, which is pretty worrisome for Denver.
Except there’s this:
