It’s been awhile since we’ve taken a deep, obscure statistical dive around here and squeezed every little dust particle of data out of it, hasn’t it? So why not put that hat back on and revisit a concept first examined in the very early days of this site’s existence?
In 2017, I wrote about the Brooklyn Nets under Kenny Atkinson, long before they got their current roster, and considered how combining an up-tempo style with a plug-awful team only managed to make Brooklyn 9-35, holders of the league’s worst record on January 25, 2017, a state of affairs that ultimately resulted in the Celtics—who had pick-swap rights with Brooklyn in the 2017 draft and who traded the top pick to Philadelphia—getting Jayson Tatum.
But that’s not relevant here. In that old article, I tried to correlate pace with winning, asking whether faster teams were in any way meaningfully better than slower teams.
The answer was “yes and no”, and the argument seemed to hinge on not whether a team played fast but whether they amplified the sample size on their possessions in any given game and, as the saying goes, “ran the opponent off the floor.” The eventual 2017 champion Warriors were one of the fastest teams in the league, and high Net Rating plus more possessions equaled 65 wins and a Game 4 loss in Cleveland in the Finals being the only thing that separated the Dubs from a 16-0 playoff record.
But there’s another way to look at pace and Net Rating, and it’s to consider whether there’s a correlation between two teams with the same (or close enough) Net Rating, one team playing fast, the other team playing slow.
After all, it’s all about sample size, right? The Nets played fast with a bad team and ended up with (well, Boston and eventually Philly did, anyway) the top pick. The Warriors played fast with a great team and won the title with one of the greatest playoff runs in NBA history in the same year.
So let’s carve up the data a bit differently and remix this idea of “suck faster.”
Exhibit A: 2019 As Baseline
Let’s start in 2019, the last 82-game season before COVID happened, and look first at two good teams at opposite ends of the pace spectrum and then at two pairs of bad teams in the same situation.
The Indiana Pacers went 48-34 with a plus-3.4 Net Rating and a 98.1 pace. The Oklahoma City Thunder, meanwhile, went 49-33 and posted a plus-3.3 Net Rating and a 102.9 pace. So playing about 5 percent faster got one team about 2 percent more wins (1/48.)
We’re beginning to see a problem here…
But no matter. Let’s look at the other comparison, and it comes in four delicious flavors instead of two.
Specifically, consider in order the Net Rating, record, and pace of each of four teams from outside the playoffs.
Sacramento: -1.1, 39-43, 103.1
Charlotte: -1.1, 39-43, 98.7
New Orleans: -1.2, 33-49, 103.3
Dallas: -1.3, 33-49, 99.0
Again, there’s a 4 or 5 percent difference between teams with fundamentally the same Net Rating, but if you divide those four teams into pairs with the same record, one played fast and one played slow.
What’s more, those divisions are so small that all four teams were expected to go 38-44 based on their overall point differential, yet two teams overachieved by one game and two other teams underachieved by five games.
At the more extreme ends of the scale, the two fastest teams in pace were Atlanta (103.9, minus-5.8 Net Rating) and Milwaukee (103.3, plus-8.4 to lead the league.) The Bucks, unsurprisingly, went 60-22, underachieving their 61-21 expected record by one game. The Hawks, also unsurprisingly, went 29-53 but beat their 27-55 expectation by two games. You’d think that fast teams would pull away from their expectation because of more chances to apply their superiority (or inferiority) over more possessions.
But that’s not what happened. The records compressed a bit as the league overall always wants to regress to its 41-41 per team mean.
So This Is Pointless?
Well, yes and no.
Yes, because it’s obvious that pace isn’t the end-all be-all of NBA performance no matter how good or bad your team is. There are just so many more factors that go into a basketball game than possessions when the difference between the fastest and slowest teams in the league is only about 6 or 7 percent and the vast majority of teams cluster in about a 2-percent range around the average.
No, because absolutely nothing I just said there takes away the fact that math is what math is and as I mentioned five years ago, this is how slow-paced college teams pull upsets in the NCAA Tournament where the variances in both Net Rating and pace are massively higher than they are in the pros. The concept is sound for an entire sport…the difference is that in the NBA, the variance just isn’t high enough to create broader trends that are enough to overcome more important factors like effective field goal percentage or defensive rebounding or the way the ball bounces at the free throw line on any given night.
The totality of NBA basketball is such that other than brutally obvious things like making shots, those subtle side stats can only define an average (expected W-L%) that teams routinely over- or under-shoot for a bunch of reasons other than how fast they play.
Is There Any Lesson Here?
There is. In situations where broad differences in overall Net Rating can affect an individual game, control of the clock on a micro, single-game level, is key. Think an NFL team that can run the ball when they have the lead in the fourth quarter and make first downs on the ground. Or a hockey team that controls the puck as the clock winds down in the third period, slowing down the number of shots on goal and playing defense in a way that takes away breakaways.
In every sport with a clock, there are ways to reduce the number of scoring opportunities (obviously, sports like baseball don’t benefit from this in any way—whether the pitcher takes three seconds to throw the next pitch or whether the pitcher’s motion could be confused for a rain delay has absolutely nothing to do with the number of pitches he’ll ultimately have to throw to get three outs.)
That’s the lesson in basketball. Over the course of a season, teams are going to regress toward the mean, and it’s certainly in a team’s best interest to try to push the tempo when they’re better than their opponent to take advantage of that small difference in sample size. You might get an extra win or two per season that way, and that can make a difference between hosting a Game 7 in a playoff round or playing it on the road.
But on a micro level, it might help to have plays in the playbook for a slow-it-down approach as early as the third quarter of a blowout, the better to avoid the rash of “no 30-point lead is safe” games that have become part and parcel of a faster-paced “threes and layups” league since the Warriors and Rockets revolutionized the game starting in 2015.
That’s the real lesson here. If you suck, don’t do your opponent the favor of sucking faster.