213hoops.com
Unraveling Clutch Offenses Through Shooting Analysis
Pondering the Question
At the start of the season, I was watching the thrilling double-overtime game between the Boston Celtics and the New York Knicks, and in the first OT I noticed that after Jayson Tatum tied the game at 128-128 with 2:58 left, the rest of the period consisted of 9 consecutive missed jump shots from both teams, most of which were three-pointers.
In a series of tweets, I’d wondered whether these modern NBA offensive schemes that emphasize three-pointers were actually the right call in such end-game situations. In my mind, the volatile, high-risk/high-reward three-ball could be less preferable to the more conservative, but reliable two-ball.
Of course, the ever-perceptive Cole Huff (@colehoops) simultaneously pointed out that in this instance, the three-point attempts and low efficiency had a much simpler explanation: tired legs at the end of a long game. And as the season’s progressed and the entire league’s seemingly been in a shooting slump, consecutive shortened off-seasons seems to be the likely cause. Classic case of Erik unnecessarily overcomplicating things. But the question remains, and it’s one I’m still interested to explore:
At what point in a game does the scoring efficiency (eFG%) of the three-pointer have diminishing returns, as its lower chance of being successful (FG%) becomes more of a factor?
(Shout out to Shane Young (@YoungNBA)—measuring the value of shot reliability in small sample sizes might be the one time FG% is actually useful in analyzing modern NBA scoring.)
The Three-Pointer and Moreyball
We’ve covered this plenty in the past, but for those just joining us, a brief history recap: in the last decade or so we’ve seen modern offenses shift to focus on shooting two kinds of shots: three-pointers and layups—a.k.a. Moreyball. The logic is that three-pointers are worth more than two-pointers, so you’d only need to shoot 33% on a three-pointer (not very hard to do consistently) to equal the eFG% of shooting 50% on a two-pointer (much harder to do outside of layups). Consequently, we’ve seen teams demolish the all-time offensive rating records multiple years in a row.
Now, with a heavy diet of shots that are missed more often than they are made, we’ll see some wild swings. Looking at individual games, we might see teams score 150 one night and 80 the next night on the same shot profile. But with large enough sample sizes over the course of a season when the Law of Averages can take hold, these Moreyball shots have nearly always led to more efficient scoring.
That being said, there are still some pitfalls that are especially amplified during the playoffs. First, the three-pointer is heavily reliant on set-up action, for most players. And in big moments teams often tighten up and stray from their complex offensive sets, instead switching to simpler action like high pick-and-rolls or isolation plays where fewer things can go wrong. Second, the long ball is unreliable. When you just need a bucket, ANY bucket, even the most elite three-point shooters are still more likely than not to miss the three—it’s just the nature of the shot.
So when sample sizes shrink in crunch-time, teams that are reliant on the three-pointer must find a plan B. And if that plan B isn’t as efficient at scoring points as the quality threes that they’re used to, they suffer a big drop-off.
Finding an Answer
To answer the question about diminishing returns of three-pointers, I ran a series of simulations, each run 100 times, comparing two imaginary teams in a scoring competition:
- Team 3PT: Shoots nothing but 40% threes (eFG% of 60%)
- Team 2PT: Shoots nothing but 55% twos (eFG% of 55%)
With a large enough sample size, Team 3PT should ultimately be more efficient and win more games against Team 2PT. But we need to find out how small the sample sizes would need to be for the volatility of the three-pointer to come back to bite Team 3PT. It’s important to note that a 5% eFG% swing is the difference between the #1 ranked efficiency last year and #23 ranked efficiency—it’s a CHASM. Team 3PT should wipe the floor with Team 2PT in this scoring competition.
Simulation 1 — 100 full Games, 88 Field Goal Attempts per game
Using last season’s FGA per game rate, this went about as expected. Team 3PT won an average of 70% of the games against Team 2PT, with an average margin of victory of about 12 points, across all of the simulations. Given enough time that volatility doesn’t matter. Team 2PT couldn’t even get close to winning more games.
Simulation 2 — 100 Five Minute Overtimes, 10 Field Goal Attempts per game
Shrinking down sample size to just 5-minute overtime periods, while Team 3PT still won the majority of the matches, it was a lot closer. In these simulations, Team 3PT only won an average of 53% of the games against Team 2PT, with an average margin of victory of just 1 point. When we’re talking about the #1 offensive team vs the #23 offensive team, that’s way too close for comfort.
Simulation 3 — 100 Last-Two-Minute Clutch Scenarios, 3 Field Goal Attempts per game
Shrinking the sample size even further, we finally saw the breaking point. Team 3PT only won an average of 49% of the games against Team 2PT, with an average margin of victory of just half-a-point. It was now more likely for Team 2PT to win or tie, an unquestionably bad outcome for the #1 offensive team in a scoring battle with a vastly inferior scoring opponent.
Main Takeaways
Let’s be clear—in NO way is the takeaway here that teams shouldn’t shoot threes or that they should avoid Moreyball shots. These are the most efficient ways to score, and teams should always try to score most efficiently.
But what this analysis has told me is that teams that downright refuse to shoot anything but threes or open layups might be getting in their own way in crunch-time.
The three-pointer’s diminishing return in the clutch mean that teams should be open to expanding their offense to allow for lower eFG% (scoring efficiency) two-pointers that have a higher FG% (chance of being successful). And having the entire floor to work with is a huge relief for a coaching staff.
This just really goes to show the value of guys like Steph Curry, Chris Paul, Kevin Durant, Kawhi Leonard, etc. who can switch it up and get high FG% buckets when they need to. And it’s a great example of analytics absolutely supporting efficient mid-range shooting, despite what you might have heard. To use a recent example, last year’s Phoenix team was the perfect embodiment of this approach. They led the league in mid-range efficiency, and between Chris Paul and Devin Booker, they were generally terrifying in the clutch.
Dude, CLIPPERS Takeaways
Oh, well after Phoenix, last year’s LA Clippers were the second team that most embodies this approach, featuring a very similar shot profile. They were top 5 in mid-range attempts and top 6 in mid-range efficiency, while maintaining a top 3 offense, thanks largely to their league-leading three-point efficiency. And this year the Clippers are built to add even more complexity to that shot profile, having added a few players with more bounce that can actually attack the paint now.
The shots will eventually start falling—as noted by Justin Russo (@FlyByKnite) the Clippers had the second best shot quality in the NBA after 5 games—and when they do, the Clippers should once again be tailor made for the playoffs.
Unraveling Clutch Offenses Through Shooting Analysis
Erik Olsgaard