We have known about the correlation between 3-point and free-throw shooting for a while now. One of the godfathers of basketball analytics, Ken Pomeroy, wrote this in a 2006 blog post:
It turns out there is a connection, admittedly weak, between free throw percentage and 3-point percentage. Moreover, in limited cases we can use free throw percentage to assess who may have had circumstances beyond his control in his favor (or against him), and thus is due for a correction in 3-point accuracy.
The idea makes sense: There’s no more controlled environment for a shot than at the free throw line, where players can get into their preordained rhythm and shoot with their set motion. That’s why the best overall and 3-point shooters in the world, like Stephen Curry, Steve Nash, and Ray Allen, are or were all 90-plus percent free-throw shooters for a large portion of their careers.
Again, it’s not a perfect correlation, but it is plenty strong — and the idea makes enough intuitive sense — that we can look at 3-point and free-throw accuracy on the year to find which players are due for regression behind the arc. To do so, I used basketball-reference.com to narrow my list to players who have played at least 500 minutes, attempted at least 10 3-pointers, and attempted at least 25 free throws. That leaves us with a player pool of 138.
To highlight players and where they stand on the regression curve, I took the sample’s percentile score for free throw percentage and 3-point percentage, and found the difference between the two. Those with high numbers are likely due for positive regression, whereas those with low ones are likely posting unsustainable shooting numbers (data as of 12/7).
And here’s a visual representation, sorted by team:
Negative Regression Candidates
Because the season is still somewhat young, we have plenty of players shooting over their heads. Like way over their heads. Take new Chicago point guard Kris Dunn, who is shooting only 57.7 percent from the line — that’s in the 17th percentile among players in this sample — but is hitting 3-pointers at a clip of 44.4 percent. For reference, Curry, who is, at this point, unarguably the greatest 3-point shooter ever, is a career 43.5 percent shooter. Dunn shot 28.8 percent from the 3-point line last year and was a 35.4 percent shooter in his three years at Providence. Regression will come soon.
Dunn is the most extreme example of this study, but it’s not like he’s jacking up a ton of shots. Plus, the Bulls aren’t incredibly exciting. The most interesting case is with the young Boston wings, Jaylen Brown and Jayson Tatum. Here are their percentages and percentile ranks for both free throw and 3-point shooting.
- Brown: 60.4 percent FT% (23.5 percentile score), 40.2 percent 3P% (66.0 percentile score)
- Tatum: 82.8 percent FT% (74.2 percentile score), 51.9 percent 3P% (100.0 percentile score)
Brown is not a great free throw shooter, and he’s likely getting lucky from behind the arc. He could become an average 3- point shooter for his career, but it’s unlikely he’ll be an elite one, and especially not a guy who can consistently exceed 40 percent. Tatum, on the other hand, is probably going to be a very solid 3-point shooter: He’s shooting nearly 83 percent from the line, and he shot 84.9 percent from the line in his one year at Duke. The issue is that he’s shooting an NBA-high 51.9 percent from the 3-point line so far this year, which is obviously unsustainable.
Interestingly, Boston has one more player in the top-10 of this sample: Al Horford, who is shooting a fine 78.7 percent from the line but a stupid 45.6 percent from the 3-point line. The Celtics are currently a league-best 22-4, although their Pythagorean win expectation — their win total based on point differential — is only 19 games. Factor in their incredibly unsustainable 3-point shooting to begin the year for most of their main players and it’s clear that the Celtics will certainly fall back a bit from their early lead in the Eastern Conference.
Another team that will likely see negative regression is the Indiana Pacers, who have four players (Victor Oladipo, Domantas Sabonis, Bojan Bogdanovic, and Lance Stephenson) who are shooting over their heads currently. Sabonis and Dipo have been red-hot to start the year and currently boast 3-point marks of 45.5 and 43.7 percent, respectively. Sabonis is probably going to be a solid shooting big man in his career, and he’s certainly benefitting from the increased spacing in Indiana compared to Oklahoma City. Oladipo gets the same spacing bump, but he’s a career 35.6 percent 3-point shooter, and his free throw mark is only in the 65th percentile. The Pacers are only one game over their Pythagorean win expectation, but their talent doesn’t suggest they should be a top-seven offensive team.
One more: I’ve been #TeamLonzo since last year, but Lonzo Ball is actually due for negative regression, even from his miserable 25.0 percent 3-point shooting mark. At least according to his awful 50.0 percent three throw mark, which is the worst mark in the sample. He was a 67.3 percent free throw shooter at UCLA, but neither mark is particularly encouraging. This will be the defining conundrum of his career.
Positive Regression Candidates
At the top of the positive regression list are two Portland guys, Evan Turner and Damian Lillard. Here are their percentile profiles:
- Turner: 85.0 percent FT% (79.2 percentile score), 19.5 percent 3P% (5.8 percentile score)
- Lillard: 92.6 percent FT% (96.4 percentile score), 33.2 percent 3P% (45.6 percentile score)
Interestingly, the correlation between free throw percentage and 3-point percentage does not seem to matter for Turner. Despite shooting well from the line over his past three seasons, he’s been awful from behind the arc:
It’s very possible that the longer distance for Turner just isn’t something he can consistently master. Lillard, on the other hand, is a career 36.7 percent 3-point shooter but is down to just 33.2 percent this season. Part of the issue for Lillard could be the quality of attempts he’s getting this season, as he’s one of the NBA’s most prolific off-the-dribble gunners. Still, this is one of the best shooters in the league; he should see positive regression sooner rather than later.
In his sophomore campaign, Nuggets guard Jamal Murray is playing 27.1 minutes per game, using 24.4 percent of his team’s possessions while on the floor, and shooting a very nice 50.8 percent from inside the arc. The issue is his 3-point shot hasn’t gone down: After shooting a slightly below-average 33.4 percent last season, he’s down to a miserable 30.4 percent this season. He profiles as a great NBA shooter, however: He made 40.8 percent from behind the arc in his one season at Kentucky, and he’s shot 88.3 and 92.4 percent from the line in his first two professional seasons. Murray has been a bit unlucky to start the year and should get back to average or higher as the year progresses.
Let’s hit on a couple more high-profile guys, starting with Joel Embiid, who has taken 54 3-pointers this season, which in itself is a good sign and helps the spacing of the 76ers’ offense. Unfortunately, those shots haven’t really dropped, as he’s shooting just 24.1 percent from behind the arc. He’s a good shooter though, as evidenced by his solid 78.3 percent free throw mark, as well as his 42.9 percent mark on long 2s:
The 76ers have already performed as a top-10 offense with Embiid on the floor; when he starts nailing 3-pointers off pick-and-pop opportunities from Rookie of the Year lock Ben Simmons, watch out.
Stephen Curry hasn’t been a bad 3-point shooter this season, knocking down 38.1 percent of his attempts. Of course, that’s only if you don’t measure him against his historically-great career marks. The dude is a career 90.3 percent from the line and 43.5 percent behind the arc. He’s currently out a couple weeks with a sprained ankle, and he has also been dealing with a right hand contusion this year. Still, he went for 30-plus points in his past two games before his ankle injury, and he hit 10 3-pointers during that time. The best 3-point shooter of all time should get back into the 40-point range once he returns from this ankle injury.
Applications
This data is useful for both DFS and gambling purposes. For the former, it’s obvious: Players who are due for negative shooting regression are likely a bit overpriced, whereas players due for positive regression are likely underpriced. For betting purposes, focus on the teams who have several players shooting over their heads, like the Celtics or Pacers. They will likely not continue to shoot the ball quite as well, which means there’s likely an advantage in fading the public on these surprising early-season teams.
