Basketball Defense: To Three Or Not To Three?
When I was putting together last week's mailbag, Brian forwarded along a question that both hit a topic I've wanted to cover for a while and required an extensive answer.
This seems so counterintuitive...wonder if college vs pro is different due to quality of shooters at pro level. @mgoblog https://t.co/H4L8lTxynY
— Peter Joseph (@PetejoeUM) April 28, 2020
The best defenses in the NBA this year, by and large, are allowing opponents to let fly from three-point range, which goes against the popular recent trend at all levels of basketball—prioritizing guarding the perimeter to help gain an advantage in three-point attempts.
Since John Beilein brought in Billy Donlon and then Luke Yaklich to be his de facto defensive coordinator, Michigan has been among the very best in the country at limiting outside shots, finishing in the top ten in 3PA/FGA the last four seasons. While that approach helped produce top-three overall defenses in 2017-18 and '18-19, the defense took a step back to 28th—not bad, but not elite—in Juwan Howard's first year.
While there were transition costs that helped explain the defensive dropoff, is there a better way to play on that end? Today, I'm going to look at the NBA's trend and whether it applies to college basketball. In my next post, I'll break down what it all means for Michigan and whether Howard's time with the Miami Heat signals that a change is coming.
Talkin' Bout The Bucks
The Milwaukee Bucks were the best team in the NBA in this thus far abbreviated season, going 53-12 with by far the league's best point differential. Having reigning MVP Giannis Antetokounmpo and the league's ninth-ranked offense (by offensive rating) will make a team quite good on its own. What has made the Bucks great, however, is their historically good defense, which ranks nearly four full points per 100 possessions better than the second-best in the NBA.
They've accomplished this even though three-point attempts make up 41.1% of opponent attempts, the third-highest rate in the league. Even when accounting for their great defensive personnel, it's clear coach Mike Budenholzer's tactics play a huge part in their success. Above all else, Milwaukee wants to ensure that opponents don't get a clean look near the rim, and by packing the paint they've become remarkably effective at doing so.
Milwaukee's defense allowed a league-low 29.1% of field goal attempts to come within six feet of the basket, and the gap between them and the #2 team was equal to the gap between #2 and #7. Opponents made only 50.7% of these shots, a ludicrous figure—the #2 team was at 57.3% and the league average if you remove the Bucks was 61.2%.
nothing at the rim, ever. [via Mike Prada]
The clip above features one small window where it appears there's a quality look but even that is by design. Number eight in white, Moe Harkless, is a 32% career three-point shooter even though he's mostly taken catch-and-shoot attempts in quality offenses. As the great Mike Prada explained in December, it's all part of the strategy:
Allowing three-pointers doesn’t seem smart, but the Bucks do and still have the best defense in the league. Last year, 36.3 percent of opponent shots were threes, the highest mark in the league. This year, 38.7 percent of them are threes, which is third behind Toronto and Miami. In exchange for turning the rim into a fortress, the Bucks leave the three-point line comparatively unattended. They have transferred all power into the rear deflector shields, so to speak.
Why does this strategy work so well? One reason is that the Bucks allow “good” threes, to the degree that any team can control them. They rank closer to the middle of the pack in corner threes yielded, which are more dangerous, and instead allow the most above-the-break attempts in the league. They specialize in surrendering the semi-open 26-footer from an average stretch big man rather than the in-rhythm corner pop from whiplash ball movement that began with dribble penetration.
(Prada was furloughed by Vox/SBNation and has started his own newsletter. If you're into basketball tactics written for a broad audience, it's greatly recommended.)
There are times when you'll see all five Milwaukee defenders positioned below the free-throw line to stop a drive. Yes, the ballhandler can kick it back out to the top of the key, but firing it to the corner is much more difficult in that scenario. This results in a defense that gives up inefficient looks on average despite so many of them being three-pointers. There are other benefits, too. Here's The Ringer's Zach Kram discussing if the Bucks have the best defense in history:
Cleaning the Glass calculates a stat called “location effective field goal percentage” that asks “if this team allowed the league average FG% from each location, what would their opponents’ effective FG% be?,” which then “gives us a sense of the efficiency of a team’s defensive shot profile.” The Bucks rank third in this metric, narrowly behind Utah and Brooklyn—meaning they would still profile as an elite defense even if they didn’t also have the best defenders at altering shots at the moment of firing. Budenholzer’s game plan works.
A rim-centric defensive philosophy affords the Bucks more advantages than merely forcing opponents into lower-percentage shots; fewer shots near the basket mean fewer realistic opportunities for the opponent to draw a whistle or grab an offensive rebound. The Bucks don’t foul often, with the fourth-lowest opponent free throw rate, and they’re on pace for the best defending rebounding season in NBA history, with Giannis leading the league in defensive rebounds per game.
Milwaukee has elite defenders, but other than Giannis, the main figures weren't necessarily regarded as such until they played for the Bucks—twins Brook and Robin Lopez, for example, are the second- and third- best rim defenders in the league by shot percentage behind, naturally, Giannis. Brook, who's been a revelation as a drop coverage specialist that stonewalls drivers with verticality, was a league-average defender until becoming a top-ten player on that end the last two seasons by most any advanced metric. That directly coincides with his time in Milwaukee.
The Bucks are unique in having Giannis, who's responsible for taking their defense from great to historic. Their success in this approach isn't an isolated case, however. Here are the top ten defenses this season along with their rank in 3PA/FGA:
Defensive Efficiency | Def. Efficiency Rank | Opponent 3-PT Rate | Opp. 3-PT Rate Rank | |
---|---|---|---|---|
Milwaukee | 101.6 | 1 | 41.1 | 3 |
Toronto | 104.9 | 2 | 43.6 | 1 |
LA Lakers | 105.5 | 3 | 37.7 | 17 |
Boston | 106.2 | 4 | 39.9 | 7 |
LA Clippers | 106.6 | 5 | 39.0 | 9 |
Philadelphia | 107.6 | 6 | 33.8 | 29 |
Indiana | 107.7 | 7 | 37.5 | 18 |
Brooklyn | 108.3 | 8 | 37.1 | 21 |
Oklahoma City | 108.4 | 9 | 36.8 | 24 |
Orlando | 108.7 | 10 | 38.6 | 13 |
Four of the top five defenses are also in the top ten by highest opponent three-point rate. There's one defense in here—Philadelphia's—that's an extreme three-point prevention squad and there are two more in the bottom ten of opponent three-point rate. While both approaches can be successful, the very top tier of defenses are taking Milwaukee's approach. (Or they have LeBron James and Anthony Davis.)
[Hit THE JUMP to see how this applies to college ball.]
Does This Apply To College?
Matt Painter is cool with this [Campredon]
To see whether the recent NBA trend also applied to college, I pulled data from KenPom from last season, then narrowed my focus to teams in high-major conferences—that's what matters to us, at least. From that set, I looked at the 20 teams on each end of the 3PA/FGA spectrum. I've sorted them here by defensive efficiency and included averages (my math skills are basic at best) to provide some context.
Here are the top 20 high-major teams from last year in suppressing three-point attempts. Michigan is among them (sorry for the extra significant figures, these are taken straight from KenPom's data set):
Adj. Defensive Efficiency |
Adj. DE Rank |
Opponent 3-PT Rate |
Opp. 3-PT Rate Rank |
|
---|---|---|---|---|
AVERAGE | 95.5609 | 64 | 32.60525573 | 54.6 |
Duke | 91.1058 | 12 | 26.66311868 | 3 |
Georgia Tech | 91.7939 | 16 | 32.9512894 | 49 |
Wisconsin | 91.8542 | 17 | 32.51287922 | 40 |
Penn St. | 92.4292 | 23 | 32.86410395 | 47 |
Texas | 92.6071 | 24 | 28.16229117 | 6 |
Michigan | 93.0352 | 28 | 28.33686441 | 7 |
Colorado | 93.3736 | 32 | 32.13114754 | 35 |
Illinois | 93.6652 | 35 | 32.48154458 | 39 |
Villanova | 93.9608 | 36 | 31.77826564 | 29 |
Minnesota | 94.2344 | 41 | 31.17233928 | 24 |
Cincinnati | 94.744 | 51 | 34.73154362 | 92 |
Arkansas | 95.4879 | 57 | 33.29646018 | 61 |
Tennessee | 95.7688 | 62 | 34.62441315 | 86 |
Marquette | 96.8613 | 73 | 34.81803368 | 97 |
Mississippi | 96.9578 | 74 | 34.68571429 | 89 |
N.C. State | 97.2983 | 79 | 32.84868067 | 46 |
Alabama | 99.471 | 114 | 34.71528472 | 91 |
Georgia | 100.773 | 136 | 34.77806789 | 94 |
Miami FL | 101.262 | 149 | 34.70437018 | 90 |
Vanderbilt | 104.534 | 221 | 33.84870237 | 67 |
On the other end, here are the 20 high-major teams that allowed the highest share of three-point attempts:
Adj. Defensive Efficiency |
Adj. DE Rank |
Opponent 3-PT Rate |
Opp. 3-PT Rate Rank |
|
---|---|---|---|---|
AVERAGE | 94.760705 | 60.15 | 42.36262556 | 298.3 |
Virginia | 85.0621 | 1 | 42.62717321 | 315 |
Kansas | 85.5375 | 2 | 41.30311615 | 288 |
Rutgers | 90.2277 | 6 | 41.61346489 | 295 |
Purdue | 90.7695 | 11 | 42.02287778 | 305 |
Florida St. | 91.491 | 15 | 40.22248244 | 266 |
Oklahoma | 93.0747 | 29 | 40.08221994 | 262 |
Louisville | 93.0868 | 30 | 41.02708804 | 282 |
Clemson | 93.4755 | 33 | 44.13671184 | 332 |
Oklahoma St. | 94.5199 | 46 | 44.4197668 | 335 |
South Carolina | 94.7198 | 50 | 39.84421809 | 251 |
St. John's | 95.2006 | 53 | 42.23044397 | 308 |
DePaul | 95.4771 | 56 | 41.31769961 | 289 |
Oregon | 97.1007 | 76 | 40.61797753 | 274 |
Texas A&M | 97.2257 | 77 | 50.27124774 | 353 |
Virginia Tech | 97.9768 | 82 | 42.54859611 | 314 |
North Carolina | 98.3509 | 94 | 41.22061031 | 286 |
Iowa | 98.6328 | 97 | 40 | 259 |
Syracuse | 99.56 | 116 | 47.91220557 | 350 |
Northwestern | 101.278 | 150 | 42.41266376 | 312 |
LSU | 102.447 | 179 | 41.42194745 | 290 |
There are three immediate takeaways.
Both approaches work. While the Milwaukee approach works a little better in this sample, it's close enough that what's best for a certain team should come down to what best fits their personnel and coaching staff. The three-point prevention squads are less than one point per 100 possessions better on average and the average defensive ranking of these teams are within four spots of each other.
The elites play like the Bucks. That said, to break into the very top tier, it appears prioritizing rim protection over three-point prevention is the way to go. Both samples have five top-25 defenses but that becomes misleading. All five of the Milwaukee-style teams are in the top 15, including the top two defenses in the country, Rutgers at #6, and Purdue at #11. The best-ranked defense in the three-point suppression group is Duke at #12.
Yes, this is like the NBA. This shouldn't come as a major surprise. The extension of the three-point arc to FIBA distance means there's now a difference in quality between corner and above-the-break attempts, regardless of how they're created, since there's now actually a change in the distance. College teams have more non-shooters than NBA squads, giving defenses greater ability to sag off shooters into the paint. College guards are less able to finish at the rim against high-level rim protectors. Think about Michigan's recent teams and how you'd prefer to defend them; I'd be going with Milwaukee's style.
Coming soon, I'll break down how Michigan has played under Juwan Howard, whether he should consider a change, and if his previous coaching stop hints at whether that'll happen.
Very interesting read. Thanks Ace.
Super interesting. I sort of assume some percentage of three pointers shot against elite defenses are late-shotclock bailout shots--in other words, having an elite defense, regardless of style, should at least slightly increase the percentage of shots that come from 3 after the offense breaks down. That assumption may be completely wrong, though. Would be interesting to look at when and where shots come in the shot clock and incorporate that into the data here.
This was my question as well.
If there's any difference amongst elite teams, it's probably that their late clock shot distribution becomes slightly skewed towards twos relative to their early shot clock distribution (i.e. for these teams that give up 40% of FGs from three overall, it's probably closer to 35% or 30% late in the clock).
A well-coached defense will tighten up their perimeter defense late in the shot clock because a drive and kick becomes less feasible and a hockey assist becomes impossible. So you want to force a drive because it takes longer than a jacked three and your teammates can help more aggressively without worrying about the play being extended.
Very interesting....and not what I would have expected. Learning about "drop coverage" earlier in the season opened my eyes toward letting some players have a jump shot that isn't all that efficient to prevent an easy bunny at the rim...but wasn't expecting this.
A few other thoughts:
- The FIBA line definitely revealed which players actually were good shooters and which ones weren't. Letting a few more of those mediocre shooters take shots isn't as painful.
- Is there data that shows how the NBA and NCAA compare for 3pt% - both on average and "number of good shooters per team"?
- Despite the logic here and the desire to mitigate the recent influx of good bigs into the league with such a strategy....it sure felt like Michigan got torched from the outside at times this year to the point where we questioned drop coverage of if there was a hex on this squad that let crappy shooters hit open shots against us.
Also noteworthy about the Bucks strategy is that the NBA has defensive 3-seconds, meaning they can't let their bigs just linger in the paint to deter drives and help out.
Maybe that actually works to their advantage.
Super interesting indeed. Definitely goes to show that it's not necessarily about generally suppressing threes or even about generally suppressing drives to the rim, but it's about having a very specific game plan based on your individual opponents which isn't surprising even if it is counterintuitive that doing so might end up allowing a relatively high rate of threes (which does surprise me somewhat).
I kept harping on this early in the season when Michigan was clearly way, WAY too three point averse. It was like they hadn't scouted their opponents. They were letting Garza, Williams and other very good bigs go one-on-one while not leaving shooters despite some of them being very bad shooters. It is a desirable outcome if you dig down on Williams and get him to pass to a sub-30% shooter. You don't even close out very hard on that guy because you want him to shoot it.
And on the flip side, this is why Jon Teske shouldn't have been shooting threes. It's a terrible outcome for Michigan. He doesn't draw his defender from the hoop because they're giving him that shot, he's not going to get fouled, he's not going to get an OREB (so the team overall is less likely than most other shots since he's the best oreb guy) and he simply made a very inefficient percentage of those shots.
BuT ThEy DoNt PlAy DeFeNsE iN ThE NbA!!!
Giannis is so good on the defensive end of the floor. I think 20 years ago a lot of people would of looked at his counting states (1 BPG, 1 SPG) and write him off as a decent defender. He was probably on his way to becoming the 3rd player to win Defensive Player of the Year and MVP in the same season this year (MJ, Olajuwon).
Are there any other Anteetokounmpo's in the pipeline that Michigan can look at?
You mean other than Isaiah Todd? I doubt it. Players with that combination of size and skill rarely go unnoticed. Unfortunately, the pipeline for proto-Antetokounmpos unexpectedly moved from college to the G League.
I think 20 years ago a lot of people would of looked at his counting states (1 BPG, 1 SPG) and write him off as a decent defender
Dennis Rodman and Joe Dumars got lots of defensive accolades despite not racking up huge counting stats. (Rodman won his two DPOY before he started winning rebound titles.) It's not that hard to figure out who's good defensively.
This description of the Bucks defense does sound very much like the approach favored by Texas Tech last season. They obvs weren't in the 2019-20 sample, but if I remember right, they had the same philosophy of keeping people out of the paint and away from the rim.
"Nothing at the Rim". Damn right. The more things change ....
After just watching "Bad Boys", I wonder how the 87-90 Pistons teams would compare statistically in terms of number of shots and eFG% close to the rim.
Or, another angle, take the top NBA defenses over the last 20 years - do they all have this in common, or are there outliers? Is it possible to have a good defense if opponent eFG% at the rim is above average (instead focus on allowing nothing from outside)?
Please and thank you?
Comments