top of page

In Search Of Positive Regression

  • lukehwatson
  • Feb 25
  • 8 min read

The most viewed article in the entire history of our blog -- and the one I'm most proud of -- remains "Millie Prior and the Math Behind Davidson's Defense." It was incredible to watch the fanbase rally around Davidson WBB down the stretch last year, as the Cats put together a win streak that earned them their first double bye and WBIT berth in program history.


Expectations were elevated this year, as Davidson secured an elite nonconference schedule that was capable of positioning them for an NCAA Tournament push. However, as promising as much of the noncon slate was, the Cats fell short of the near perfect mark they would need in order to break through for an at large. Davidson needed a signature win against one of Mississippi State, Baylor, or NC State, but it eluded them. While they defeated several top 125 level opponents like Ball State and High Point, they kept falling short in quite heartbreaking fashion in the big ones. The Baylor loss, for instance, saw the Cats post their worst 3 point shooting performance of the season. NC State saw the Cats rally from down huge early only for the Pack to hit tough bucket after tough bucket down the stretch to keep Davidson at bay.


This apparent bad luck hasn't reversed during the A10 schedule. The Cats lost a second multi-overtime game in barely a month at Richmond in what was probably one of the greatest games in A10 history. Davidson led at Rhode Island in the fourth quarter before going cold offensively at the worst possible time, and a pedestrian outside shooting team lit it up from deep on a cold night for the Cats.


I was starting to worry I had let the cat out of the bag by highlighting Davidson's defense last spring. I thought that I had given everyone a scout on it (as if Division I basketball coaches, who are paid to do this every day, didn't already see what we were trying to do). But I kept feeling that Davidson was still executing well and just getting very unfortunate breaks.


I wanted to find out if the numbers matched the eye test. It was obvious Davidson's opponents were shooting much better than before; however, the Cats actually have a higher rated defense this year (44th) than last (48th). That led me down a rabbit hole that has me feeling much better about the season and the Cats' A10 tournament hopes. It's a testament to the culture that Coach Fulks and the staff have built that 20 wins might somehow feel underwhelming, but I think the team is legitimately better than their record -- and better than ever.


Davidson's defense is getting better -- its luck is just getting worse. If the former holds, the latter should reverse.


The Optimization of Davidson's Defense (A Brief Refresher)


Davidson's trademark switch and drop-heavy defense really emerged in their breakout 2023/24 campaign, when the Cats were 2nd in opponent assist rate and 18th in opponent assist rate nationally. The scheme emerged further in 2024/25, as a full strength Cats squad got even better defensively, and the staff's vision came more into focus. Davidson was 1st in the entire country for lowest opponent assist rate, and 9th in lowest three point attempt rate against. Teams were taking the exact shots we wanted them to, opting for dribble jumpers and post ups instead of catch and shoot opportunities and backdoor cuts, which are the most valuable shots in the game. My favorite Davidson WBB stat of all is that the Cats went 58 consecutive games without an opponent cracking 10 made threes from 2022 to late 2024. After allowing 11 to Ball State, they then went the final 24 games of 2025 without doing so. That meant that only once in 81 tries did Davidson's opponent make 10 or more three pointers from late 2022 to the end of 2025. That math advantage produced a lot of wins on its own.


The shot type data and efficiency numbers (via Hoop Explorer) bear out just how optimized Davidson's scheme was for the modern game:



It jumps out that Davidson was extremely efficient guarding drive and kicks (0.69 PPP against) and backdoor cuts (0.76 PPP). Davidson hardly ever gave up attack and kicks (10th percentile in frequency), and the 3 point numbers allowed also show that for you.


It's also unsurprising that Davidson's opponents made their hay from attacking the rim, posting up, and taking dribble jumpers. All three were used at frequencies above the D1 averages (see each's f %tile). And while teams actually fared comparably well compared to the average efficiency for each respective play type, the important thing is that all these play types inherently possess lower average efficiency. For instance, Davidson allowed 0.84 PPP on opponents' drives to the basket, but even below average efficiency guarding those shots rendered those plays less efficient than the average play in college basketball, which is worth about 0.89 PPP. In short, opponents took the inherently inefficient shots Davidson wanted them to take, like one-on-ones in the post, instead of more efficient shots, like catch-and-shoot threes.


A key part of Davidson's defensive identity is that it's better to let a big play against a small one-on-one than let the big play inside out by sending too much help, and the numbers display why that's smart basketball. But I already wrote all about that, and I want to instead focus on the "dribble jumpers" numbers this time -- and the benefits and drawbacks of drop coverage.


In Defense of the Drop


As you'll remember (or have re-read), even prime Steph Curry shot worse on unassisted jumpers (aka dribble jumpers) than assisted jumpers. It's also important to note that in a drop-heavy scheme, dribble jumpers are going to be unguarded, as the player guarding the screener in a pick and roll will... you know, drop. That entices the ball handler to take the shot themselves.


Of all the options on a pick and roll, you want the ball handler to shoot off the dribble. It's less efficient than allowing the roller to catch the ball, and, of course, it's much less efficient than allowing another player to shoot the ball off the catch. An unassisted shot is, again, inherently more difficult. Drop coverage also makes it easier to rebound since a) the big has deeper position in the paint closer to the rim and b) it's easier to rebound off midrange shots than shots at the rim or threes.


However, drop coverage does mean that you are leaving a certain type of shot open: How else would you entice players to take a shot that everyone in the game now knows is less valuable than a three or rim attempt? And that, in turn, creates variance. Elite off dribble shooters -- as rare as they are -- feast against a drop.


Davidson's shot location data from last year is absolutely beautiful. Opponents took 30.1% of their shots from the midrange compared to just 26.2% from three. Twos are, once again, less valuable than threes. It paid off, too. Davidson allowed a slightly above average 0.78 PPP on off dribble jumpers as a whole, which makes sense. Even though these shots were usually more open, they were often longer twos instead of threes, which mitigated the raw points per possession figure.


Now I want to present Davidson's defensive numbers this year. What stands out?



Teams are still taking plenty of dribble jumpers and midrange shots, and while they've fared decently on attack and kicks compared to last year, they still barely get those looks (13th f %tile). Davidson is one of the best defenses in the country against off ball cuts and inside out play. The extreme switching / no help philosophy is doing its job there by keeping Davidson attached to off ball players and enticing isolation plays like post ups (63 f %tile) and dribble drives (86.7 f %tile).


However, as Davidson has gotten more reliant on drop, one thing sticks out like a sore thumb. Davidson's opponents have managed to produce 1.08 POINTS PER POSSESSION off of dribble jumpers.


Think about that difference. Last year, with a very similar scheme and many of the same personnel, Davidson's opponents were very pedestrian on these shots. Now, they're collectively shooting at the level of a highly ranked WNBA prospect.


I offer three explanations:


  1. Davidson's schedule is much tougher than last year, and we've played more elite off dribble shooters. This is undoubtedly true, but it's not so significantly harder that every opponent should be shooting like a future pro.


  1. Millie Prior was one of the best defenders in the A10 and the PERFECT center for a drop-centric defense. This is also extremely true, since centers with a massive wingspan are perfect for this scheme. They can sink deeper and still contest the shot of someone coming off a screen (see: Ryan Kalkbrenner). However, that implies Davidson's current centers are incompetent defensively... and Ines Garcia is a highly gifted, albeit slightly shorter defender. Edina Strausz is also a gifted defender, without Garcia's incredible mobility but with a more comparable wingspan to Millie. So while this would explain a slight regression (you do have multiple freshmen stepping into a quite complicated scheme), I don't fully buy that it would do something this dramatic.


  1. Davidson has been incredibly unlucky. 1.08 PPP is a simply ridiculous and unsustainable number, even if it's a shot the defense is designed to allow. Bad variance happens in basketball all the time, sometimes for a while, and you just have to keep at it and look at the body of work that existed before to get the whole picture.


I buy a combination of the above, but mostly number 3. Allow me to inject some positivity.


When Do the Turn Tables Turn?


Once again, the eye test says that Davidson's opponents have routinely made very difficult shots all year long. The Cats' combination of switching, forced isos, and drop coverage made them a prolific 3-point defense for the three seasons before this one. Am I supposed to believe that the same scheme that allowed 10+ made threes only once in 81 games is now getting diced up, despite producing very similar shot location data with the same highly skilled coaching staff and players? I don't think so.


Just look at the raw shooting percentages. Davidson's opponents shot 26.2% from three last year (3rd nationally in % against) and 35.7% from the midrange. This year, they're at an insanely hot 33.2% from three (307th) and 39.8% from the midrange (339th!!!).


Despite all this insane shooting from opponents, I have to reiterate that the Cats' adjusted defensive rating is better this year than last. That's because the Cats are an ELITE defensive rebounding team (28th nationally) and adept at turning opponents over (25.3 TOV% against, 84th nationally). That makes for very good defensive shot volume. Davidson had a very good one last year, but this year's is even better. When you combine the fact that opponents get fewer shots than Davidson with the fact Davidson almost always puts up more threes than their opponent, you create a massive math advantage. (But, once again, I've already written about that.)


When you have this numbers advantage in every game, you lose if your opponents are nearly perfect with the shots you give them. And, unfortunately, Davidson's opponents have often been so. Miami went 8-16 on midrange shots; NC State went 13-24. On unassisted threes, Richmond went... well, I don't want to think about that game anymore.


The important thing is that Davidson is still preventing opponents from shooting the most valuable threes -- open, catch-and-shoot ones. Keep playing smart basketball and karmic justice will be served. Davidson's defense might end up being much better than the number 44 sticker indicates.


Will it be served this season? Will I be treated to Maggie Doogan going 1-7 on unassisted threes against Davidson on Sunday at the A10 tournament? I sure hope so. I would love nothing more than to crush our rivals' souls. With the all the hard work put in to build such an intelligent scheme, and all the defensive discipline that the coaches have instilled, it would be well deserved. The team has won some big games over the past couple years, but they are so painfully close to the ultimate breakthrough.


I believe it will happen. If not this time, then very soon.

 
 
 

Comments


bottom of page