Will Michigan's "three point defense" improve?


Pomeroy’s conclusion absolutely fits common sense. The more open looks a team gives up, the more threes an opponent will take. Therefor, over a season, it’s a much better way to measure how good a team is at preventing open looks, which is how one would measure a team’s three point defense. I mean Pomeroy and the other links I’ve provided prove this statistically, showing that there is a correlation but it’s less than 2%. Unless you have something math wise to back this up then I don’t see how this is controversial.

Also I don’t know what your point about MSU is trying to say. First of all I’m pretty sure your numbers are wrong. They didn’t lead the Big Ten either of the years, the were 2nd in 2015 and 3rd in 2016. I assume you’re talking only Big Ten play since earlier you made a point about how his conclusion was wrong since he counted non conference too which has variable schedules. The difference between MSU and the worst 3PT% defense those two years was 6% and 7%, respectively. With margins that small I 100% believe it was luck that Michigan State opponents happened to shoot 3% worse than the mean those two years.


Are your MSU numbers in the big ten? I just quickly did stats for all games from the big ten website…


I touched on that in my post a bit.


No. I used stats for the whole year from the big ten website.

I am not trying to disprove Pomeroy. I am just saying it “smells” fishy…

It still smells fishy…

15 years ago everybody was showing how smart they were by reciting the “hot hand fallacy”…People rarely do that anymore…and for good reason…


Look, I’ve said, and Pomeroy agrees, that a team’s 3pt% against is slightly affected by their defensive ability… But I’ll concede a 2% difference and 1% above the mean on that. I simply believe it’s logical that the best three point defenses give up the fewest attempted threes, as the more open looks a team has the more they will shoot it. They 2% is what we were talking about with not getting hands up once they start their shot and the other issues we have. Everything else is just luck. I think you can agree with me that 41.5% against is not even close to sustainable.

Here’s the last thing I’ll say… I think everyone on this message board would agree that our defenses in the 2015 and 2016 seasons were just as bad as this year. Despite that, we were very middle of the road and actually above average in terms of opponent 3pt% at 34% against. Just luck.


Not all contested shots are contested equally.

The notion that an outside observer can reliably quantify proximity to a shooter as he releases the ball and how that proximity did or did not affect the shot is just wrong. To a large extent the shooter is in control of proximity at time of release. Good defenders rush shooters shots. Good defenders take away the luxury of time and space. Just because I stop someone from shooting a 3 does notmake me a good defender of 3 point shots–that is an absurd fragmentation of defense as a whole that no coach, who wants to play good all around defense, would subscribe to.

The lame way Walton and MAAR have contest 3’s this year is a 3 point shooters dream. No amount of reading Pomeroy’s theory (that it is mostly luck) will quell a shooters glee when they realize that Walton or MAAR are closing out on their 3 point shot…


Michigan has been unlucky so far this year. Sometimes it is hard to believe in luck or statistics. We don’t want to believe that things are haphazard.

Nobody is arguing that UM’s defense is great or that the closeouts are awesome or that people aren’t getting lost on screens. But with six minutes left in the game, Michigan down by four, 30% 3pt shooter freshman Anthony Cowan takes a step-back off-the-dribble 3 pt with Walton doing some form of contest. If you’d shown that shot at the beginning of the possession to Turgeon and Beilein and asked if they’d take it or would play a new possession, it would be an easy decision for both of them. If you’d shown NIckens shots (but not the results) to Turgeon and Beilein before the game and offered 50% (2 of 4) to both, it would have been an easy decision for both.

This doesn’t mean Michigan is losing because and only because of bad luck. It means that they’ve played well enough to be in position to win their B1G games but bad enough that some bad luck could swing some of the results. If they played those games 100 times at exactly the same level, sometimes they’d be 3-0, sometimes 0-3, sometimes somewhere in between, based purely on luck. My guess is that the most common result would be 2-1, but who knows.

Luck tends to even out over time and even over a season, although of course not always. I’m still hoping Michigan’s luck improves over the next sixteen games. But I’m also hoping they play better.


I will continue to pray that Walton and MAAR do not read and believe that there is much truth at all in Pomeroy’s claim.


Why? This doesn’t mean they should play bad defense. In fact, a proper understanding should lead to better defense and less 3s – for instance, running Nickens off the line or not letting him take 3s at all, or getting up in Cowan’s grill with 4 seconds on the shot clock so that he is forced into a tough 2. Acknowledging some bad luck doesn’t mean it’s okay to get caught on screens or make lame attempts to swipe at a ball while leaving their players open on the 3 point line.


Is any of this a by-product of the Pack Line defense?


This is so right. 28 seconds of good defense forced the Cowan and Trimble pull up 3s; they weren’t contested perfectly; them making both, or even 1, is unlucky, even if we could have defended it better.

Duncan was culpable for Nickens open looks, that was bad defense. But we were still unlucky that he went 4/4. We are at ‘fault’ for not eliminating the possibility of those shots being made, but we are also unlucky that a staggering percentage of 3pt shots are being made. Especially considering Iowa, PSU, and Maryland aren’t good shooting teams.

  1. I don’t believe that the quality of 3 point defense as a subset of defense on the whole can be measured by how well a team “runs” guys off the 3 point line. For the sake of argument let’s call running a guy off the 3 point line “excessive pressure”.

  2. I don’t believe proximity to a shooter was quantified properly. It would be very difficult to quantify proximity in a meaningful way. To an outside observer proximity resembling pressure would be hard to distinguish from proximity which forced a shooter to rush his shot and change his form.

MAAR and Walton look to me like players that create proximity to a player merely resembling pressure. Am I able to quantify it? No.

A good 3 point defense will occupy that “space” between pressure which merely resembles pressure (too little) and pressure which runs a 3 point shooter off the line (too much).

My intuition tells me that the best 3 point defenses would result in teams taking an average number of attempts against them, while having their opponents shoot a low percentage. (See MSU the last 2 years.)

It’s a delicate balance between too little and too much…

I have my doubts for other reasons too.


So you disagree with the premise that the best three-point defenses take away attempts?

FWIW, I do think the pack line defensive principles could be a big part of taking away 3PA. Wisconsin teams have always done a really good job of this (and their 3P% def rank has fluctuated between 8th and 308th in the last five years while their 2PT D has always been top-100).


As a coach, I would find it to be a red flag if opposing teams were taking too many threes. I would also think it was a red flag if opposing teams were taking too little threes.

I am not in a position to quantify this stuff. I am not sure anyone is. I think we can all agree that it is undesirable to just give open 3’s. I also think we can all agree that it is undesirable to run 3 point shooters off the 3 point line at all costs. So, yeah, I don’t think adopting a practice of limiting 3 point attempts(at any cost) is necessarily desirable at all.

Has Pomeroy quantified proximity in terms of too much and too little and how it affects PPP?


Why is this undesirable? This is what most defenses are designed to do? Not sure what exactly you are arguing now.


It gives driving opportunities.


Well yes, but that would be incorporated in 2-point defense.

If you compare this year’s defense to last year’s, you see that the two-point defense has improved from 51.1% to 46.4%, but the three-point defense has regressed from 34.7% to 41.4% with 3PA% dropping from 36.4% to 29.6%.

Theoretically, better 2-point FG% defense while allowing fewer three-pointers should be a net gain. Obviously the ridiculously hot three-point shooting is proving that to be wrong.

The question this thread is asking is whether those numbers will regress toward the mean.

I’m not sure either way, but I think KenPom’s argument is pretty solid logically.


More undesirable if you are a help defender. You could be leaving a man 5 feet from the basket to fly out (often too late and/or too out of control) at an open player 20+ feet from the basket.


My strong hunch is that the almost impossible to quantify space between too much and too little defensive proximity is where good 3 point defense lives. Not coincidentally, I think the almost impossible to quantify space between too much and too little defensive proximity is the space where Pomeroy is claiming luck lives.


That would be the 1.9% that kenpom says is the difference in percentage between good and bad three point defense. It’s there, it’s just mostly negligible.