Will Michigan's "three point defense" improve?

Do you know the categories he uses to express different defensive proximity?

Not sure what you are saying exactly, heā€™s not measuring defensive proximity.

The whole article was posted above which you can see here (itā€™s free)
http://kenpom.com/blog/3point-defense-should-not-be-defined-by-opponents-3p/

Iā€™m not sure, but I think the problem here is that you think weā€™re arguing that defense does not affect the likelihood that any particular three point shot goes in. Thatā€™s not what weā€™re saying. Not all 3s are created equally. We know only too well that Zak Irvin is way more likely to hit an open catch and shoot 3 than a contested off the dribble 3.

What we are saying is that good three point defense over the course of a game and season will most accurately predict 3 pointers taken. Why? Offenses want open 3s and lay-ups/dunks, the most efficient offense. Take those away and the offense is left with other things ā€“ some of which will be contested 3s, but many of which will be long 2s or contested 2s. Thatā€™s why ā€œa lot of long 2sā€ or ā€œgood 2 pt defense and a low amount of 3sā€ are generally recipes for good defensive outcomes.

But Maryland shot 18-39 on 2s (46%), and 15 3s; Iowa shot 22-50 on 2s (44%) and 19 3s in an overtime game. These arenā€™t dominating defensive performances but play those games 100 times and more times that not MD and Iowa will score less than they did in the two games just played ā€“ and would not be performances that would cause Michigan to drop 40 spots in kenpomā€™s defensive rankings.

So with Walton defending Cowan ā€“ sure his defense played a role in how likely it was for the shot to go in. Hell, if Anthony Davis is out there maybe the shot gets blocked altogether. But the fact that the shot went in doesnā€™t mean that it was a good shot or even necessarily bad defense - it may have been. It doesnā€™t mean that Michigan didnā€™t get somewhat unlucky that it went down.

And I think proper communication and understanding of these principles could help Michiganā€™s players ratchet up their intensity throughout the game. They have to understand that even if they play good defense late in the game, some of those shots are going to go in ā€“ in fact, sometimes a lot of them will. Theyā€™re not good enough and luck is haphazard enough that they need to play every possession (or at least the vast majority of them) with focus and intensity if they want a high probability of a good season.

1 Like

He just looked at it statistically to use only the evidence at hand, not what one would theoretically expect. The other links I gave had more stuff on proximity but they said thereā€™s really no difference in wide open and semi open or contested and really in your face.

Look, it can be boiled down to this: sure, there may be a correlation to how good a teamā€™s 3pt defense and an opponentā€™s three point percentage. There may be, and the evidence suggests itā€™s very small. However, we know there is a correlation between three point attempts given up and how good a teamā€™s three point defense is because LOGIC. In every way possible it makes sense that teams shoot more threes if they have more open looks. With that in mind, thereā€™s absolutely no reason to use a stat that is mostly luck, this is not debatable, and may or may not reflect how good a defense is when we have something like 3PA/FGA which definitely does.

1 Like

A good defense isnā€™t going to affect how a team shoots on open 3pt attempts. A good team will limit the amount of open 3pt attempts taken.

There are exceptions to the rule, which he covers in his articles, where some teams are good at forcing teams to take contested 3pt shots (he cites Kentucky and Baylor). In general, though, offenses wonā€™t force 3pt shots when they are being guarded tighter..

KenPom provides the example of a ballscreen. If a player gets a ball screen and is open for the 3pt attempt, he often takes it. If heā€™s guarded, however, he is more likely to PASS or DRIVE out of the screen instead of take the shot. This is common sense. Saying otherwise defies playing experiences..

1 Like

Say Team A shoots 20 3pt FG per game at 40%.

If Team A plays Cupcake team B that doesnā€™t close out well, they are likely to shoot an increased number of 3pt FGs. They get an increased number of open looks, but still shoot at a rate of 40% on those looks.

If a Team A then plays Elite team C that closes out well, they are likely to shoot a decreased number of 3pt FGs. They get a decreased number of open looks, so they take less 3pt shots, but still shoot around the rate of 40% of the shots they do take.

A good defensive team means decreased 3pt FGs, not decreased 3pt%. A bad defensive team means increased 3pt FGs, not increased 3pt%.

1 Like

"To say that the defense has little control over defensive 3P% is probably not the most precise way to put it. If a defense stopped guarding the three-point line or decided to never contest a shot, they would surely see their numbers suffer

To illustrate why a team might have little control over its opponentsā€™ three-point shooting, consider a ball screen situation. If a defender goes under a ball screen, the ballhander, assuming heā€™s a good shooter, will be inclined to shoot. If the defender goes over the screen, the ballhandlerā€™s response will not be to shoot a more difficult shot, it will be to not shoot at all. In this way, defensive strategies tend to impact the number of threes taken and not the percentage of threes made. By the end of the season, opponents have taken a mix of open and contested shots based on their own decisions, and from the defenseā€™s point of view the distribution of these decisions isnā€™t going to differ much from team to team. Thus, the resulting rankings of defensive 3P% are largely random, influenced some by opponents shooting ability." - Ken Pomeroy

2 Likes

What sucks is ā€˜usingā€™ ā€˜good luckā€™ against ucla for a half. (Sorry for the ā€˜ā€™-Iā€™m big on air quotes).

Consider this: Could it be that the truly lucky stat here is Michiganā€™s 3FGA/FGA?

Weā€™ve talked for some time about how Michigan is doing a good job of preventing 3FGA, but what exactly are they doing to make that happen? Common sense says nothing. I mean, look at the shots Dylan clipped in his power rankings story. I know itā€™s just one game, but that hardly looks like a team thatā€™s strong at denying three-point attempts.

It could just be that A) Michigan happens to have played several teams (like South Carolina) that donā€™t shoot the 3 much and B) opponents are just choosing to attack us inside rather than shoot from deep. After all, if any of us were game-planning against Michigan, the last thing youā€™d advise is shooting a bunch of 3s. You would attack attack attack because the perimeter defenders lack quickness and there isnā€™t an eraser down low.

The thought is that you affect 3PA/FGA because players usually shoot open threes a lot more than guarded threes. Those (made) threes were definitely open for the most part, but thatā€™s because they shot them, right?

The other thing that doesnā€™t really add up about that theory is that teams have been far less effective inside the arc this year. I think the lower 3PA/FGA is coming from some of the defensive shifts that Donlon has implemented.

1 Like

Yeah, youā€™re a lot better at explaining this than I am.

ā€œIt couldnā€™t be much simpler than it is right now,ā€ Beilein said this afternoon. ā€œWeā€™re not switching any screens. Weā€™re doing a classic hedge. Weā€™re tagging guys. Weā€™re probably doing less than weā€™ve done before.ā€

Very unlikely. First, if your theory is correct, the low 3FGA/FGA isnā€™t luck, itā€™s a rational reaction by actors (opposing players) to a situation (our bad defense).

Second, could double-digit teams and their dozens of players have all chosen to forego open 3s for (a) lay-ups, but then missed a significant percentage of them, or (b) long two-point jumpers? I guess. But it is far more likely that these teams and players have collectively acted in a way consistent with how players and teams generally act.

Some contested 3s go in anyway. Some open 3s donā€™t. To some extent which happen against you is luck. Generally that evens out over a season. But not always.

Itā€™s interesting how hard people will try to argue against this. People really hate the idea of chance.

2 Likes

Wait, I think youā€™re looking at the stat for last year. This year weā€™re listed 5th in the conference in free throw percentage against.

Yep ā€“ my bad.

Who says our defense isnā€™t improved :smirk:

3 Likes

Noā€¦i donā€™t think it will

terribleā€¦

My opinion has not changed, but Michigan simply breaks any statistical laws.

2 Likes

Oh, no. The studies speak for themselves. Iā€™m thinking Michigan might be an outlier/exception to the rule.

In the way KenPom says that Kentucky and Baylor are exceptions because they force a lot of contested 3s, maybe Michigan is an outlier on the other end of the spectrum.