Let's Chart Some Threes Because We're Nervous
[JD Scott]
Michigan's offense during the opening weekend of the tourney was… not great. Michigan failed to hit a point per possession in either game. A sloppy, weird, late-night game against Montana with 14 turnovers rather explains itself. The Houston game not so much.
While Houston does sport one of the country's better defenses, Michigan turned the ball over just seven times. They shot 45% from two, nine points under their season average; they were a grim 24% from three before Jordan Poole's miracle pulled them up to 27%. If they'd hit their not-very-impressive season average of 36%, the end of that game is Michigan putting the clamps on whilst up 6-8 points.
Game-to-game shooting variance is of course the very heart of basketball but I wanted to see if Houston had done anything that warranted that kind of performance or if it was just one of those things. So I started poking around and got quite deep in the weeds, because quantifying what's a good three pointer and what's a bad one is tricky. But I'm willing to give it a shot after checking out this paper from the 2014 Sloan Conference. It uses NBA data to create a model of what a good shot is; that model is way beyond the scope of this post but there were a couple of graphs that confirm everyone's eye test.
The first: catch and shoot is better than off the dribble.
Second, and possibly counterintuitively, every foot matters when you're closing out.
As the paper authors put it it, "it is not simply a matter of a shot being “contested” or not but ... there is significant marginal value in every foot of space between the shooter and the closest defender." I wouldn't necessarily have expected that. (Also, I assume that the big uptick in long jumper eFG when a guy is in your grill is because he's fouling you.)
These are NBA numbers but there's no reason to expect that college basketball players would deviate from either of these assertions. So, here's a bunch of three pointers charted. Spoiler alert: the large majority of the attempts Michigan got off were good looks with reasonable space between the shooter and the defender. A fairly typical look:
| Y/N | Length | Shooter | State | Defender distance | Contest | Notes |
|---|---|---|---|---|---|---|
| MISS | 22 | Mathews | catch and shoot | 3 | light | Late clock jack is surprisingly clean look |
| MISS | 23 | MAAR | catch and shoot | 2 | heavy | Curl screen gets vg closeout from Houston defender |
| MISS | 24 | Simpson | catch and shoot | 6 | none | PNR switch w big sagging off Simpson, no contest |
| MISS | 22 | Livers | catch and shoot | 3 | moderate | drive and kick rhythm three from wing |
| MISS | 22 | MAAR | off the dribble | 4 | light | switch confusion for UH gives MAAR opportunity to gather and go straight up |
| MAKE | 28 | Robinson | catch and shoot | 3 | heavy | late clock deep jack is worst look of game so far, goes down |
| MAKE | 24 | Robinson | catch and shoot | 7 | light | V-cut and Teske screen gets Robinson clean look |
| MISS | 22 | Wagner | catch and shoot | 3 | moderate | pick and pop open-ish, decent contest |
| MAKE | 22 | Robinson | catch and shoot | 2 | moderate | Robinson's defender is lax for a second and DR just rises over him |
| MAKE | 22 | Poole | catch and shoot | 4 | light | penetrate and kick to Wagner, extra pass, open corner |
| MISS | 22 | Simpson | catch and shoot | 8 | none | drive and kick, Poole upfakes and dribbles once to draw a second guy, Simpson wide open in corner |
| MISS | 22 | Poole | off the dribble | 2 | heavy | bad heat check stepback w/ 21 seconds on clock |
| MISS | 24 | Poole | catch and shoot | 4 | light | drive and kick from MAAR |
| MISS | 27 | Robinson | off the dribble | 2 | heavy | late clock, initial contest and one-dribble lanch |
| MISS | 24 | MAAR | catch and shoot | 8 | none | numbers for M, shot fake from MAAR gets wide open look, airball |
| MISS | 22 | MAAR | off the dribble | 2 | heavy | last ditch attempt at end of half |
| HALFTIME | ||||||
| MISS | 22 | Matthews | catch and shoot | 4 | light | Wagner kick to Simpson over for open look |
| MAKE | 23 | Wagner | catch and shoot | 5 | light | Simpson pick and pop |
| MISS | 23 | Wagner | catch and shoot | 5 | light | Near identical pick and pop from same spot on the floor |
| MISS | 23 | Simpson | catch and shoot | 4 | light | Poole drive and kick, token contest |
| MAKE | 23 | Wagner | catch and shoot | 5 | moderate | Pick and pop from left wing, and one |
| MISS | 24 | Robinson | catch and shoot | 6 | none | wing pick from Matthews opens this up. No contest... airball. |
| MISS | 25 | MAAR | off the dribble | 4 | moderate | PNR switch, pretty good contest from the big MAAR rises up on |
| MISS | 23 | Robinson | catch and shoot | 6 | light | Simpson drives baseline and kicks to left wing |
| MISS | 22 | Robinson | catch and shoot | 5 | light | OREB to scramble drill to corner 3 |
| MAKE | 22 | Matthews | catch and shoot | 4 | moderate | Simpson drive and kick; lucky bounce off heel and straight up and down |
| MISS | 25 | Matthews | off the dribble | 2 | heavy | late clock jack is not at all a clean look |
| MAKE | 30 | Poole | catch and shoot | 2 | heavy | THIS IS A BAD LOOK BUT OKAY |
Leaving out the two must-launches at the end of each half and Michigan had just 5 off the dribble looks on 28 attempts, two of those from MAAR in pretty good situations. One was a pick and roll switch with the big playing off, the other an opening when Houston got confused on another pick and roll.
Houston did force five heavily contested late-clock jacks, one of which went down when Robinson hit a deep one. There was also one more heavily contested three as Corey Davis came around a screen really well on an early MAAR attempt. The other 20 attempts I charted were all reasonable to excellent looks that simply didn't go down. Eight attempts from Michigan's worst three-point shooters, Simpson and Matthews, isn't a particularly unusual ratio. Those guys have about 20% of Michigan's attempts on the year. 8 of 28 non-desperation threes is a couple more than you'd expect, but not outlandish.
Verdict: just one of those things. One that happens to a team like Michigan that's not exactly Villanova.
but...it would be interesting to see Z play the middle of the zone. The downside is that he's a small target, but it gets him off the three point line (where teams don't really guard him anyway) and let's him be the distributor, which he's very good at. He can catch it and take a dribble and hit the floater if the bigs don't step up to gaurd him and he can pass it if they do.
I don't think they'll do it, but I'd love to be a fly on the wall in practice if they've tested it out.
the main problem is the person shooting the 3. matthews, simpson and livers 3pt shot % has fallen off a cliff since the beginning of the year.
livers is 2-12 since Feb 1. he was 12-21 in Jan.
matthews 7-27 since Feb 1. he was 8-22 in Jan.
simpson 6-36 since Feb 1. he was 8-19 in Jan.
that's the problem. very simple. those guys need to stop taking 3s unless it's very late in the clock and they have to.
You can't just as 3/5th of a 5 out offense to not take 3s. I mean, they shouldn't be the bulk of the offense, but man, they need to take 3s and need to make them. I noticed against Houston that they were really sagging of Simpson. He's a great driver to the basket and they dared to take 3s and he didn't punish them. Having that help defender sag off Simpson really hurts the ability of other players to take it to the lane. The fact that Matthews and Simpson can't hit 3s consistantly is really what limits this offense.
they just need to start making them. None of them are as bad as they've been since Feb 1 (as you see from what they did in April). You don't change your strategy because of randomness.
Duncan was a pretty bad 10-32 in January. It would have been stupid for him to stop shooting under the assumption that would continue. He's been his usual self since then.
Those three guys are also better than what they've done since Feb 1.
You don't change your blackjack strategy just because you lose a few hands playing by the book. Play the odds and stick with it.
They called it. That was the 5-point play.
and perfect screen grab. Man, that play was huge.
how many teams in the remaining field, shot better their second game from three than their first. And if thety did, how did that performance differ from seasonal averages and prior tournament numbers in second day games.
Texas A&M: 10/24 for 42% which was better than their first game (38%) and season avg (33%)
Gonzaga: 9/22 for 41% which was better than their first game (22%!!) and season avg (37%)
FSU: 8/23 for 35% which was worse than their first game (47%) and right on their season avg (35%)
Kentucky: 7/15 for 47% which was better than their first game (0%!! on 0/6 shooting) and better than their season avg (36%)
Kansas St. 1/12 for 8%!!! which was worse than their first game (47%) and their season avg (34%). Nice to play a 16 seed so you can score 0.76 points per possession and win.
Nevada: 6/18 for 33% which was worse than their first game (38%) and their season avg (40%)
Loyola: 8/20 for 40% which was better than their first game (38%) and right on their season avg (40%)
Michigan: 8/30 for 27% which was worse than their first game (31%) and seeason average (36%)
So it's exactly as one would expect from random trials with no trend: four teams shot better the second game, four shot worse.
Three shot better than their season average, three shot worse and two were spot on.
Totally random, people.
Of all those teams, Michigan shot the most. I feel like we were driving a lot more at the end of the year and have settled for jumpshots in the NCAA tourney. Maybe they were hoping outside shots would open up the drives but may need to drive to open up the outside shot (and shot rhythm).
I think the weekend boils down to a combination of: 1. Randomness being an issue, most teams are going to have some mediocre games shooting the ball; and 2. Michigan not being loaded with deadly 3-point shooters this year.
We have some guys who are good. We have one guy you can argue is great. But this is not a vintage Michigan team when it comes to shooting, and I feel like we're more vulnerable to this kind of thing with the current roster. Robinson is the only guy you expect to be really good, and while he usually is defenses don't have to worry about him creating his own shot.
We have guys that CAN shoot but also can be off. We have guys that CAN be dangerous off the dribble but can also get shut down. We don't have killers on offense.
And that's fine, this is baked into the cake. Michigan CAN get hot but they can also win when the 3s aren't falling. They did so twice this weekend. Because we have killers on defense.
I dare you to chart the MSU-Syracuse game. 8>)
charting the MSU game would even more strongly result in, wow, that was some crazy bad luck. MSU missed sooooo many wide open, easy threes. MSU had a three point shooting night that had a 1.5% chance of happening or being worse (assuming each trial had a 40% chance of success, their season average, which is reasonable given all the threes were taken by good three point shooters and a normal or higher proportion of them were "good looks").
Total outlier event. Not a lot a coach can do when you get your really good 3 pt shooters a bunch of good looks but they just don't go down.
I didn't read through all the comments so appologies if this has been discussed. Any chance it is the type of ball they are using? I thought I heard the ball they use for the tournament is different thean what is used all year (most places, as schools use their own balls during regular season). I heard this brought up when a couple announcers were talking about how many balls have been wedged against the rim this tournament, saying the ball used for thr tourny is stickier than that of those normally used. I never played any basketball at all so maybe this has little to no affect, but possibly either the ball itself being more sticky or the players just not as used to it.
Can we talk about why NBAers are taking shots from 4-8 ft away when the defender distance is even further? Wouldnt that imply the path to the basket is open and they could go in for a dunk/layup? But instead they are taking short J's and making only 50-80% of them
March 22nd, 2018 at 12:57 PM ^
and they're probably mostly end of shot clock/game clock cases in which the shooter didn't have time to get closer.
March 22nd, 2018 at 10:00 AM ^
if you sag off one guy, you can pay more attention to the others.
And compounds if another, nrmally dependable guy is off. Then the whole offense looks discombobulated.
Comments