I prefer the shot on the left. So does Beilein. [Marc-Gregor Campredon]
After the Nebraska evisceration, I wanted to take a closer look at something we discussed on this week's podcast. Michigan generated 12 more three-point attempts than the Huskers, which added to the growing pile of statistical evidence that the Wolverines have undergone a fundamental shift—not on offense, but on defense. John Beilein gave the money quote on it after the Purdue game:
We’ve made a conscious decision to defend the three-point line, knowing that a tough two is much better to give up than an open three, which we were giving up like crazy in our earlier struggles.
The key number to look at is 3PA/FGA: the percentage of each team's field goal attempts that come from beyond the arc. The offense is shooting threes at the usual Beilein offense rate: 45.3%, 16th nationally. Before this year, Beilein's Michigan defenses haven't been good at preventing opponent three-point looks; his best finish in 3PA/FGA was 108th in 2014, and most of his M teams have been in the 200 range.
This season, Michigan opponents are attempting just 29.0% of their field goals from beyond the arc. That puts the Wolverines tenth in the country.
The shift in defensive philosophy, likely a product of adding Billy Donlon to the staff, has created a massive gap in points generated from the three-point line between Michigan and their opponents. Critically, the Wolverines aren't forcing shots to make it happen. I put together a video of Michigan's three-point attempts (two garbage-time attempts excised) against Nebraska with freeze-frames just before the point of release; there are only a couple questionable shots among the 25:
I did the same for Nebraska's shots. While they had a few wide open looks, Michigan did a much better job of closing out on Husker shooters than vice versa, and that's not even the most telling part of this video—that would be the length of the video itself. What's not in there is the number of times Michigan defenders ran potential shooters off the line, forcing them to take those tough twos instead.
Even if Nebraska had hit their open looks, they had little hope of keeping up with Michigan's offense. Their second three-point attempt of the game came with under five minutes left in the first half; by that point, M had opened up a 20-point lead while shooting 8-for-12 on triples.
As conference champion Purdue found out, it's hard to close the three-point gap on Michigan with two-pointers, even when they're going in at a relatively high rate. It helps, of course, that Beilein's offense also generates great looks inside the arc; Michigan is 12th nationally in two-point percentage. This leaves opponents in a bind. Do they try to match Michigan three-for-three, even though the Wolverines have superior shooters to almost any team they face? Or do they run their normal offense and hope to either hit twos at a remarkable rate or get an off game from Michigan's shooters?
I'm not sure there's a good answer.
[Hit THE JUMP to see the numbers behind the three-point gap.]
The Defense, For A Given Definition Of The Term
Slicing through M's defense with little resistance. [Marc-Gregor Campredon]
Do you have a stick? Throw it. Congratuations, you have hit a horrifying Michigan defensive stat.
The Wolverines may have pulled out a victory against a Nebraska team playing without its only viable post player, but they didn't do it by solving any of their problems on defense; the Huskers scored 1.21 points per possession, a hair below the average performance against M's defense in conference play. Michigan is now 185th in adjusted defensive efficiency; their worst finish under John Beilein was 120th in his first year in Ann Arbor.
Through five conference games, Michigan has the worst Big Ten defense by 8.9 points per 100 possessions; B1G opponents are making 52.7% of their twos and 55.3%(!!!) of their threes—and they're rebounding 34.7% of their misses. Michigan is great at not fouling and above-average at stealing the ball; they're somewhere between below-average and terrible at everything else.
Dylan has a post today that goes into further, gruesome detail on Michigan's defense, with one area of focus being the collapse of their pick-and-roll defense:
Michigan’s pick-and-roll defense has completely fallen apart. In the last six games, the Wolverines have allowed .986 points per possession (including pass outs) in the pick-and-roll game. Compared to seasonal numbers across Division I, that would rank 336th nationally.
Only the first half of the Nebraska game is available on the YouTubes, which is probably for the best. This actually came out better than I expected and it's still far from good:
The issue, as Dylan mentions in his post, doesn't appear to be the scheme; no matter how Michigan approaching defending the high screen—usually either with a soft hedge or ICE technique—they're allowing baskets because of individual player breakdowns. Passes into the post, like in the first play, are too easy to make. Blown rotations, like in the second, lead to wide open three-point attempts. Michigan commits the cardinal sin of allowing the P&R ballhandler to split the hedge at the 0:34 mark, something that occurred at least once more in the second half.
They did a little better towards the end of the half, as you can see in the video, but I also forgot to include this abomination:
It was more of the same in the second half. There are two common threads: Michigan has zero rim protection, which allows opponents to attack without fear, and their help/rotation off the ball is awful. I grew up on the suffocating team defense of the mid-aughts Pistons. This is the opposite of that. The problems are so widespread that it's impossible to suggest one or two solutions that could turn things around.
[After THE JUMP: That said...]
MAAR probably gets a pass for not contesting this one.
I regret responding to this with "that's easy enough":
I went back through the UCLA game and charted each three-point attempt by both teams save for the last couple minutes of garbage time. The no-late-heavy shot contest system is relatively self-explanatory and looks at how well the defender guarded the shot attempt. Heavy contest shots, especially from beyond the arc, are bad ideas; late contest is enough of an opening to get a good look but isn't completely wide open; no contest is wide the hell open.
When breaking it down by halves, the story of the game emerges:
|No Contest||Late Contest||Heavy Contest|
|Michigan (1st Half)||3/4||7/10||2/2|
|Michigan (2nd Half)||0/1||2/6||0/1|
|UCLA (1st Half)||5/6||5/6||0/2|
|UCLA (2nd Half)||3/3||3/4||0/2|
I expected a bit more NBA Jam (i.e. drilling heavily contested shots) in UCLA's first-half results; instead, I saw a series of errors that led to good looks, and those errors got way worse in the second half. Meanwhile, Michigan's offense stopped generating easy looks beyond the arc in the second half at the same time they cooled off on tougher shots.
[Hit THE JUMP for blood, oh god, so much blood.]
As time winds down on the regular season, Michigan finds itself squarely on the bubble – the classic meh major-conference team that gets sent to Dayton as an 11-seed. Big Ten play is almost over and we know the Wolverines were a middle-of-the-road team (as of right now: 7th in Sagarin’s and Pomeroy’s ratings, 7th in conference efficiency margin, and are likely to tie Ohio State as the 7th place team in the league). When your best resume asset is that you haven’t lost to any bad teams, it hasn’t been a great season. The very real possibility that Michigan misses the NCAA Tournament would qualify this year as a big disappointment.
Still, even though Caris LeVert and Spike Albrecht have combined to play just 48 minutes against Big Ten opponents, Michigan will finish with a winning record in Big Ten play and (probably) a positive efficiency margin in league play. This season’s path was very comparable to last season’s: the teams had uninspiring non-conference performances and Caris was lost near the beginning of league play (and Spike and Derrick Walton were injured this year and last, respectively). A year ago, the Wolverines finished 8-10 in Big Ten play – losing four overtime games and winning one – and this year, 10-8 or 11-7 will be the final result. Since Michigan didn’t drop a stinker like NJIT or Eastern Michigan like last season, we’re going to be nervously watching on Selection Sunday – can 3 good wins and a bunch of chalk get us in?
The reason why Michigan hasn’t been better is fairly obvious.
I don't expect Michigan to be actually good at defense for a lot of different reasons, but there's a difference between Michigan's usual meh and this. The trend is worrying. Defensive efficiency in the Beilein era:
- 2008: 100th
- 2009: 69th
- 2010: 58th
- 2011: 37th
- 2012: 61st
- 2013: 48th
- 2014: 109th
- 2015: 107th
- 2016: 145th
This is the third straight year of a triple-digit ranking. While you may remember things as "not good" even when the larger picture was much prettier, this is a whole new era of ineptness only matched by Beilein's first team of castoffs and runaways. This year's team is in fact considerably worse despite than those guys despite having a reasonable amount of experience. For the first time in a while Michigan doesn't have a freshman playing major minutes; for the first time in a while they've crawled out of the 300s in Kenpom's experience stat. This was the first year in a while you could reasonably expect year to year improvement, and yet.
Big Ten 2-pt defense:
1. MSU (41.4%)
13. Rutgers (52.4%)
14. Michigan (55.3%)
WORSE THAN RUTGERS
— Alex Cook (@_ac616) February 29, 2016
When you’re worse than Rutgers at something as critical as 2-point FG % defense, you have a major problem.
[After the JUMP, a lot of graphs]
This appears to be an effective hedge. [Fuller]
Brian directed me to an excellent Vantage Sports article detailing how NBA teams defend the on-ball screen earlier this week and suggested it would be a good idea to take a closer look at how Michigan does it. Before getting into the Wolverines specifically, a look at the three basic ways to defend this:
- Hard Hedge — The way M's done it the most under Beilein. The defender guarding the screener (usually a big man) aggressively slides out on the ballhandler to cut off a drive to the basket and make quick passes more difficult. This temporarily commits two defenders to the ballhandler and usually requires quick rotation from the other defenders on the court.
- Soft Show — A less aggressive approach that still briefly commits two defenders to the ballhandler, in this case the defender guarding the screener moves next to the screener, cutting off a drive directly to the hoop; he doesn't move all the way out on the ballhandler, however, and dives back to the screener after cutting off the initial drive. This still requires some weakside rotation.
- Drop Back — The conservative tack. The defender on the screener drops back (surprise!) into the paint, discouraging the ballhandler from driving while also lessening the burden on other defenders to rotate onto the roll man. This does require the defender on the ballhandler to fight over the screen well, otherwise there's room for a pull-up three.
As best I can tell, college teams favor the more aggressive approaches. This is likely due to two things: pro point guards are really damn good, and there's less space inside the arc to cover in college, making it easier to recover after a hard or soft hedge.
I went through the last three games—Rutgers, Wisconsin, and Nebraska—to see how Michigan defended the pick-and-roll. I found nine instances in which Michigan was in man defense against a P&R*; six times they hedged hard and three times they played a soft show. The results:
A few takeaways with picture pages after THE JUMP.
On the surface, Michigan's defense shouldn't have experienced the falloff it has this season. While Michigan's young, they're actually a bit older than they were last year. Mitch McGary has not been available, but there has been a groundswell of semi-indignation at Jordan Morgan's omission from the Big Ten's All Defense team.
But backslid they have. Last year's Michigan team finished the year 48th. This year's #48 defense is giving up 97.2 points per hundred possessions, adjusted for schedule. Michigan is well short of this number, at 100.6.
You'll note that this isn't actually that much. Michigan's about 6.6% worse on their possessions this year. The average NCAA defense is in fact 4% worse than last year, what with the rule tightening and virtual elimination of charges. A big chunk of the backslide is everyone's backslide. The rest, well…
The McGary Factor
watching the tourney run prompted this section, yes [Eric Upchurch]
Michigan entered last year's NCAA tournament 11th in the Kenpom rankings. Unfortunately, Kenpom doesn't keep individual running O/D rankings, but Michigan's surge to 48th on D and fourth overall coincided with Mitch McGary beasting up in the tourney. Michigan held a selection of very good teams to under a point per possession. They faced the #32, 21, 34, 12, 29, and 4 offenses in the tourney and held them to 0.97 points a trip.
McGary rebounded everything and stole everything. Michigan kept in contact before their late surge against Kansas thanks to his 14 rebounds. He picked up three steals, as well. McGary had five(!) steals against Florida and 12 rebounds against Syracuse. Jon Gasaway was tossing out stats I can't quite remember but were pretty much "Mitch McGary's DREB rate in the tourney is ALL THE REBOUNDS."
But that was five games. Before that McGary had been limited for much of the year. His impact on the stats is far smaller than his impact in our minds. If you're looking for a reason Michigan's not going to run to the national championship game again, he applies. In a discussion of why Michigan's statistical profile on D is grim he's not a primary driver.
Transition defense is a primary driver, probably the primary driver.
You've probably eyeballed this whilst exclaiming AAAAARRGGGHHH during the year, and your intuition is borne out by the stats. Michigan's actually been fine at preventing transition possessions—defined as shots in the first ten seconds of the shot clock—but they've been a lot worse at preventing dunk-and-open-three city.
This is partially because shots have migrated from two-point jumpers to shots at the rim and threes. They've also been considerably worse at preventing teams from both high-profit areas. While some of this is the new rules emphasis, transition is the part of the game where that has the least impact. Hoop-math doesn't have overall trends, unfortunately. Nor does it fold in free throws. Oh well.
With what we have to work with we can figure that a just over a fifth of Michigan's defense has gone from 1.08 PPP to 1.24 PPP. That is most of the statistical decline right there.
|Morgan committing a block under 2014 rule-type substances. [Eric Upchurch]|
The Insane Near-Abolition Of The Charge
There was a ton of speculation as to whether the new rules would help or hurt Michigan. Survey says: probably both. The good: offense takes off, foul trouble becomes more prevalent without touching Michigan, and Michigan's excellent free throw shooting is more prominent. The bad: Michigan's primary way to defend the rim has become more fraught with peril than ever.
FTAs have gone up nationwide, of course, and Michigan remains one of the country's least foul-prone outfits. They've dropped from first to third in that department. While that doesn't seem like a significant move, remember that thing I said in This Week's Obsession about how things tend to get stretched out at the ends of these Gaussian-ish distributions. Michigan's FTA/FGA allowed last year was preposterous 22.7, 13 points lower than the national average. This year FTAs are about 13% more common nationwide. Michigan is seeing opponents shoot 23% more FTAs.
If Michigan was in the middle of the pack that effect would feature a 40 spot dip in FTA/FGA; since Michigan was the nation's best by some distance a year ago it looks like they're basically the same. They are not.
Most of this is Jordan Morgan clutching his head and shooting imaginary eye lasers at the refs. His fouls per 40 minutes have leapt from 3.5 to 5.3, and one dollar says almost all of that is the charge random number generator being recalibrated away from defenders. The other difference that doesn't seem to be this year's whistle emphasis is increased playing time for the relatively foul-prone Spike Albrecht, who also gets whistled for a lot of ARE YOU SERIOUSLY HIGH RIGHT NOW SERIOUSLY blocking calls.
Free Throw Defense
Michigan was pretty good at it last year (68.5, 118th) and is miserable at it this year (72.9, 321st). Just one of those things. Every time I mention this someone asks about whether the distribution of shots between posts and guards is impacting this, and every time I say "maybe, but if so that is probably just luck as well."
This post was going to be longer. But:
- Michigan is a better defensive rebounding team this year, both in conference and overall.
- Michigan's TO force rate has dropped, but again so has the rest of D-I's. They were 240th last year. This year they are 243rd.
- Michigan's eFG allowed on half-court possessions has gone from 46.3% to… 45.9%. IE, it has improved in a tougher environment to play D.
They're not fouling more, they're not allowing more shots per possession, they're not allowing teams to shoot better in their half court sets. 100% of the defensive regression from last year to this year is on crappier transition D and charges being broken.
Is This Good Or Bad?
Well, it indicates what kind of team you'd like to see Michigan deal with in the tournament: slow ones. Failing that, it seems good that there's such an obvious problem that Michigan can try to mitigate by dumping a ton of practice time into.
On the other hand, we just saw Indiana chew Michigan up in transition, and they're not an efficient team in that department. They are a frequent team, with 28% of their shots coming quickly. But a big chunk of that is Indiana taking debatable shots quickly because they know their half court offense is going to suck. That's an obvious reaction, one Michigan should have seen coming. And yet there were multiple Indiana transition baskets of of Michigan makes. Almost 40% of Indiana's attempts were in transition*. This is not a waning issue.
Michigan has been able to slow down transition-oriented teams this season. Iowa and Michigan State are 6th and 13th at putting up early shots, respectively, and Michigan is 3-1 against those teams with three respectable defensive showings. (The two MSU games look bad because Izzo spent two solid minutes at the end of each game in a foul/matador cycle, but prior to that both games featured MSU at right around one PPP.) In the fourth, Iowa ran out to a big lead with a bunch of threes from Roy Devyn Marble, some of them in painfully wide open transition. 30% of Iowa's shots were fast, they went in at a 75% eFG clip, and Michigan got blown off the court.
I'd rather have one issue that Michigan can mitigate by sending waves of guys back than a big dip in half-court D, so I tentatively suggest this is a hopeful sign.
*[And of course Indiana was crazy efficient in half-court situations in that game. The overall trend is decent—or at least the same—half-court defense, though. Consider it stipulated that if Michigan plays half court D as badly as they did against Indiana, they're dead meat.]