Hoop Math: Five-Man Unit Stats, Or The Rise Of Mitch McGary Comment Count

Ace January 29th, 2013 at 12:24 PM


Michigan dodged a bullet today when X-rays revealed no broken bones in Jordan Morgan's ankle, but the Wolverines likely will have to make do without their starting center for the next couple games, at least. How much will his absence hurt Michigan?

If the numbers from conference play are any indication, not nearly as much as you'd think.

I spent yesterday compiling the statistics for each five-man unit John Beilein has deployed in Big Ten play (garbage time excluded) to see if I could spot any trends. The entire spreadsheet of all 40(!) different lineup combinations is available for your perusal as a Google Doc. Here are the five most common lineups the Wolverines have used, divided up by offensive statistics...

Burke-Hardaway-Stauskas-GRIII-McGary 78 45 32 15 27 22 16 22
Burke-Hardaway-Stauskas-GRIII-Morgan 58 33 37 14 35 21 18 18
Burke-Hardaway-Stauskas-GRIII-Horford 13 11 8 4 9 5 3 1
Burke-LeVert-Hardaway-GRIII-McGary 13 6 8 4 2 1 9 7
Burke-LeVert-Stauskas-GRIII-McGary 11 6 8 6 4 3 3 3

...and defensive:

Burke-Hardaway-Stauskas-GRIII-McGary 70 32 34 11 17 11 23 18
Burke-Hardaway-Stauskas-GRIII-Morgan 73 30 40 15 18 13 17 24
Burke-Hardaway-Stauskas-GRIII-Horford 14 7 11 2 6 5 5 5
Burke-LeVert-Hardaway-GRIII-McGary 12 7 11 4 4 2 5 3
Burke-LeVert-Stauskas-GRIII-McGary 12 3 6 1 2 2 5 3

The raw numbers are tough to compare, so this is where tempo-based stats come in handy. I've calculated each unit's number of possessions using KenPom's standard formula (2PA+3PA+(0.475*FTA)+TO-OR). From there, it's easy to calculate points per possession, which I've multiplied by 100 to give the standard offensive and defensive efficiency numbers. Also included is plus/minus, for those curious.

  Off Poss Off Eff Def Poss Def Eff +/-
Burke-Hardaway-Stauskas-GRIII-McGary 116.8 134.4 117.1 92.2 49
Burke-Hardaway-Stauskas-GRIII-Morgan 111.6 115.6 114.6 103.0 11
Burke-Hardaway-Stauskas-GRIII-Horford 27.3 143.0 27.9 89.8 14
Burke-LeVert-Hardaway-GRIII-McGary 24.0 104.3 26.9 104.1 -3
Burke-LeVert-Stauskas-GRIII-McGary 20.9 157.9 21.0 52.5 22

Well, then. Only the top two lineups have enough data to really rely upon (Michigan averages around 65 possessions per game, so even those have less than two full games of data)—the Horford lineup's numbers come almost entirely from the Illinois game.

Caveats aside, there's little doubt that Michigan's starters play are playing far, far better—on both ends of the court—with Mitch McGary at center than Jordan Morgan. The offensive efficiency with that lineup is off the charts*, and that defensive efficiency number would put Michigan just behind Ohio State, one of the best defensive teams in the country, at third in the Big Ten.

To see if this trend bore itself out regardless of the surrounding lineup, I calculated the offensive and defensive efficiency numbers for any lineup featuring Morgan/McGary/Horford...

  Off Poss Off Eff Def Poss Def Eff
ALL Morgan Lineups 180.6 109.1 179.2 102.1
ALL McGary Lineups 199.1 126.1 202.6 90.3
ALL Horford Lineups 43.3 124.7 43.8 89.0

...and the team's four factors statistics when each of the three centers is on the court:

  Offense Defense
Morgan 52.6 34.1 15.5 36.4 48.2 32.3 15.6 19.0
McGary 58.5 44.6 16.6 20.1 44.6 28.7 19.2 20.9
Horford 64.5 37.5 16.1 23.7 40.0 34.6 20.5 20.0

Here is where you can really see the difference between Morgan and McGary. When McGary is on the floor, Michigan rebounds over 10% more of the available misses on offense, and while they get to the line far less frequently, they shoot better from the floor. This could be chalked up as an anomaly, since the two-point shooting numbers are virtually equal for Morgan lineups and McGary lineups, while Michigan shoots 46.6% from three with McGary compared to 34.0% with Morgan.

There's a possible explanation for that, however, in the defensive numbers. The Wolverines force more turnovers with McGary on the court (19.2% to 15.6%), and of late many of Michigan's best looks from three have come off their transition game. That probably doesn't account for a 12-percent difference, but even if that's normalized there's still a gap in offensive production between the two; I consider McGary the better passer, a factor that may also contribute.

The difference between the two defensively is easier to figure out. McGary's activity defensively helps the team force more turnovers, while his excellence on the glass leads to a better rebounding rage. While McGary fouls more often than Morgan, the team fouls so rarely as a whole that the foul rate isn't affected greatly.

As for Horford, the sample size issues make it tough to take away anything concrete, but thus far the team hasn't missed a beat when he's on the floor—in fact, they're doing unsustainably well on both ends, with an eFG% of 64.5 on offense and 40.0 on defense. He's an interesting case defensively; like McGary, he's disruptive on defense, leading to more turnovers, but opponents are rebounding better with him on the floor than either Morgan or McGary.

I've said this before, but I'll make it clear in this post: John Beilein has stated repeatedly that he's very happy with the rotation as it is, and it could take some fantastic play from McGary paired with sub-par performances from Morgan for him to consider making any changes. When Morgan returns, I fully expect him to slide back into the starting lineup, and that's fine—given the physical demand of the position, regardless of who's starting McGary and Morgan are going to split minutes relatively down the middle anyway.

What this shows, however, is that Michigan has something special in Mitch McGary. Not only that, but Horford's solid work in limited time means the Wolverines shouldn't be in trouble if Morgan misses more than a couple games.

I'll have more notes from this five-man lineup data tomorrow, including insight on Caris LeVert's impact and how Michigan fares when they go to the bench.

*Michigan's overall conference-only offensive efficiency is 118.9, which is over ten points clear of Indiana for the Big Ten lead.



January 29th, 2013 at 12:38 PM ^

I, too, enjoy McGary's rebounding rage.

Also, small sample sizes and all, but those are some tasty numbers from the Burke-LeVert-Stauskas-GRIII-McGary lineup.


January 29th, 2013 at 12:39 PM ^

All year long I've been advocating including McGary in the rotation more, simply because Morgan drives me nuts with his constant mistakes and inability to finish.  This data finally makes a strong point that McGary should be in the starting lineup instead of Morgan.  Looks like we will get that wish this week and see if it makes a difference.


November 29th, 2013 at 10:58 PM ^

I'd agree that McGary should start getting the majority of the minutes, but I think it's unnecessary to start him and risk picking up an early foul. . Even if Morgan is hurt, I think Beilein  starts Horford (already did it in the 2nd half of the Ilinois game) In the end, McGary will play more minutes and  finish the game as long as his foul situation allow him to.


January 29th, 2013 at 12:48 PM ^

Get more minutes? Sure. But the rotation does a few things besides give someone the prestige of starting. There's a reason that Horford started the 2nd half with Morgan out... and I bet Horford gets the start tomorrow.

Mitch coming off the bench does a few things:

He brings new energy, faces either a tired starter or a bench guy, and Michigan won't have many fouls yet.

There are many NBA 6th men who don't do as well as starters. I think for his Freshman year, Coach B likes the advantages of having Mitch come off the bench. Bulls did something very similar with Omer Asik in his first year. When Noah got hurt, Kurt Thomas started at center so that Asik could still do his Asik rotation things that he was very good at. I'd expect Mitch's minutes to go up, but along the same lines of the current rotation.

Blue in Seattle

January 29th, 2013 at 1:29 PM ^

I think the excellent play of McGary is not just his talent and growth to this point, but also his coach putting him in situations where he can succeed.  Morgan has learned so much more of Beileins system, and basketball is so fluid, that I think it's likely we don't notice how the system is more limited with McGary.  That said, it's clear game by game that the freshman are improving acroos the board.  And the more teams shut down Burke, the more the rest of the  team responds.

I love that Brian/Ace are trying to keep himself from getting high expectations, but statistics are really better at predictions when things stay static.  This team is growing rapidly because it is so young and because it is so well coached.



January 29th, 2013 at 12:50 PM ^

This is really cool, great work.

I wonder if Morgan is more frequently going up against an opponent's top center while McGary might be playing against a backup more often.  Could be a part of the difference - i.e., Morgan is 'less better' than the top center and McGary is 'more better' than the backups?

gpsimms not to…

January 29th, 2013 at 1:00 PM ^

First, everyone should read kenpom's take on +/-  :


There's a ton of noise in these numbers.


Second, everyone should read:


McGary is a nice defensive player, but he can't move his feet like JMo can, and that causes him to get more than 5 fouls per 40.


January 29th, 2013 at 1:01 PM ^

pretty small and messy. But this is another interesting contribution to a McGary v. JMo debate that I kinda find regrettable. (Too bad you don't take into account any of the recent history of conversation about it here.) Generally, both players bring pretty different things to the table. And you don't get into what may prove the most crucial factor Saturday: being able to guard Zeller one-on-one. For that job I'll take Morgan in the starting role, hands down.

It still fascinates me that Morgan catches so much hell for early-career missed bunnies when he has been in the top two offensive leaders for the conference three years running. I think what it may mean is that it's hard to be a tall person and perform all these roles in the crowded area beneath the basket--that in fact JMo does it extremely well. We see McGary miss his share of such shots too. Love both of them myself, but think Morgan is really under-appreciated. Don't think he has quite reached the level of confidence on offense this year as last--or the year before, when Darius Morris found him with such pinpoint accuracy--but I hope to see that return. 

In truth, the big man blessings for Michigan are extraordinary; but Saturday we will see why we need Morgan. 


January 29th, 2013 at 1:02 PM ^

As a follow-up, perhaps you could look at the assist rates and transition baskets when McGary and Morgan are on the floor to better understand the offensive efficiency numbers? If McGary passes better and creates more offense off of turnovers, you should be able to find it in there, especially as the year goes on.


January 29th, 2013 at 1:04 PM ^

I wonder if you could combine this analysis with transition data. My subjective impression is that it often takes M a while to get into fast break mode,and this may coincide with McGary entering the game. Whether McGary is the direct impetus or not, his numbers would benefit from that. It does seem true that he's a good outlet passer, so he's probably responsible for at least some of the variance.


January 29th, 2013 at 1:05 PM ^

... McGary's excellence on the glass leads to a better rebounding rage. I always said Morgan needed to work on channeling his anger toward the offensive glass.


January 29th, 2013 at 1:07 PM ^

And I don't know if it showed up in the stats or not, but it seemed like Illinois punished us on the offensive glass when McGary was in. Did anyone else notice this? Are there any stats to back this up?


January 29th, 2013 at 1:42 PM ^

Illinois is a weird case to judge def rebounding by because their shooting is so perimeter oriented. Sometimes getting the proper inside rebounding position actually leaves you at a disadvantage for long and irregular rebounds generated from off target perimeter shots.


January 29th, 2013 at 1:05 PM ^


This is outstanding anlysis.  Thank you for putting in the time to produce & publish it.

But what about Bielfeldt?  No statistical  love for the big guy?


January 29th, 2013 at 1:30 PM ^

First of all, excellent work Ace.  I do have a couple of comments:

  • Individual defensive metrics are to be used carefully, but UMHoops did an analysis recently that showed Morgan was definitely better as an individual defender than McGary:  LINK
  • Obviously McGary seems to make more big plays with blocks and steals, but the eye test tells me he's a much more variable defender than Morgan.  He loses his man trying to help more, he fouls more trying to go for the block or being aggressive, and he lets his man get better position.  I think Morgan does most of his work without notice because he's denying his man the ball and never letting him get a good shot.  I also think Morgan sets better picks.  Those are probably the reason he'll remain the starter - along with the fact that having McGary's energy come off the bench is a big boost. 
  • If Morgan or McGary can ever fix their issue making shots from within 2-feet of the basket they will get a bulk of the minutes.
  • What's so amazing about this team is that we have three centers who are relatively interchangeable, plus we have Bielfeldt that can hold his own in small doses.  Think about last year - the second post player was Smotrycz playing out of position and the third was McLimans.  In order for McLimans to see the floor this year we'd have to get through FOUR other solid options.  And next year McLimans is replaced by a top-100 power forward who is a great fit for this offense! 

gpsimms not to…

January 29th, 2013 at 2:00 PM ^

I'm curious how synergy sports gets their numbers, because they obviously directly disagree with the numbers Ace comes up with.  

I feel like the "Jordan Morgan is a superior B10 defender" hypothesis is supported by the look test and the synergy numbers.  He moves incredibly quickly for a big.  He is dangerous on hedges, and quick to get back, etc.  But Ace's numbers (although they have tons of noise, see my post above) obviously are tempting us to make the opposite conclusion.

gpsimms not to…

January 29th, 2013 at 3:12 PM ^

I didn't really get how a man can have xx ppp, but I guess it must just be the ppp on the possessions they use.

In some ways, I can see the numbers make sense, then.  McGary's shot blocking makes him a better (potentially) team defender, whereas Morgan's speed/athleticism/3 years of DI S&C make him better equipped to man-up on a big.  But even that explanation doesn't really cover the huge discrepency in the two sets of numbers.  Like I've said, the easiest explanation is the +/- is noisy.

Against teams with strong bigs (IU, and MSU if Payne keeps playing out of his mind) I think it's really important to have Morgan out there.

We'll probably see Sunday, but I think Morgan's contributions to this team are severely underrated by most.  McGary's sexy right now, so to speak.

Blue boy johnson

January 29th, 2013 at 2:26 PM ^

I think there must be a few other things going on here. When I watch the games, and I watch them all, I just don't see the disparity between Morgan and McGary these numbers indicate.

I am not very familiar with these stats, but if they are indicating M is defending eFG% better by a  rate of 23%  just by substituting Morgan for McGary, something is amiss in the numbers.


January 29th, 2013 at 4:19 PM ^

Who McGary and Morgan play against are not random draws; they are endgenous decisions by the coach.  Adjusting for the quality of opposition would be one approach.  Overall, as one person said, this is reassuring.  But, it's far from definitive.

Matt S

January 29th, 2013 at 6:08 PM ^

It's almost impossible to over-state how sensitive this kind of +/- stuff is to sample size issues.

The article briefly comments on the fact that even the most used line-up listed accounts for less than two games' worth of plays, but it's even less reliable than that, because the line-ups compiled their stats against different opposing line-ups.  Stats analysts for the NBA typically prefer to include two seasons worth of data when comparing +/-, and they have twice as many games per season against more consistent opposition.

I like McGary, but this data just really isn't useful.


January 29th, 2013 at 6:43 PM ^

I've felt that McGary had outplayed Morgan the last two games and could push him for being the starter by next year if not the end of this year.  When Morgan went down against Illinois, besides feeling bad for JMo, I was not at all worried and a little bit excited to see what McGary would do with the extra minutes.  I felt McGary didnt play all that well and got benched a lot and ended up with his normal amount of minutes.  Horford impressed at times, like with his first post move that he sadly couldnt replicate, and Bielfeldt looked a lot better than expected. Overall though, I like things the way they are with McGary comign off the bench for 16 minutes and JMo starting and getting similar minutes with Horford taking up the rest.