Podcast: How Michigan flipped the script against Michigan State

1 Like

The stat Dylan shared about the “turned the season around” narrative is still bobbling my mind. I can’t fathom that Michigan was #30 in Kenpom when we were 7-7 and now that we’re 16-12 we’ve moved all the way to #29. I realize we haven’t exactly gone on a crazy run but there’s no way you can convince me we haven’t made significant improvements.

  • Second half Hunter > First half Hunter
  • Second half Jones >>> First half Jones
  • Second half Houstan > First half Houstan
  • Second half bench = First half bench
  • Second half offense > First half offense

I guess you could say Eli has been quiet in these last few weeks and Diabate has been just as up/down throughout the entire season.

Is it possible that when you talk about their Kenpom rank it doesn’t capture actual changes in efficiency because other teams have changed during that time as well? Meaning that the #30 team in December was less efficient than the #29 team in March?


Preseason projections we’re propping up those early numbers. If you removed those from the equation there would probably be a noticeable difference.

1 Like

On he who shall not be named’s site, Michigan was 40th up to the Illinois game and 18th since.


I was using Torvik splits, so no preseason projections. I did Nov/Dec and Jan/Feb as the splits and it was 28/29th ish for both.

Adding in the Rutgers/Illinois games to that split makes it more pronounced like @kturnup noted. Probably makes a bit more sense.

The team defense has been worse overall.

You mean as the season has gone on? I would have guessed that the defense has mostly been bad all year and more erratic than trending one way or another - especially when adjusted to opponent given how tough the B10 is.

This is maybe a better way to look at it… The numbers on the left are the split data (numbers on the right are the season-long).

Start of the season through UCF:

Remove Rutgers/Illinois COVID games…Here’s the same data for the games since:

IMO I do believe most of Eli’s negatives are the result of nagging minor injuries he’s probably been dealing with all season. A real gamer and team player as we all know​:thinking::wink::grinning:


I also think the best version of Eli is when he’s quietly doing the little things on the margins rather than being a focal point. Chase a guy around screens all night and don’t let him shoot, make a couple of off-ball threes, score a layup on a back-cut that no one sees coming, put people in the correct spot, etc.

So I have to point out that Eric does not help his brand by calling the decision to keep Jones on the bench in the first half inflexible. All of the specifics of this situation point in favor of keeping Devante parked on the bench:

  • Jones proved that he wasn’t ready to play through foul trouble by picking up his second foul before the first TV timeout.
  • The team went on a huge run without him.
  • The opponent is thin at point guard for this game.
  • The team is in the middle of playing four games in eight days, so any opportunity to play the bench without losing ground is pure gold.
  • Jones in particular looks pretty banged up - every time he goes to the floor in a dead ball situation, he looks like he needs paramedics to get him up again.

This is not a hill to plant the Death to the Autobench flag on.


You’re correct in this case for sure on all accounts. It was a horrific start for Jones and he needed to be sent to the bench. The unknowable question that they discussed on the pod is what would Michigan have done if MSU got back into the game while Frankie was on the court? Thankfully, we’ll never know.

Let’s hope that in any significant, close game the coaches don’t resort to the full autobench.

I’m also just more sympathetic to the autobench in games that figure to have more possessions. Especially early on in that game, it was a track meet, so the risk of Jones getting #3 was a lot higher than usual.

Yet another reason why the Beilein autobench didn’t make much sense. Some of his best teams would play, like, 55 or 60 possessions instead of 70.

To be honest, besides that Michigan went on a big run almost right as Frankie entered the game, none of these points have a ton of validity to me. I think I acknowledged as much on the podcast after Dylan made the point about Michigan going on a run. That did give them more flexibility, even though I’m not sure that would have changed how Michigan approached it regardless.

I don’t really change my stance on it though — I still think it was poor process despite the results (which includes the Caleb decision, too). I also think it is telling that Frankie played probably his best half of the season, then didn’t play the entire 2nd half until garbage time.

I disagree with this line of thinking. I mentioned this on the first podcast but there are many ways around this — for one, you can alter normal substitution patterns so he sits in the short term, but comes back later in the half (say, return at the 10 min mark, just as an example). Also, it is just reactive to assume he picks up a third just because the pace is faster. He played the entire second half and picked up one more foul.

This is in general and not specific to your comment, but by benching DeVante (or Caleb), you are making the decision that you won’t let him pick up a third foul in the 1st half rather than seeing if it actually happens… and if it does, he still has two more fouls to give. The goal isn’t to avoid a player picking up 3 fouls in the first half — it should be getting your best players on the court for as many minutes as possible. That is the most important thing. And to a much lesser extent, avoid them picking up a fifth.

I just think the patterns of when to bench players with fouls is so engrained in coaches and the public that we sometimes forget the bigger picture. That goes for player behavior too — they may think they’re in foul trouble when they pick up one foul at the beginning of the 1st half because they could get benched for picking up another. That just does not make sense, in the large majority of cases.

You are sitting players before they actually come close to the foul limit, which is 5 fouls. Let there be more information about whether they will come close to a 5th foul before you bench a player with perceived foul trouble.


Interesting debate.
I wonder if there is any correlation between tempo and foul rate.

I agree that a strict autobench regardless of context doesn’t make sense.
But, I’m not sure that means that you always want more info about being close to a 5th foul…

In the context of the MSU game, Frankie’s spell of good play (as well as the score) meant Jones could rest. Getting Frankie minutes is another goal for this season as well as next. (I thought Martelli’s comment that Frankie would be working on his free through shooting that night interesting – he’s teaching Frankie that a PG missing free throws is a huge liability going forward.)

Also, players adapt to foul trouble to some extent. If Frankie had picked up 3 fouls and Jones had then come back in (still in first half), he’d know that the TEAM’S PGS already had 5 fouls …

I agree with your general position about the autobench and your monologues about it on the recent podcasts.

But I think that game was an example where the results dictated the process rather than a “poor process despite the results”.

Martelli basically rolled a 6 on ten consecutive rolls of the figurative Collins’ die. There was never a need to resort to Jones and risk getting 60% of your allotted fouls less than 50% into the game.

Michigan was up 1 when Jones went out and stretched the lead to 10 within 4 minutes of Collins being on the floor. I think everyone can agree that playing Collins for those 4 minutes was reasonable.

For the remaining 13 minutes of the half MSU cut the lead to 7 at 10:32, then Johns hit a 3 at 10:10 to bring the lead back to 10. It was back to 8 at 9:05, then immediately back to 11 at 8:45. The lead ballooned to 16 for the remainder of the half. I just don’t see the value in forcing Jones back on the court when Collins in the lineup was clearly clicking extremely well. Michigan had a choke-hold on the game the entire first half and putting Jones on the floor again is at least partly risking the maximum number of minutes you can get out of him in the second half if the game is closer, and thus higher leverage (if he picks up a 3rd immediately).

Yeah I admit that was an assumption on my part, that more possessions = more of an opportunity for Jones to pick up fouls. I kind of figured they went hand in hand, but maybe not, maybe foul rate doesn’t work that way. But in general I definitely agree that autobenching someone in the 1st half is usually overly cautious and often detrimental, I was just trying to make sense of Martelli’s logic. Another point in favor of not autobenching could be U-M’s PnR coverage… If Michigan was committed to going under and sagging and dropping instead of blitzing or icing or being more aggressive, maybe that tactic removes some of the risk of our guards picking up defensive fouls? In which case 2 fouls is less of a risky spot than it otherwise might be.

1 Like