Elite Coaches in their Third Year

Submitted by Eye of the Tiger on October 25th, 2017 at 11:20 PM

As a part of the current "sky is falling/future so bright" conversations, I decided to look at how Harbaugh's performance (and projected performance) stacks up to that of other highly regarded coaches. I chose 10 coaches across four conferences (Big 10, SEC, ACC and Pac-12), as well as Brian Kelley at Notre Dame. Most I chose because they are widely considered elite; a couple others I chose because the programs have faced challenges in the past that remind me of Michigan's. 

Among those 11 coaches, 5/11 (Meyer at UF, Dantonio, Nick Saban at LSU, Dabo Swinney at Clemson, Brian Kelley at ND) replaced their predecessors due to poor performance. Among the remainder, 2/11 (Meyer at OSU*, Saban at Alabama**) replaced their predecessors because of minimal-impact scadals, 2/11 because their predecessors left for another job (James Franklin at PSU and Chris Petersen at Washington) and 2/11 because their predecessors retired (Chip Kelly at Oregon, Jimbo Fisher at FSU).

*Counting Fickell as an interim coach, so Meyer was replacing Tressel not Fickell.

**In both cases, I do not recognize the NCAA's decision to vacate wins. Wins are wins.  

Given the small sample size and lack of randomness, you can't generalize from anything I write here. But it still points to some interesting patterns that potentially have some interesting implications for our program. 

Here is what I found.


1. Not all situations are analogous

The mean wins of all coaches in their first two years is 9.11 wins/season. The mean wins for their third year is 10.05 wins/season. However, if we look narrowly at coaches hired to replace poor performers (like Harbaugh), mean wins in the first two years is 8.88, while mean wins in the third year is 8.20.

By contrast, elite coaches hired for reasons other than the poor performance of their predecessors won an average of 9.15 wins/season in their first two years, and 11.59 in their third. That is an average of +3.39 wins/season in their third year versus coaches hired to replace poor performers, though only +0.27 wins/season more during their first two years.

Thus we see that coaches hired to replace poor performers win less in their first 3 seasons than coaches hired for other reasons.

This is almost certainly tied to how the previous regime ended. Poor performing predecessors won -2.04 games/year during their final two years as compared with coaches who left for other reasons.


2. Coaches hired to replace poor performers generally win less in their third year than in their first and second

Elite coaches hired to replace poor performers average +2.46 wins/season in their first two years relative to the last two years of the preceding regime. However, they also average -0.68 wins/season in their third year relative to their first two. By contrast, elite coaches hired for other reasons average +2.44 wins/season relative to their first two years.

As it happens, only 1/5 coaches hired to replace poor performers (Brian Kelly) won more in their third year than they averaged across their first two years. All the others declined. Win totals for Meyer at UF and Dantonio declined by15% relative to their first two seasons. Nick Saban at LSU: 10%; and Dabo Swinney: 16%.

To me this suggests recruiting problems. Specifically, it suggests that the poor performance by and transition from the previous regime impacts recruiting over a two-year cycle, as has been the experience at Michigan—a down year prior to their final season and then a transition year. By contrast, coaches who take jobs for other reasons appear to benefit from continuity. Notably, Meyer did appreciably better in his third year at OSU, where he replaced a successful coach because of a minimal-impact scandals, than he did at UF, where he replaced Ron Zook.


3. How Harbaugh Fits in

Harbaugh averaged 10.00 wins/season during his first two, which is +0.89 wins/season relative to the mean for all elite coaches and +1.12 wins/season relative to coaches hired to replace poor performers. This despite the fact that Hoke’s win total during his final two seasons ranks 10/11 on our list, beating only Jerry Dinardo at LSU. Notably, these two preceding coaches are also the only ones on our list to win fewer than 50% of games during their final two seasons.

We do not yet know what Michigan’s win total will be for 2017, but 8 wins seems like a reasonable bet at this point. If that projection stands, then it would put Harbaugh at -2.05 wins/season relative to the mean for the whole field, but only -0.20 wins/season relative to the mean for other coaches hired to replace poor performers. In terms of rank, Harbaugh would be tied for 3/6: under Kelly (12.00) and Meyer at UF (9.00) but equal to Saban at LSU (8.00) and ahead of Dantonio (6.00) and Swinney (6.00).


4. Fourth Seasons

The 9 coaches who have completed their fourth seasons recorded an average of 11.56 wins/season, which is +1.68 wins/season relative to the average for their third season. Those hired to replace poor performers recorded an average of 11.20 wins/season, which is a whopping +3.00 wins/season relative to their third season average.

And 4/5 coaches in this category improved in their fourth year relative to their third year—the sole exception being Brian Kelley, who was the only coach to improve in his third year relative to his first two.


5. Conclusions

Harbaugh’s performance in his first two years was above average for both elite coaches as a whole and those specifically hired to replace poor performers. A projected 8-5 finish would be significantly below the average for elite coaches as a whole but fairly average for those replacing poor performers, a group that includes both Urban Meyer at UF and Nick Saban at LSU.

It is impossible to know what the future brings, but if one considers Harbaugh to be an elite coach, then one can hypothesize—from these other examples and our 16+ returning starters—that in 2018 we will improve our win total by a significant margin.

This is why I’m not ready to press the panic button, and why--regardless of our current offensive ineptitude--I’m still bullish on our program. Even Dantonio and Swinney, who only won 6 games apiece in their third years, won 10 or more in their fourth. And in both cases, that success was sustainable. MSU, one notes, has won at least 10 games in 5/7 years since his third; Clemson has done so in all 6/6. In fact, of our 5 coaches hired to replace poor performers, only Kelley has failed to win 10 games in a majority of seasons after his third.

If Harbaugh is not, in fact, an elite coach, then there is more reason to worry. But given his track record and how we are recruiting, I think it's more reasonable to assume he is. In that case, truly sustained success begins when we can field an experienced team and also have credible depth at all or nearly all positions. Even better, when we start replacing outgoing seniors and juniors with incoming seniors and juniors, year in and year out (particularly on the offensive line, where we have struggled since 2011). In that case, we'd look a bit like Wisconsin, but with better talent. 

That's a good recipe. 




October 26th, 2017 at 8:56 AM ^

My only problem with this type of comparison is really a timing issue.  When Harbaugh was hired he had two weeks to put together a recruiting class that had imploded when Hoke was fired.  His second and third year's classes were outstanding.  Most of the referenced coaches had a couple months rather than weeks to put their first classes together.  In the case of Meyer at OSU, it was nearly three full months.


October 26th, 2017 at 12:39 PM ^

I cannot stand any chicken littles. This season's roster and Hoke's subsequent recruiting in '14, as well as the '15 class that was ultimately JH's (but really Hoke's fault), told me from the get-go that 8-4 was the ceiling for 2017. And that's ok!!! Year 3 isn't some magical year -- scheduling and roster situations matter so much more. Next year we return a lot of starters (albiet the o-line and the QB spot are scary) but what makes next year tough is the amount of tough road games. If we have QB continuity from '18-'19 and improved o-line play in 2018, I honestly think 2019 is looking much brighter for us than either this year or next, based on experience and the schedule.

If people would chill the fuck out that would be great. BREATHE people, BREATHE. Dave Brandon doesn't work here anymore, Brady Hoke and RR don't work here anymore. We have a coach who is HIGHLY competent and isn't out just for himself. We will be fine.






October 26th, 2017 at 5:07 PM ^

Better talent wise?  What about it screams better talent wise?  You will not have Mo Hurst, Chase W, Rashan Gary.  Maybe Bush Jr, maybe Hudson if he blows up next season.  On offense you will have Cole gone.  

I agree its good to have players continue to develop but there is some great talent that we are either losing after this season, or will be after next season.


October 26th, 2017 at 1:02 AM ^

Sorry but not counting vacated wins is dumb in my opinion because we know they actually won those games on the field. So while it probably doesn't throw the numbers off by a lot, it still does throw them off.



October 26th, 2017 at 1:32 AM ^

Defense is probably going to be better next year as crazy as that is. With only McCray and hurst leaving everyone is back and the youth has a full year in the program. Defense will be deep.

The scary thing is the oline. Who the hell are the tackles going to be? It could get worse. Really wish harbaugh had taken a dual threat QB in Peters class.

Eye of the Tiger

October 26th, 2017 at 2:19 AM ^

Ideally Newsome is back and either he, Bredeson or both can play tackle. In a perfect world, our OL is somethig like: Bredeson, Runyan Jr., Ruiz, Onwenu, Newsome. 

Beyond that, JBB is likely to get a 5th year, and though he's never been great at pass pro, he's been okay this year at run blocking and might improve enough to keep the RT job, or at least be a solid contributor at the position. And then there's Filiaga, who has ideal size for a tackle and will at least have his redshirt year done. He *might* be ready to start. 

All that said, like you, OL is the position group that makes me most nervous. We've missed on too many recruits over the past six years. Plus it's very hard to project OL performance based on limited snaps and practices we don't see, so it's all a bit mysteries of the unknown. And who knows whether or to what level Newsome can play after his injury. 

But at least we'll have some depth, in the form of multiple guys with solid talent and multiple years in the program plus a couple high ceiling RS freshmen. 




October 26th, 2017 at 11:46 AM ^

I would be willing to bet Newsome is at least as good as Cole is this year, faith I have in the UM health system. Im not sold on Bredeson being able to play tackle at this point but if he can that does help the line tremendously as that interior would seem to be much improved over this year. (Yes I think a RS fres ruiz is as good if not better than our current center and Runyan would be better than this years Bredeson.) But I think best case scenario would be one of the project tackles finally has the light switch on and makes a solid jump so Bredeson can stay at guard. Even if that doesn't happen I think we are a better line next year with the starting 5 you proposed. 


October 29th, 2017 at 12:44 PM ^

Isn't Winovich gone too?


And the OL is playing significantly better right now (amazing how much better an OL looks when they're executing the plays their talent dictates they'd do better and the QB throws the ball on time).


I think it's waaaaaay too early to be worrying about the 2018 OL, but to answer your question, I guess the 4 leading candidates for 2 OT spots next season would be Filiaga, Bushell-Beatty, Steuber, and assuming health, Newsome.


Isn't Hudson being projected as a T too?


October 26th, 2017 at 6:55 AM ^

I appreciate your work in putting this together.  It nicely counterbalances the (false) narrative that all elite coaches win a national title by year three (!), and the samples you identify support the intuitive conclusion others have suggested - the recruiting drop-off of the very end of Hoke's regime and the transition year are naturally going to be felt about three years later, when those classes are juniors or seniors.  I will cross my fingers that some of Harbaugh's OL recruits hit, and that we have a breakthrough at the QB position.


October 26th, 2017 at 8:34 AM ^

Nick Saban replaced Mike Shula at Alabama for "poor performance." Period. 

Bob Stoops also ought to be in that mix. He also replaced a "poor performer."


Also: "Meyer did appreciably better in his third year at OSU, where he replaced a successful coach because of a minimal-impact scandals, than he did at UF, where he replaced Ron Zook." 

I mean, that's true. But Meyer won an f'n NATIONAL TITLE in his SECOND year following the poor-performing Zook. It took him three years to do it OSU.



Eye of the Tiger

October 26th, 2017 at 9:28 AM ^

Mike Shula won 10 games in his second to last year, and it's not clear he was getting fired without the scandal. So it doesn't count according to my criteria.

Yes, true about Meyer at UF winning a national championship in his second year, but this analysis is about whether a third year dip relative to the first two is normal or not for elite coaches--and what this may or may not say about subsequent years. Meyer, like 3 of the other 4 coaches in his category, experienced a dip in his third year relative to his first two. Then he won another national championship.

Yes I could have used Stoops but decided not to since that was almost two decades ago. Like I said in the preamble, this isn't statistically rigorous. If I was getting paid for it I'd make a dataset with 200+ datapoints samples randomly. But I'm not, so I picked the first names that came to mind and went with "quantified illustrative anecdotal evidence" instead.


October 26th, 2017 at 10:28 AM ^

Mike Shula barely had a winning record...at Alabama!...in four seasons. He was, without a single shred of doubt, fired because of performance. He lost his final 4 games and was 0-4 against Auburn. 

Honestly, it seems more like an effort to justify Harbaugh as an elite coach. I mean, he IS an elite coach. That's not really up for debate. But it's not because he's "just like" or "doing better" than UFM when he was at Florida in his 3rd year. That strikes me as fanciful.

Stoops is one of the "elite" coaches of this era -- I mean he coached last season.  Jimbo Fisher is probably another one.


Eye of the Tiger

October 26th, 2017 at 6:06 PM ^

LIke I said in the piece, you have to assume Harbaugh is an elite coach in order to say his results fit. Brady Hoke won 19 games in his first two seasons. So his performance would also be ahead of Dantonio or Saban at LSU up to that point. But we know Hoke is not an elite coach. This data doesn't tell us anything about whether Harbaugh is elite.

What it does tell, on the other hand, is that down third years are normal for elite coaches who take over for poor performers. And it also suggests that if--and only if--Harbaugh is in fact an elite coach, then we can reasonably expect a major uptick in wins next year relative to this year. 

But of course not all coaches who have success are elite. We might have thought RR was elite in 2007, but it's clear at this point he's not. Maybe we thought/hoped for the same with Hoke after 2013, since he also overperformed in his first two years. But then it went the other way in 2014.

At the end of the piece, I state that I believe Harbaugh is an elite coach; based on his track record and recruiting, I see grounds for optimism. That said, I was very clear that this is not something you can prove or disprove from the data I presented. All the data suggests is that elite coaches often have down third years, relative to their first two. 


October 26th, 2017 at 5:12 PM ^

Just had three little losses get in the way of that from happening.

Last year stings.  With how close they were, I think last years team has an ESPN 30 for 30 vibe for what should of happened but didn't.  I don't think most of us here understood how stacked that team was last year.  Just a few bad breaks.


October 26th, 2017 at 8:38 AM ^

a real coaching challenge. Coaches only get 20 hours per week to teach these kids, so 2+ years is often needed for OL and QB to mature enough to start. Coming out of camp, this team looked sharp compared to now. Youth needs more than 20 hours per week to keep up with the expected learning curve, thus, it takes another bowl season, spring and fall camp to reach full potential. Next year will be better.


October 26th, 2017 at 9:48 AM ^

My only problem with the analysis?


You look at a bunch of coaches, and only two of them are elite.


Urban Meyer and Nick Saban.


Everyone else on your list is just "good", not "elite".



October 26th, 2017 at 1:38 PM ^

There's not much debate that right now there are two "elite" active coaches -- Saban and Meyer.

Banging on the door is Swinney.  Maybe Jimbo Fisher, but I'm not convinced of that.

And immediately behind that is a host of really good coaches who have yet to punch the ticket to "elite" -- Harbaugh, Kirby Smart, Chris Petersen, etc.


October 26th, 2017 at 9:21 PM ^

Part of the reason there are so few elite coaches in college football is because many of the best coaches go to the NFL. Guys like Chip Kelly, Bill O'Brien, and Pete Carroll were all elite coaches at the college level, and are now in the League. Harbaugh, of course, coached the 49ers for four seasons before returning to college. I've always found it difficult to swallow the contention that Saban is elite and JH is not, when Harbaugh's NFL coaching stint produced three NFC championship game appearances and a Super Bowl, while Saban flamed out in Miami.


October 27th, 2017 at 7:55 AM ^

Alright, I see that he's not coaching this season. I don't really watch NFL so I tend to lose track if guys once they leave college ball. But if he does return to college I think he should be presumed an elite coach until he proves otherwise.


October 26th, 2017 at 9:49 AM ^

Thank you.  This is one of the more interesting threads I've seen on this site.  It may not be statistically rigorous, but frankly those kind of posts make my eyes roll up into the back of my head (and I was a Math major).  This post also stood out because it told us something we hadn't considered.  Many times I feel like we see posts that tell us about what we'd expect, but to four digits to the right of the decimal point.


October 26th, 2017 at 11:00 AM ^

my own research on Wikipedia (ducks). I was interested to find out that Alabama has never lost as many starters on offense or defense as Michigan has this season for as long as Saban as been their coach. In Saban's 11 years as coach they have never had any less than 10 returning starters (and that happened only once in 2010). On average, they return 13 STARTERS a season (between offense and defense), which is insane to me. There wasn't a difference IMO in returning starters with the years they won the National Championship, but they won it in 2011 with 17 and in 2017 they have 16.....so there's that. Also consistently paired with the #1 recruiting class ever year. I digress, my point is that if Michigan can start to get meaningful contributions from Freshmen right away, we can start to build the depth and experience that Alabama has maintained for the past decade. I believe we are already there with our defense, once the offense catches up, we will set up to be successful for years to come. 


October 26th, 2017 at 12:13 PM ^

UM needs to find a way to balance out their recruiting classes. The guys from the larger 2012 and 2013 classes have graduated leaving the 2017 team void of upperclassmen. The '14 and '15 classes were light in numbers so there are no juniors and seniors ready to fill in the void.

Hoke faced a similar situation in 2013 when the graduating class left the team full of freshman and sophmores.

These ups and downs in recruiting has lead Michigan to "re-build" twice in recent times, 2013 and now 2017.

Harbaugh has taken in large classes in '16 and '17. 2018 is shaping up to be lower in numbers. They are set up again for another dip around 2020/2021. If the 2018 and 2019 guys can be instant impact type players they may be able to smoothen out these peaks and dips.

EDIT: I think I said the same thing you did in a round about kind of way.

Coach Carr Camp

October 26th, 2017 at 7:58 PM ^

We need to avoid the crater that is our 2014 and 2015 classes though in 2018 and 2019. 2014 was not horrible, and we are shaping up to be pretty similar in 2018. Thankfully, while people are whining about the 2018 class being mediocre, the 2019 class is actually off to a very solid start with 3 top 100 guys already. 


October 27th, 2017 at 10:06 AM ^

I think between this season and next they'll begin digging out of the hole the 2013 OL busted class left. They had 6 OL recruits in that class, 5 of which were four star, top 170 players. It was probably one of the better OL hauls in UM history, on paper anyway, with Kyle Bosch, Logan Tuley Tillman, David Dawson, Patrick Kugler and Chris Fox.

Only Kugler made it and he was in a real competition this season with a true Freshman. Not all of those guys would have been five year guys if a couple more had made it, we'd be in a vastly different OL situation.

2014's class consisted of only Mason Cole and JBB. That is understandable as you took a high number of OL in 2013 and you could never guess that nearly every guy in the previous class would bust. Thank goodness they hit on Mason Cole.

Regarding 2015, Newsome is hurt and who's future is unknown, we all saw what happened with Ulizio, and Runyan has bounced in and out.

2016 guys are only in their second year.

I think they have some quality depth starting to build but unfortunately it's a slow process.