Coach-em-up Rankings, Pt I

We sometimes debate here the relative merits of elite recruiting
vs. the abilitiy to find under-the-rader talent, or system guys,
and coach them up.  So I decided to take a look at the
numbers for the teams in the ACC coastal to see if I could draw
any conclusions. In part I (of a possibly one-part series), I take
a look at VT, LOLUVA, and UNC...

Methodology
For each coach, I used the 247.com
composite recruiting rankings for each year. Using a formula I
devised (so it's not in the least bit scientific), I calculated
the expected performance of the team, based on recruiting class
rankings. Finally, I used either the final AP
poll
(or sagarin
ratings
if the team was unranked by AP) to determine the
actual performance for that year.

My formula for expected results is the sum of the expected level
of contribution for each class multiplied by the class ranking:
expected = (0.1 * 1st-year) + (0.15 * 2nd-year) + (0.3 * 3rd
year) + (0.3 * 4th-year) + (0.15 * 5th-year)

where 1st-year are true freshmen, 2nd-year are redshirt freshmen
and true sophomores, etc. Argue with my formula all you want, it's
just a guess on my part. Hope that makes sense. If not, leave a
comment, and I'll try to better explain.

Virginia Tech

Coaches:

Coach Recruiting Expected Actual Diff Years Wins Loss Win Pct.
Beamer (Good) 29.50 28.50 12.75 15.75 2004-2011 60 15 0.800
Beamer (Bad) 24.50 27.75 44.75 -17.00 2012-2015 28 22 0.560
Beamer (All) 30.93 28.25 23.42 4.83 2002-2015 32 25 0.561
Fuente 25.00 26.00 16.00 10.00 2016-pres 9 4 0.692


Recruiting Rankings, Expected Results, and Actual Results:

Year 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Recruiting 32 20 32 26 20 24 34 36 21 21 27 29 40 25
Expected 36 30 31 27 26 26 26 26 28 30 28 25 26 28
Actual 10 7 19 9 15 10 16 16 53 42 39 45 16

VT Coach-em-up Chart

Grades:

Frank Beamer:
This is really a tale of two Beamers. The good Beamer, who
coached from 2004-2011, and Franken-Beamer, who coached from
2012-2015. The good Beamer beat expectation every.single.year.
And by a hefty margin to boot. Then things fell off a cliff. In
his last four years, Franken-Beamer teams failed to live up to
expectations by an average of 17 positions. I'd really like to
go back and find the data for pre-2002, because I believe it
would reinforce my belief that the real Beamer is the good
Beamer. Overall, he did a great job getting his teams to
overachieve, but you can't just ignore the last four years.
Recruiting: B, Results: B+,
Coaching-em-up: B
Justin Fuente:
Coach Fuente really got things back on track in his first year
at the helm, beating expectations by a solid 10 positions.
Unfortunately, there's only one year of data to work with, which
isn't enough to draw any definitive conclusions.
Recruiting: B+, Results: A,
Coaching-em-up: Incomplete

LOLUVA

Coaches:

Coach Recruiting Expected Actual Diff Years Wins Loss Win Pct.
Groh 34.38 30.67 37.17 -6.50 2001-2009 59 53 0.527
London 37.17 39.83 68.00 -28.17 2010-2015 26 45 0.366
Mendenhall 59.50 36.00 92.00 -56.00 2016-pres. 2 10 0.167

Recruiting Rankings, Expected Results, and Actual Results:

Year 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Recruiting 35 16 77 29 60 36 64 23 25 29 32 50 63 56
Expected 16 20 26 31 44 47 50 45 45 38 32 29 36 44
Actual 23 33 27 26 45 69 79 34 83 92 49 71 92

UVA Coach-em-up Chart

Grades

Al Groh:
Groh started off with a couple of top-15 recruiting classes,
but could never field a team that lived up to that hype. Still,
his teams hovered around top-30 status, and for three years met
or exceeded expectations.
Recruiting: B, Results: B,
Coaching-em-up: C
Mike London:
A string of top-30 recruiting classes never showed it on the
field. London's 'Hoos exceeded expectations only once in his
tenure, and failed miserably most years. His -28 differential is
pretty terrible, as were his teams.
Recruiting: B, Results: D,
Coaching-em-up: F
Bronco Mendenhall:
His first year was a spectacular failure on the field, missing
expectations by 56 spots. Recruiting wasn't great
either. It's only one year, but the early results look good for
the Orange and Maroon.
Recruiting: C-, Results: F,
Coaching-em-up: Incomplete

UNC

Coaches:

Coach Recruiting Expected Actual Diff Years Wins Loss Win Pct.
Bunting 30.8 29.00 55.00 -26.00 2001-2006 27 45 0.375
Davis 18.5 24 42.25 -18.25 2007-2010 28 23 0.549
With 19 19 53 -34.00 2011- 7 6 0.538
Fedora 29.8 29.5 41.5 -12.00 2012-pres 32 19 0.627

Recruiting Rankings, Expected Results, and Actual Results:

Year 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Recruiting 19 46 23 11 28 11 24 19 43 28 31 28 35 27
Expected 33 27 27 27 28 22 19 19 22 24 30 32 32 30
Actual 35 43 87 66 29 33 41 53 42 44 70 15 37

UNC Coach-em-up Chart

Grades

John Bunting:
Good, but not great recruiting. Bad teams. Only came close to
matching expectations once.
Recruiting: B, Results: C-,
Coaching-em-up: F
Butch Davis:
An ace recruiter, pulling in top-15 classes twice. But the
talent never showed up on the field, with average rankings
around 40th. Came close to meeting expectations only once.
Recruiting: A-, Results: C+,
Coaching-em-up: D
With:
One year isn't much to go on,  but going 7-6 with top-20
talent aint good. UNC agreed.
Recruiting: A-, Results:C,
Coaching-em-up: F
Larry Fedora:
The numbers are somewhat skewed by the terrible -40
performance in 2014. Since then, though, his teams have beat
(2015) or come close to (2016) expectations. It's hard to tell
if he's trending upward, or if 2015 was a fluke. Still, the
numbers don't lie, and a -12 overall differential is not great.
Recruiting: B, Results:
C+, Coaching-em-up: C-

Conclusion
So there you have it. At the top of our coach-em-up list are
Justin Fuente and Frank Beamer, which should come as no
surprise.  In the roll of dishonor, the worst coaches were
(excluding one-and-dones) Mike London, John Bunting, and Butch
Davis. If there's interest, part II will look at Duke, Georgia
Tech, and Pittsburgh.

DISCLAIMER: Forum topics may not have been written or edited by The Key Play staff.

Comments

Great write up.

Isn't Fuente 10-4? Are bowl games being removed from results?

You're correct about Fuente's record: I will correct that. Bowl games are not intentionally removed from the results. I used google to find coaches records, with searches like 'list of uva head football coaches'. Mostly this took me to Wikipedia pages, and I scraped the numbers without any particular care. So, the numbers may or may not be precise, depending on the source. They should be accurate, though, for the most part.

I for one would love to see the results for Miami and Clemson. The last few years Clemson has seen the fruits of the recruiting labor, but they seemed to underperform for about a decade before then.

Plan for the worst and hope for the best, not the other way around.

I was saving Miami/Clemson/FSU for part three. Three elite-recruiting teams with vastly different results.

For the coaching them up "estimated", are you averaging the past 5 years recruiting classes evenly? The Class contribution % tends to run something like 10/20/25/25/20 for a college team (Fr to r-Sr), so that might be a more realistic cut for determining the predicted success per class. (Beamer ran lean on true Freshman and used the redshirt more than most "successful" coaches do.)

I really like Sagarin, too. Keep in mind, Sagarin is like a voter- he prefers weighing recent results higher, so he's great at predicting how good the team is "currently"; less accurate with how the entire season shaped up.

FEI and S&P look at the entire season at no particular time, so might be a better metric to weigh a season score with.

You definitely have my interest to proceed.

This is brilliant.

I estimated using 10/15/30/30/15. I'll rerun the numbers with your weighting, since (as I noted in the writeup) I was making up my ratio out of thin air. Thanks for that feedback. Thanks also for the FEI/S&P tip. I hadn't thought to use those. Somewhere out in the Internet I ran across a web page that scored the different computer rankings based on predictive results, but for the life of me I cannot find it again.

Note also that since my data only begins in 2002, the 2004/2005 'expected' suffer from some ramp-up error. Still, as a rough-cut, I think it's OK.

FEI only goes back to 2007 and S&P to 2005, so those won't get you any further than you've already gotten. I went back to 1987 and stumbled upon the Simple Ratings System (SRS), which uses W-L record vs SOS and the results tend to be fairly consistent with S&P & FEI.

I truly wouldn't change anything you've done. Your numbers are solid. I just ran the numbers for VT and the ratios you are using are just as accurate than the ones I put up:

5.21/18.23/27.24/27.87/19.99/1.47 for Beamer & Fuente going back to '87, (where true Freshman were less likely to play). Looking back 10 years, they look like 7.39/20.98/28.2/25.3/17.12/1.00

Then you have their polar opposite; the SEC model - coaches like Urban Meyer often don't have fifth year production (his Ohio State 2012 class the most recent, as that whole class was gone last season). Taking the averages from Some of the more successful recent classes (Alabama 08-09, UF 06-07, OSU 11-13), those ratios come out to 10.1/23.5/33.2/25.8/7.3.

I know you've said that you are looking at the ACC and you are breaking things apart by the different coaching era's but I think an interesting addition to this would be to look at the previous stop for all the 'new' coaches in the league. That would give us some insight into what to expect those coaches to do in their new positions in the ACC. I'd be interested to understand how 2016 coaching class were performing with their previous talent - Fuente, Richt, Mendenhal, Babers.

I told him I’d crawl on my hands and knees to be the DL coach at Virginia Tech. Now, all of a sudden, I’m sitting in this chair and I told him I’d still crawl on my hands and knees to work here. I just want to be here.
JC Price

My formula for expected results is the sum of the expected level
of contribution for each class multiplied by the class ranking:
expected = (0.1 * 1st-year) + (0.15 * 2nd-year) + (0.3 * 3rd
year) + (0.3 * 4th-year) + (0.15 * 5th-year)

You put those words together, those are my favorite words, Popeyes and bahama
- Mike Burnup

Would you be willing to revise your ratings based on your new rubrik?

Never Forget #1 Overall Seed UVA 54, #64 UMBC 74

Leg for commenting on the thread so I could find it

I gave him the leg, as it seems you forgot to.

Sometimes we live no particular way but our own

Oops. So he got two.

This is really interesting analysis. Objectively speaking, I found UVa's distribution the most interesting to look at of the six teams (from both parts 1 and 2). It's crazy how much variance they seem to have both in actual performance and in their recruiting rankings year to year.

From a non-objective (Hokie) perspective, it's also interesting and humorous to see them vary so much year to year from mediocre to complete sucktitude with a clear trend towards even worse sucktitude.

How I Learned to Stop Worrying and Love the Jet Sweep