Tuesday, June 05, 2007

Ranking Coaches X: Screw it, here's the full thing

OK, Intro was here. Longevity Post here. The original Sporting News list here.

I decided to better myself and try to load the spreadsheet I prepared for this onto Google Docs so I could just link to it here and do the full analysis in a single post, so here goes.

The full, sortable spreadsheet is here.

Analysis below the fold!


Covered this before.

National Titles

On its own, this category in a way speaks to longevity, since the coaches that have won titles tend to also be around the longest (or perhaps they stick around longer because they've won a title). Also, national titles necessarily include a factor called luck. Tommy Tuberville does not have a national title, though he has an undefeated season. Urban Meyer and Nick Saban have national titles, but not undefeated seasons. One ball bounces a different way, one call goes a different way, and a number of coaches that have titles might not, or coaches that don't have titles might have them.

Only 12 coaches have won a national title. That's a small group. 7 of the 12 have at least 12 years experience. Only 4 coaches have more than one title, and the average tenure of those 4 coaches is 24 years. The vast majority of coaches at BCS programs do not have a national title. The reason: it's hard to win a title, and luck plays a large factor. Some other quirks:

  • Dennis Erickson, whose high ranking from Dienhart made me scoff in the introduction, is one of the 4 current coaches with more than one title. That's probably reason enough to put him in a top tier. So I didn't give Erickson enough credit. Damn you EDSBS and your awesome volcano/golf cart picture!

  • Beginner's Luck? Bob Stoops won his title in his second year. Lloyd Carr in his third. Will they get another one? Their chances are probably as good as anyone else.

  • On Dienhart's list, the lowest ranked coach with a national title is Phil Fulmer at 31. 19 coaches without a title are ranked ahead of him.
Conference Titles

I think this is actually a better gauge of coaching talent than national titles. How a coach compares against his closest rivals and most direct competitors (who know him best) is a strong method. Now, for a few of these coaches, this category is useless. Charlie Weis won't ever win a conference title at Notre Dame. Joe Paterno has just 2, but Penn State was independent for a long period. Another flaw, as said in the introduction, is that not every conference title is of equal value. An undefeated season in a tough conference is far stronger than a shared title among 3 5-3 teams in a weak conference.

  • Pete Carroll has won at least a share of the conference title in every year but his first at USC.

  • Bob Stoops, Urban Meyer, Rich Rodriguez, and Butch Davis have won a conference title 50% of the time.

  • Spurrier and Carr have won titles better than 1/3rd of the time.

  • Erickson and Richt have won titles every third year.

  • Bobby Bowden has won a conference title about every third year, but 16 of his 37 years were spent as an independent. So really, he's going at a rate better than 50%. Only Carroll can beat that.

  • I find it fascinating that Mack Brown hadn't won a conference title until his 18th season. The longest another current coach has gone without a title is Al Groh, 12 years.

  • Tyrone Willingham is the lowest rated coach on Dienhart's list that has won a conference title. If you consider longevity and conference titles, it sure looks like Dienhart is unfairly downgrading Willingham.
Winning Percentage

I think this is another good, but not perfect method of objectively comparing coaches. Different programs have different inherent advantages and disadvantages. Programs play schedules of varying difficulty. But if anything matters in football, it's wins and losses. One drawback with sorting this category is that short-term success or failure screws with it. Bret Bielema tops the list with one great season. Dan Hawkins is near the bottom with one awful season.

  • Dienhart must definitely think Bielema's first year was beginner's luck. Bielema tops the sorted list of winning percentage, but he's way down at 39 on Dienhart's list, the biggest drop between the two categories.

  • Charlie Weis doesn't fare any better. By winning %, Weis should be 8th, but Dienhart has him down at 42nd.

  • Phil Fulmer also gets downgraded by Dienhart when considering winning percentage. Fulmer is 7th among BCS coaches in winning %, but Dienhart has him at 31. Considering Fulmer's national title and multiple conference titles, it sure looks like Dienhart is underrating Fulmer.

  • Jim Grobe got the biggest lift by Dienhart. Grobe is 39th among coaches in winning %, but Dienhart has him at 4th. Raising Grobe is definitely reasonable considering the disadvantages of coaching at Wake Forest, but 4th is still pretty high for a guy who has barely won half his games.

  • Bobby Johnson at Vandy is another coach that got the benefit of the doubt from Dienhart based upon difficulty of winning at his program. Only 3 coaches with at least a game under their belt have a worse winning percentage than Johnson (and one of them is Dan Hawkins, who has just a single bad year), but Dienhart has him in the top 30. Just so we're clear, Fulmer, who has a national title, 2 SEC titles and 7th in winning percentage, is worse in Dienhart's view than Bobby Johnson, with 0 national titles, 0 SEC titles, and 56th out of 59 in winning percentage.

  • Tyrone Willingham's winning percentage is, like every other category, better than where Dienhart rates him.

  • Coaches rated worse by Dienhart than their winning percentage would have them rated (by at least 10 spots): Bret Bielema, Urban Meyer, Mark Richt, Phil Fulmer, Charlie Weis, Ralph Friedgen, Les Miles, Tommy Bowden, Karl Dorrell, Ron Prince, Bill Doba, Ty Willingham. That's 3 of the 4 African American coaches (who have coached a year - Randy Shannon gets an incomplete). Also, 4 SEC coaches in the top 22 for winning % are downgraded at least 10 spots.

  • Coaches rated better by Dienhart than their winning percentage would have them rated (by at least 10 spots, not counting guys without experience), with my guess on why he's raised them in parentheses: Dan Hawkins (small sample), Bobby Johnson (weak program), Mark Mangino (???), Greg Schiano (weak program), Jim Grobe (weak program), Mike Riley (???), Kirk Ferentz (??? 34th in winning %, 11th with Dienhart), Tom O'Brien (???), Frank Beamer (longevity?), Nick Saban (Title?), Rich Rodriguez (???)
Winning Percentage As Against School's Historic Winning Percentage

This system attempts to isolate the difficulties and advantages some programs have. Basically, I looked at a coach's winning percentage and subtracted his school's historic winning percentage. If a coach served at multiple schools, I accounted for that. For example, USC has a historic winning percentage of .702. That means that it's a very successful program, and there are natural advantages to coaching there. If Pete Carroll's winning percentage is .844, he's +.142 over the historic percentage. So the program is good, but Carroll's abilities in isolation shows that he's made the program even better. Likewise, with a program that historically hasn't been that good, a coach might not top the winning percentage category, but he'd still look pretty good in this one.

I think this is the most accurate objective method of comparing coaches. The only flaw is small sample size - Bielema at the top, Hawkins at the bottom for example.

  • Dennis Erickson is the big winner in this category. He's third overall, only behind two guys with small sample sizes (Bielema and Meyer).

  • Bielema, Friedgen, Richt, Miles and Tedford are in the top 10 of this category, but Dienhart doesn't include any of them in the top 20.

  • Biggest downgrade by Dienhart from where a coach would be rated by this objective category: Ron Prince (from 13th down to 57th), but that's a small sample size. Bielema was downgraded the second most (1st down to 39th), but again, a small sample size. For a guy who has coached more than a single year, the most downgraded is Les Miles (from 9th to 41st).

  • Biggest upgrade by Dienhart from where a coach would be rated by this objective category: Bill Callahan (from 50th up to 21st).

  • Coaches downgraded by Dienhart by more than 10 spots: Bielema, Meyer, Friedgen, Richt, Miles, Tedford, Prince, Fulmer, Joe Tiller, Mark Dantonio, Tommy Bowden, Bill Doba, Charlie Weis, Dorrell, Mike Gundy, Randy Edsall, Pat Fitzgerald.

  • Coaches upgraded by Dienhart by more than 10 spots (my guess at a reason, and here I don't consider a bad program as a good reason since it's already accounted for): Hawkins (small sample), Bobby Johnson (???), Bill Callahan (???), Mark Mangino (???), Greg Schiano (???), Jim Leavitt (Shouldn't be listed here - he's the only coach at the program, so his comparison to history is neutral), Houston Nutt (???), Lloyd Carr (title), Kirk Ferentz (???), Tom O'Brien (???), Tommy Tuberville (???, near-title?), Frank Beamer (longevity), Nick Saban (title), Mack Brown (title), Rich Rodriguez (???), Jim Tressel (title)

Some Overall Thoughts

Looking at the various objective criteria, I think Dienhart overrates and underrates a few coaches, based upon their accomplishments.

OVERRATED: Mark Mangino, Bill Callahan, Bobby Johnson, Rich Rodriguez, Kirk Ferentz, Tom O'Brien.

UNDERRATED: Phil Fulmer, Ralph Friedgen, Mark Richt, Charlie Weis, Jeff Tedford, Les Miles, Tommy Bowden, Bret Bielema, Tyrone Willingham, Karl Dorrell, Bill Doba.

A few more specific nits to pick considering all the categories discussed:

  • At the top, Rich Rodriguez is not defensible at #3. Bob Stoops bests him in every single objective category. Spurrier too. And Erickson. And Bobby Bowden. Even Lloyd Carr tops him in every category but one.

  • Kirk Ferentz at #11 and Tom O'Brien at #13 aren't defensible either. Coaches who best or equal them both in every category but are ranked below: Bellotti, Paterno, Butch Davis, Richt, Fulmer.

  • Houston Nutt at #20 isn't defensible. Guys behind him that best or equal him in every category: Tedford, Richt, Leach, Friedgen, Fulmer, Tiller, Bielema, Miles. Nutt's objective rankings put him right in line with Tommy Bowden, whom Dienhart ranks 47th (though, arguably he shouldn't be that low).
So there you have it. I can't remember which site discussing this had the comment "Dienhart's fine through #1, but then he falls apart." That's almost right.

The problem, as suggested in the introduction, is that Dienhart's subjective ratings mean something. When he overrates a certain coach, that coach develops an undeserved reputation, which could affect the way his program is perceived.

EDITED for Grammar and a few small analysis mistakes.


Anonymous said...

How do you calculate the win% spread for coaches like Spurrier? How do you weight the Duke spread and the Florida spread?

LD said...

Boil the comparison % down to a per-game number. This is the equation:

(((Winning % for Coach at school A) - (Winning % historically at school A) * number of games coached by Coach at School A) + ((Winning % for Coach at school B) - (Winning % historically at school B) * number of games coached by Coach at School B))/ Total number of games coached by coach.

For example, Les Miles:

At Oklahoma State, he went 28-21 over 49 games. That's a .571 winning percentage. The historic winning percentage at OSU is .490, so Miles' comparison figure there is +.081.

At LSU, he's gone 22-4, an .846 winning percentage. The historic winning percentage at LSU is .638, so Miles' comparison figure there is +.208.

Since he coached almost twice as many games at OSU, we can't just average the two comparison figures.

So this is the calculation:
OSU:(+.081 * 49 = 3.969) +
LSU:(+.208 * 26 = 5.408) =
Divide by total number of games coached (75)

9.377/75 = .125

With Spurrier, Erickson, or the other coaches with more than one job, the equation just gets bigger.