Sunday, January 07, 2007

Big Game Hunting

So with the hiring of Nick Saban last week by Alabama and the way Alabama folks (and others) have been talking about "experienced coaches", I started thinking about whether particular coaches are "sure things" while others aren't. Is it reasonable for Alabama to (probably) overpay for Nick Saban if he's a "sure thing", when spending market rate for someone who isn't a "sure thing" could be more expensive in the long run?

But what is a sure thing? Is it someone with several years' experience as a head coach? I'd say no, because someone could have experience over time, but never elevate to greatness. A coach could do just enough to get by over time, and that wouldn't make him a "sure thing". Is a sure thing someone who has a great overall record for an extended period of time? Maybe, but in college football an overall record could be misleading - there are too many teams, and too many bad teams that a guy could potentially win nearly 70% of his games, but lose against almost every good team he played (a guy at Georgia in the late 1990s, perhaps?). Is that a sure thing?

So I started thinking about how particular coaches performed in "big games". Perhaps that's a better gauge of the talent of a coach. How his teams performed against top talent.

First, a few disclaimers before I get to the data.

  • For this study, I looked at a bunch of the coaches I thought were closest to "sure things" in college football today. I looked at guys with multiple years of experience, with big programs. Some guys who have recently become unemployed but still have experience. If I missed someone you think should be on the list, tell me in comments. Or feel free to run the data yourself...
  • Big Games - this is very difficult to define. Rivalry games are "big games". Conference games are "big games". I recognize that my definition is purely arbitrary and is cleary incomplete.
  • I used games against ranked opponents at the moment the game was played. Now, first off, I recognize that rankings are fluid during a season. A team that is ranked in the opening week might end up with a terrible record. Rankings during a particular week may not be a good indicator of how good that opponent is in the abstract. That said, rankings in a particular week is a good indicator as to whether the game is "big" - the perception of the opponent is good, the event is bigger. For example, in 1998, Georgia played at LSU early in the season. LSU at the time was undefeated and ranked in the top 5 or 6. Georgia won the game, and LSU fell apart, pretty much not winning another game and ending up with a losing record. That LSU team was not great, but the game against Georgia (I was there) was a huge game.
  • Another qualifier that has to be said is that not all ranked teams are the same. It's one thing if you're playing against the #1 team in the country. It's another entirely if you're playing the #25 ranked team. My numbers below are highly simplified, but I just didn't have the time to do more extended mining.
  • Finally, the numbers don't take into account how good the coach's team was or whether the wins and losses were against teams that were particularly better. For example, Jim Tressel and Pete Carroll have some of the best "big game" records. That may be because they're great in-game coaches. It also might be that they were coaching great teams against teams that, while still ranked, were ranked lower. So when looking at these numbers, take into account the teams these coaches coached. A losing record for someone like Ralph Friedgen at Maryland might not be as bad as the record suggests.
  • Some of the coaches have very small samples of games against ranked opponents. Take those numbers with a grain of salt.
  • Basically, these numbers don't really mean all that much. They're arbitrary, and small samples in some cases.
Here's how the current numbers look, ranked by winning percentage. The numbers are based upon the last 6 years, but some coaches haven't coached that long. The records are games against ranked opponents. The numbers are as of today (Tressel and Meyer naturally will add a win and a loss tomorrow). For ties, I ranked the guy with more games first.

1. Pete Carroll (USC) 20-6; .769
2. Jim Tressel (OSU) 20-7; .741
3. Mack Brown (Texas) 18-7; .720
4. Larry Coker (Miami) 21-10; .677
5. Urban Meyer (Florida, Utah) 6-3; .667 (2003-2006)
6. Bob Stoops (Oklahoma) 15-9; .625
7. Tommy Tuberville (Auburn) 14-10; .583
8. Gary Patterson (TCU) 4-3; .571
9. Steve Spurrier (Florida, USC) 9-7; .563 (2001, 2005-2006)
10. Nick Saban (LSU) 13-11; .542 (2000-2004)
11. Mark Richt (Georgia) 15-13; .536
12. Lloyd Carr (Michigan) 14-13; .519
13. Frank Beamer (Va. Tech) 11-11; .500
14. Bobby Bowden (FSU) 12-14; .462
15T. Kirk Ferentz (Iowa) 10-12; .455
15T. Barry Alvarez (Wisconsin) 10-12; .455 (2001-2005)
17T. Bobby Petrino (Louisville) 4-5; .444 (2003-2006)
17T. Charlie Weis (Notre Dame) 4-5; .444 (2005-2006)
19. Phillip Fulmer (Tennessee) 11-15; .423
20. Les Miles (LSU, Okla. St.) 10-15; .400
21. Tommy Bowden (Clemson) 8-12; .400
22. Mike Bellotti (Oregon) 9-14; .391
23. John L. Smith (Louisville, Michigan State) 7-11; .389
24T. Tom O'Brien (BC) 7-12; .368
24T. Chuck Amato (NC State) 7-12; .368
26. Pat Hill (Fresno St.) 5-9; .357
27. Tyrone Willingham (Stanford, Notre Dame, Washington) 9-18; .333
28. Ralph Friedgen (Maryland) 8-16; .333
29. Rich Rodriguez (WVU) 6-12; .333
30. Mike Price (Washington St., UTEP) 3-6; .333 (2001-2002, 2004-2006)
31. Jeff Tedford (Cal) 5-11; .313 (2002-2006)
32. Dennis Franchione (Alabama, Texas A&M) 8-18; .308
33. Houston Nutt (Arkansas) 8-20; .286
34. Mike Riley (Oregon State) 4-10; .286 (2003-2006)
35. Al Groh (Virginia) 7-18; .280
36. Gary Barnett (Colorado) 6-18; .250 (2001-2005)
37. Frank Solich (Ohio, Nebraska) 3-9; .250 (2001-2003, 2005-2006)
38. Bill Callahan (Nebraska) 2-6; .250 (2004-2006)
39T. Mike Leach (Texas Tech) 6-19; .240
39T. Joe Paterno (Penn State) 6-19; .240
41T. June Jones (Hawaii) 2-8; .200
41T. Jim Leavitt (South Florida) 2-8; .200 (can't find 2001)
43T. Jeff Bower (Southern Miss) 1-4; .200
43T. Steve Kragthorpe (Tulsa) 1-4; .200 (2003-2006)
45. Ron Zook (Florida, Illinois) 5-21; .192 (2002-2006)
46. Dan Hawkins (Colorado, Boise State) 2-10; .167
47. George O'Leary (UCF, Georgia Tech) 1-6; .143 (2001, 2004-2006)
48. Sonny Lubick (Colorado State) 1-8; .111
49. Glen Mason (Minnesota) 2-17; .105
50. Jim Grobe (Wake Forest) 2-18; .100
51. Greg Schiano (Rutgers) 1-13; .071
52. Gary Pinkel (Missouri) 1-14; .067
53. Dirk Koetter (Arizona St.) 1-18; .053
54. Paul Johnson (Navy) 0-7; .000 (2002-2006)


Like I said before, this list shouldn't be read into too deeply. The sample sizes are small, and much of these records are tautological in that good teams provide good records to coaches, and good coaches lead to good teams. It's hard to tell what's the chicken and what's the egg. But there are some interesting things to get from this list.
  • The guy with the most wins against ranked opponents since 2001 doesn't have a job right now: Larry Coker. I have to say I'm a bit surprised that his name didn't come up for any of the job openings this year. Maybe he didn't want another job. In any event, Coker should be in line for another job somewhere next year.
  • Chuck Amato and the guy who replaced him, Tom O'Brien, have identical records against ranked opponents since 2001.
  • Another thing about Chuck Amato (and something that shows how these numbers don't tell us all that much) - this past year he had a pretty good year against ranked opponents, going 2-1. He went 1-8 against the rest of his schedule.
  • The Notre Dame question? In his first two years, Charlie Weis has gone 4-5 against ranked opponents while getting crowned a coaching genius. Tyrone Willingham was 5-6 against ranked opponents in his first two years at Notre Dame (a better winning percentage). Take note, too, that under Weis, none of the teams Notre Dame beat while ranked finished the season ranked (none last year, Penn State was the only win this year and they are not currently ranked, though they may end up in the polls).
  • Looking at these numbers, I find it odd that no larger school has backed up a money truck to try to get Gary Patterson away from TCU.
  • There is sort of a pattern for coaches that have been fired recently - the inability to beat ranked opponents seems to be a large factor. Look at Dirk Koetter. 1 win over a ranked opponent in 6 years isn't going to cut it. Gary Barnett hadn't beaten a ranked opponent in his last three years at Colorado (and surprisingly, Dan Hawkins' record isn't much better). Glen Mason rode Laurence Maroney to two wins over ranked opponents, but that's it since 2000. During Ron Zook's tenure at Florida he lost at least 4 games against ranked opponents in every year, and never beat more than 2 in a year (Urban Meyer's bested that in both of his seasons). If these numbers mean much, I think Gary Pinkel should watch out (1 win in 6 years).
  • Nebraska fired Frank Solich and replaced him with Bill Callahan. They have the same winning percentage against ranked opponents (though Solich had a better record against them while at Nebraska).
  • It might seem a little odd that John L. Smith is out of a job when guys listed way below him were in line for promotions or better jobs this season. Slappy is kind of a mystery. His numbers against ranked opponents aren't too bad. But his numbers against unranked opponents aren't much better. Against unranked opponents over the same timeframe, Smith is only 33-23. 23 losses against unranked foes (about 4 a year) won't let you keep your job at a school that thinks they can compete with the best.
  • Some numbers are pretty skewed by coaching moves. Les Miles' numbers are much better at LSU than at OK State. Spurrier was naturally better at Florida than at South Carolina. Mike Price was better at Washington State than UTEP. Zook's numbers took a huge nosedive after heading to Illinois (and they weren't all that great to begin with). Willingham's numbers have tumbled because of the last two years at UW (0-9). Interestingly, Dennis Franchione has been better at A&M than he was at Alabama (I was surprised by that because it's kind of conventional wisdom that he's sucked in College Station). Half of Dan Hawkins' losses came this year at Colorado.
OK, those are trivia-level facts. What about the initial premise as to "sure thing" coaches? As I look at the list, I think there's probably a small list of coaches that fit the category of a "sure thing". Surely guys who win 70% of the games against top opponents have to be considered sure things, though an argument can be made that for Carroll, Tressel and Brown, who knows how they'd fare if they were coaching significantly less talented teams (and the response is that these guys recruit the players, so who's to say those teams would be significantly less talented...). In fact, I'd go a step further. Considering the small sample sizes and the inability to differentiate between games against higher-ranked opponents and games against lower-ranked opponents, I think a coach that's reasonably close to .500 against ranked opponents is good enough to be considered a sure thing. I think most college football fans want to go into every game against a top team with at least a 50-50 chance of winning - and the top 20 or so are the coaches that give their teams that chance.

So if the question posed is whether it was worth it for Alabama to spend $32M to sign Nick Saban, and he's one of those guys who gives you a 50-50 chance or better to win every game against top opponents, I think the answer is yes. There are very few of those kind of coaches out there. If you have one or have the chance to get one, it's probably cost effective to overpay, rather than spend nearly as much (market rate) for uncertainty or worse.

Consider: some of the top coaching candidates at various schools this year are towards the bottom of that list. Guys like Mike Leach, Greg Schiano, Steve Kragthorpe, Jim Leavitt, Jim Grobe and Paul Johnson. Their positions on this list are lower mainly because of the difficulty in winning at those locations, but also consider that the fact that they are "hot" candidates largely because of their winning records - which must be inflated by wins against unranked opponents. Each of these guys might be able to win consistently against top opponents at a school where it is easier to bring in top talent, but they're clearly not sure things - there's some level of uncertainty in hiring any of them. I think spending $4M a year for a 99% chance of getting a great coach (Nick Saban) is probably a smarter decision for Alabama than spending $2.5M+ for a guy who might only be a 60-75% chance of being a great coach (with a decent sized risk that in a few years you'll have to spend money to fire and rehire another coach - and yes, there's that risk with Saban too).