Wednesday, January 24, 2007

What's the Best Conference in College Football?

I've spent time on and off for about the past month working on this post, and it's finally here. I've done an analysis of all of the BCS conferences to find the best one out there. It's pretty long, and Blogger isn't conducive to letting me mess with the CSS like I want to, so the full text follows after the jump.

1.  Foreword

On Saturday, December 2 of last year, as people debated whether Florida or Michigan should be invited to face Ohio State in the BCS National Championship game, the perceived historical strength of the SEC and weakness of the Big Ten was taken as a given by some commentators and voters.  This was not a new statement, and I'd heard it so many times before that I almost believed it.  But as a lifelong fan of a Big Ten team, it irked me and I wanted to see for myself if it was true, or just another myth.  To answer my own question, I've prepared a comparison of the BCS conferences over the past 15 seasons (and shorter intervals) and here I'll share with you my results.

Continue reading...


Kyle King said...

This is some impressive and illuminating work. Nice job.

I've been addressing out-of-conference scheduling over at Dawg Sports, but my approach has been more qualitative than quantitative, so it's good to see some solid number-crunching in this area.

I have one question about methodology, though. While I agree that scheduling Baylor is more respectable than scheduling North Texas, is there a way to make allowance for scheduling good teams from non-B.C.S. conferences?

In 2005, Georgia played Boise State, a W.A.C. team that wouldn't register in your calculations, but the Broncos certainly qualify as "playing somebody" to a much greater extent than scheduling, say, Duke, which would count under your approach.

If some allowance could be made for this, even crudely (e.g., giving partial credit for non-B.C.S. conference teams that played in bowl games), what is already an insightful and nuanced look at out-of-conference scheduling could be refined further still, painting an even truer portrait of how the respective conferences schedule.

Geoff said...

Thanks, Kyle, I appreciate it.

[I]s there a way to make allowance for scheduling good teams from non-B.C.S. conferences?

I struggled with that question too when I was doing the analysis, and it's an important point. In the end, I thought that the best way to get an idea of how the BCS conferences stack up against each other, top to bottom, was by excluding non-BCS-conference teams except for ND (the key words here being "against each other"). It was a nice bright line, and it made sifting through the data a lot easier. As a practical matter, double-checking the conferences' records against each other became possible and that gave me a much better feeling about my math.

So that trumped the scheduling implications, at least for now. The good news is that I still have a shiny pile of data and there's a lot of time left before September comes around. I really like the idea about giving credit for putting bowl-bound teams on the schedule. Everything's already in the spreadsheets, so it's a matter of some creative compiling and sorting. I'll let you know if I come up with something interesting.

Anonymous said...

Excellent analysis! Enjoyed the read.
Now if we can get Michigan to play us (UGA) in Athens in September.
I don't think it will ever happen.


Anonymous said...

I can't entirely fault your conclusion here, but I will question some of your methods. The 90s belonged to the SEC, from 2000 on, it's not been the case. The PAC-10 has probably been the best conference of the 2000s, numberswise when you take into account schedule strength. Your initial impression that the last 90s was not kind to the Big Ten was correct and you should have stuck with that.

I'm not absolutely possitive here what your are doing, but I'll make the point that using the method of eliminated only major opponents can be useful, but also deceptive. The obvious advantage is to measure opponent that are at a similar level. However, if you are assuming that all teams that are currently major have always been major, what's the point of making the breakdown? Not only are some of the Big East teams new to the "major" ranks, some were not even in division I-A a few years back. Comparing "major" opponents but including minor teams will skew your results.

A disadvantage to this kind of approach is that it assumes that all opponent are basically the same quality, which is really not the case. For instance, of the PAC-10's non-conference major opponents, 53.5% of their opponents finished the season ranked and they won 62.6% of their games. Compare that to the ACC who's major opponents who finished ranked made up only 33.8% of their schedule and won only 54.0% of their games. Of course, you'd expect the ACC teams to win more of those games.

It's not that this isn't a valid way of looking at things, but I would say that it's not enough information to draw conclusions from, particularly in the absense of comparing similar teams from the opposing conferences (A Team that went 7-1 in Conference A should beat a team that went 1-7 in conference B). College football teams only play 3-6 non-conference games a year, and most of those against non-major teams. You can't assume that there's enough info for the irregularities to iron themselves out.

Anonymous said...

Everyones statement is well thoughtout, but I would like to add that some teams that play more in conference games could be increasing the difficulty of their schedule.

I am a product of the Big 12, so it hurt when I say this... The SEC is by far the strongest conference if you remove all of the quatitative confusion. Many times models lose their value when you have too many variable that don't correlate.

I think we should all just take a step back and go through a few teams in the SEC. (I AM ONLY CONSIDERING THE LAST 5 years)


If the teams listed above did not beat the hell out of each other throughout the season they would all be top 15 teams almost every year.

Ole Miss, Miss. State, South Carolina, Vandy, and Kentucky are not great, but they are do have solid programs that can show up and beat you on any given day.

Jeremy said...

I don't know if this helps, but--

As of 12/2006, here is NCAA D1-A Football over the last 10 years:

1.Ohio St. 97-27 .7823
2.Michigan 96-27 .7805
3.Texas 96-29 .7680
4.Miami FL 93-29 .7623
5.Va Tech 95-30 .7600
6.Fla St. 95-31 .7540
7.Tenn 93-31 .7500
8.Oklahoma 94-32 .7460
9.Florida 92-32 .7419
10.USC 90-34 .7258

Oddly, Lloyd Carr has the most wins and best winning percentage among every coach who's been at the same school the last 10 years. If he'd won the damn Rose Bowl, Michigan would've had the most wins and best winning percentage of any school in the nation bar none over the last decade.