Working as a consultant means I’m always trying to analyze stuff. Being a huge fan of college football makes me very interested in all things related to the BCS (especially when my beloved Gamecocks are nationally relevant - it was fun seeing that logo on the board behind the talking heads during the admittedly farcical BCS selection show Sunday night). Marry the two, and voila: A nerdalicious blog post!
First, a preamble. In my perfect world, we’d have an 8-team playoff with a selection committee that used something like the BCS standings as one criterion. I’m also an SEC guy, so I’m biased. But given the current system, I believe an LSU-Alabama rematch is the best possible outcome. Here are my reasons:
- Alabama has the best single win between the two teams. This is because I think Arkansas is a better team than anyone on Oklahoma State’s schedule. Yet Alabama handled the hogs handily. I will grant, however, that Oklahoma State has beaten a larger number of quality teams than has Alabama.
- Alabama has by far the better loss than Oklahoma State. They lost by 3 in overtime to LSU, as opposed to losing to an average Iowa State team. Heck, even Stanford’s loss to Oregon is better than that.
- With the exception of that loss, nobody seriously threatened Alabama all season. Penn State needed a garbage-time touchdown and 2-point conversion to come within 2 scores of the Tide, and even that’s a stretch, as the final margin was 16. Nobody else even came that close. Their average margin in their 11 victories was THIRTY! OSU was nearly as dominant, but they almost didn’t make it all the way back against schizophrenic Texas A&M, eking out a 1-point victory, and Kansas State put a scare into them in a 52-45 shootout.
- When in doubt, go with defense.
Now, most of my reasoning is subjective. What’s fascinating about the BCS is that it is a combination of subjective (human rankings) and objective (computer, or, more accurately, algorithmic rankings). The end of the regular season is a good time to look at the interaction of these two factors. I think it’s really fun, actually.
Rationale for This Exercise in Nerdiness
The underlying assumption of what I’ve done here is that the computer rankings are an objective, unbiased measure of what a team has accomplished during the 2011 season. The are certainly not a perfect measure. In fact, much better measures are possible. But some factors (such as margin of victory) are considered no-nos by the BCS’ silly rules. Dan Wetzel, author of Death to the BCS, has written a great article that only begins to address this issue here. Nevertheless, I’m going to go with what we’ve got.
Human polls, by contrast, measure both accomplishment and potential. In August, it’s all about potential. By December, it’s probably mostly about accomplishment, but still somewhat about potential. Fundamentally, that’s why we have controversy. One team may have accomplished more, yet another might present a bigger challenge to an opponent. In an infinitely long season, these factors would eventually converge. Even a 12-game season, they do start to converge - there’s relatively difference in the top 10 teams between the humans and the computers, for instance. But they don’t get all the way there, which is why some subjectivity is in order. After the 2004 season, the BCS formula was revised to put more weight on the subjective. Unfortunately, that leaves it even more open to (justifiable) criticism about human biases. Joy.
What I’ve Done
Anyway, I’ve chosen to do some basic analysis of the BCS data, comparing the purely objective computer rankings to the more subjective human polls. (Here’s a link to an explanation of the formula and the sources for the inputs: The Evil BCS) To do that, I did this:
- Since the official BCS standings only report ordinals for the top 25 teams in each poll or formula, I had to back-fill the missing rankings for any team ranked 26th or lower by digging into each one individually. (Google was my friend here.)
- Because Southern Cal is on probation, it is not eligible to be ranked in the BCS. Of course, their games are relevant to all the other teams, so the computer rankings include them - the BCS just bumps everyone below the Trojans up one spot. I did the same for the ones outside the top 25.
- For the human polls, I used the “others receiving votes” notes to fill out the list for those ranked outside the top 25. The problem is, Auburn did not receive any votes in the coaches’ poll. Because 35 other teams did, I assigned them a ranking of 36 for purposes of my analysis.
- I calculated an average of the ordinal rankings from the 2 human polls, as well as an average of the 6 computer rankings. The BCS tosses out the highest and lowest computer ranking for each team, but I have not done this because I wanted to capture as much objective data as I could.
What I Found
Before I get into some conclusions, here is my data sheet. OK, that’s done. Here we go…
Over- and Underrated Teams
By simply subtracting the “human average” from the “computer average” and sorting in descending order, the humble analyst can determine which teams are perceived more positively by humans than their on-field accomplishment would suggest. In fanspeak, this means the ones that are overrated. One can, of course, also see which teams have accomplished more than they’ve gotten credit for - the underrated.
Here are the five most “overrated” teams in the BCS top 25:
My fellow SEC compatriots will note with a smirk how many of these come from the B1G conference. As it turns out, the conference’s other representatives, Nebraska (20th) and Penn State (22nd) have almost no difference in their human and computer scores. Nor do my South Carolina Gamecocks, who are “overrated” by only half a spot. (Aside: the Gamecocks’ BCS ranking feels about right to me as a fan of my team. I think 9 is a good spot. We might be better than K-State and worse than Wisconsin, but 9 passes this fan’s sniff test.) Another team that’s rated the same by both: LSU at numero uno (duh!). The most obvious of these to me is Virginia Tech, whom I consider to be grossly overrated by the humans. In fact, I’m surprised the difference here is only 4-ish spots.
On the other side of the equation are the most “underrated” teams:
Texas and Auburn had a handful of losses each, but played brutal schedules. Heck, Auburn played 4 of the top 9 and 6 of the top 17! Oklahoma’s story is interesting. To me, they are the biggest underachiever of the season. Even so, the computers have them ranked 7th, on average, so they are still underrated. The astute reader will observe that 4 of these hail from the Big XII. This is the biggest story in the data to me. Fans of the Big XII have a legitimate beef in that every single team from the conference is ranked higher by the computers than by the humans. Put kindly, this means the humans essentially think the conference’s teams have overachieved. Put lest kindly, it means voters are biased against the conference. I can’t imagine why…
Who’s Number 2?
If the question is, “Who deserves to be LSU’s opponent based on what they have accomplished this year?” then the answer, according to the computers used by the BCS, is Oklahoma State. But it’s not as cut-and-dried as some talking heads who say things like “Clearly, Oklahoma State has the better resume” would have you believe. The Cowboys’ average computer rank is 2.3, while the Tide’s is 2.7. What that means is that 2 of the algorithms (Jeff Sagarin’s and Peter Wolfe’s) actually think Alabama has the better resume, while the other 4 say Oklahoma State does.
Besides, that’s not really the question. Or is it? See, one of the problems is that the very question of being “deserving” is a subjective one. Should it be based only on achievement? Potential? Some combination? What about consistency of performance? How about who would be favored on a neutral field? Is offense or defense more important? What about injuries and overcoming them? As far as I know, the guidelines for voters in the human polls don’t really give clear direction on this. I’ve given my reasons for thinking Alabama is the right choice, and I’m sure every voter who put them at number 2 has his or her own reasons.
I’ll close with something completely worthless: my own ranking of the BCS top 25. Note that there are a few teams not listed that I think are among the best 25 in the country. Southern Cal is a top-10 team, in my opinion, and Notre Dame and Florida State are both somewhere in the 20-25 range. But I wanted to limit this to BCS-ranked teams. Here we go: