Now a BCS Tournament?
In the wake of Boise State's thrill-ride finish to its undefeated season, many in the popular press are calling for the NCAA to implement a playoff for the college football season in order to be certain to crown the true national champion. Under the current formula for BCS (Bowl Championship Series) games, the contest that pits the top-ranked BCS teams is designated the national championship game. Implementing a more elaborate playoff series wouldn't be too hard, conceptually at least: just push some bowl dates around, pick an even number of teams, set them in brackets, and begin. Or the NCAA could just tack on one additional game (for 2006, Florida vs. Boise State) to determine the champion.
All of these approaches would suffice to crown a winner. But would the winner of the tournament be any more the "true champion" than the team that wins today's BCS championship game? (I've written on the trouble with tournaments before, here.)
Of course not. There's no way to determine a college football champion. So give it up.
1. The current BCS uses a complex formula (aren't they all?) to rank teams. The formula includes a heavy dose of opinion polling, computer rankings, comparisons of schedule strength and game outcomes. In short, although everything in the end is expressed numerically, there's a ton of subjectivity built into the rankings. Even computer-generated ranks have necessary subjective element in their design. Subjectivity is inevitable; anyone who has ever put a few nickels on a football game knows just how little even experienced observers can tell about the strength of a football team. "Upsets" happen all the time; one explanation for the frequency of upsets is that we really cannot tell before the fact which team is the better team. Imagine the difficulty of putting a couple of dozen plausible candidates into a rank order. Now imagine that rank order being meaningful (especially financially) and we can see what the BCS is up against.
2. But assume perfect information and perfect analysis. (Those BCS people are wicked smart.) What are they measuring? What does it mean to be the "best" team? Even a significantly better team won't ordinarily be able to beat the next-best team all the time. Assume Team A would beat Team B eight out of ten times: Team A is decidedly better. Yet when the teams actually face each other (if they do), the game could be one of those two in which Team B wins. So, if the job of the BCS is determine which of these two teams would beat the other more often, then the actual game outcome should be rather irrelevant to the question. If Team A is substantially better than Team B, then even if Team B wins the actual contest (as it might, twice in ten games) , Team A is still the better team and should be ranked higher. In other words, should rankings be determined "on paper"? The actual outcomes of games will only tend to mislead us. (That's why I seldom watch college football; I don't want to be distracted in determining which team is best.)
3. Most would say that the "best" is determined by who wins on the field, not on paper, and so game outcomes must determine rankings. Luckily, college football is pretty widespread and the leading teams don't play each other too often. If they did, then we'd have a mess. Team A beats B, B beats C, then C beats A. So we have to go back to eyeballing the teams and deciding that Team B, although it lost to A, is actually better than A. So we're back to deciding rankings on paper, so to speak.
4. With that said, at this point everyone's pretty convinced that, were the NCAA to add just one more game, Florida should play Boise State to more accurately determine the champion. But assume Boise State had lost to Oklahoma and was out of the picture. Arguably, the next best team in America after Florida would be Ohio State, which is ranked second. Should Florida and Ohio State play again? What if Ohio State wins? Now we have a problem.
5. Strangely, college football rankings are made easier by the fact that the top teams are spread out among the six BCS conferences, a few others, and Notre Dame. The result is the top teams don't play each other enough to determine which team is best (assuming that we can tell which team is best by the outcome of the games.) If they played a lot, they'd beat each other willy-nilly and we'd never sort it out. So opinions count, a lot. One problem with opinions is that they are in part the consequence of one's view at the outset. We form opinions, defend them and are slow to abandon them. So, to a certain extent the rankings at the end are a product of the rankings at the beginning. Here's an argument to that effect. For example, in 2004 Oklahoma, USC and Auburn all finished the season undefeated. Oklahoma and USC were ranked high in the preseason; Auburn started in the teens. So Oklahoma and USC got to play for the national title while Auburn got to watch. (Why do we rank preseason?; this only exacerbates the problem.)
6. So, let's assume we can push past all these problems of definition, subjectivity and the rest and actually rank the teams. What would a tournament resolve? Surely no one believes the winner of the NCAA tournament is truly the best team in the nation. It's called Madness for a reason. Any team, even the best team, will lose some percentage of its games to inferior opponents. The brevity of the contest and the element of luck ensures it. In fact, the odds are probably against the best basketball team winning the tournament. Why would a single-elimination tournament in football be any more likely to identify the nation's best team? Footballs are oblong and take funny bounces. I think luck has more to do with determining (close) football contests than any other of the big American sports. And scoring is hard in football; as a result, lots of football games are close, ending with a margin of victory of less than a single score.
7. My point is that college football is inherently resistant to the kind of organization and structure that the BCS beefers want to impose. We don't know what we mean by the best, nor how to go about measuring it even if we did. And beyond that, we don't really have a mechanism plausibly available to determine the champion. People may want a championship tournament, as so soon will the NCAA, I presume. A tournament may be justified for entertainment's sake or to make more money. But let's not say we need one to determine which team is the best in the land.
8. Boise State by 3.
All of these approaches would suffice to crown a winner. But would the winner of the tournament be any more the "true champion" than the team that wins today's BCS championship game? (I've written on the trouble with tournaments before, here.)
Of course not. There's no way to determine a college football champion. So give it up.
1. The current BCS uses a complex formula (aren't they all?) to rank teams. The formula includes a heavy dose of opinion polling, computer rankings, comparisons of schedule strength and game outcomes. In short, although everything in the end is expressed numerically, there's a ton of subjectivity built into the rankings. Even computer-generated ranks have necessary subjective element in their design. Subjectivity is inevitable; anyone who has ever put a few nickels on a football game knows just how little even experienced observers can tell about the strength of a football team. "Upsets" happen all the time; one explanation for the frequency of upsets is that we really cannot tell before the fact which team is the better team. Imagine the difficulty of putting a couple of dozen plausible candidates into a rank order. Now imagine that rank order being meaningful (especially financially) and we can see what the BCS is up against.
2. But assume perfect information and perfect analysis. (Those BCS people are wicked smart.) What are they measuring? What does it mean to be the "best" team? Even a significantly better team won't ordinarily be able to beat the next-best team all the time. Assume Team A would beat Team B eight out of ten times: Team A is decidedly better. Yet when the teams actually face each other (if they do), the game could be one of those two in which Team B wins. So, if the job of the BCS is determine which of these two teams would beat the other more often, then the actual game outcome should be rather irrelevant to the question. If Team A is substantially better than Team B, then even if Team B wins the actual contest (as it might, twice in ten games) , Team A is still the better team and should be ranked higher. In other words, should rankings be determined "on paper"? The actual outcomes of games will only tend to mislead us. (That's why I seldom watch college football; I don't want to be distracted in determining which team is best.)
3. Most would say that the "best" is determined by who wins on the field, not on paper, and so game outcomes must determine rankings. Luckily, college football is pretty widespread and the leading teams don't play each other too often. If they did, then we'd have a mess. Team A beats B, B beats C, then C beats A. So we have to go back to eyeballing the teams and deciding that Team B, although it lost to A, is actually better than A. So we're back to deciding rankings on paper, so to speak.
4. With that said, at this point everyone's pretty convinced that, were the NCAA to add just one more game, Florida should play Boise State to more accurately determine the champion. But assume Boise State had lost to Oklahoma and was out of the picture. Arguably, the next best team in America after Florida would be Ohio State, which is ranked second. Should Florida and Ohio State play again? What if Ohio State wins? Now we have a problem.
5. Strangely, college football rankings are made easier by the fact that the top teams are spread out among the six BCS conferences, a few others, and Notre Dame. The result is the top teams don't play each other enough to determine which team is best (assuming that we can tell which team is best by the outcome of the games.) If they played a lot, they'd beat each other willy-nilly and we'd never sort it out. So opinions count, a lot. One problem with opinions is that they are in part the consequence of one's view at the outset. We form opinions, defend them and are slow to abandon them. So, to a certain extent the rankings at the end are a product of the rankings at the beginning. Here's an argument to that effect. For example, in 2004 Oklahoma, USC and Auburn all finished the season undefeated. Oklahoma and USC were ranked high in the preseason; Auburn started in the teens. So Oklahoma and USC got to play for the national title while Auburn got to watch. (Why do we rank preseason?; this only exacerbates the problem.)
6. So, let's assume we can push past all these problems of definition, subjectivity and the rest and actually rank the teams. What would a tournament resolve? Surely no one believes the winner of the NCAA tournament is truly the best team in the nation. It's called Madness for a reason. Any team, even the best team, will lose some percentage of its games to inferior opponents. The brevity of the contest and the element of luck ensures it. In fact, the odds are probably against the best basketball team winning the tournament. Why would a single-elimination tournament in football be any more likely to identify the nation's best team? Footballs are oblong and take funny bounces. I think luck has more to do with determining (close) football contests than any other of the big American sports. And scoring is hard in football; as a result, lots of football games are close, ending with a margin of victory of less than a single score.
7. My point is that college football is inherently resistant to the kind of organization and structure that the BCS beefers want to impose. We don't know what we mean by the best, nor how to go about measuring it even if we did. And beyond that, we don't really have a mechanism plausibly available to determine the champion. People may want a championship tournament, as so soon will the NCAA, I presume. A tournament may be justified for entertainment's sake or to make more money. But let's not say we need one to determine which team is the best in the land.
8. Boise State by 3.
posted by TSLP at 1/09/2007
Comments on "Now a BCS Tournament?"
post a commentLinks to this post:
Create a Link