The Problems with Bowl Season

The hot question for years in college football has been, "Should there be a playoff for the national championship?". I have been putting off answering this question for years. I truly believed that there was no correct answer. Each way to decide the national champion is marred by problems. For example, the current BCS model leaves out teams that deserve to be there (i.e. Auburn in 2005, possibly Utah even), without guaranteeing there still won't be a split title (USC/LSU 2003). The old poll system may result in split national championships (Nebraska/Michigan in 1997) as well, and there is no guarantee of the number 1 and number 2 playing each other. Then there's the old, old poll system that awards national titles before any bowl games have been played, resulting in "mythical" national championships awarded to teams that lost their bowl games (Oklahoma 1950, Alabama 1964 & 1973). Finally, the playoff system, though we have yet to see it, would probably lead to teams that are a lock for the playoffs resting starters in their last game (or two). Maybe it's just me, but this practice would taint the relative purity of the college game. So there you have it. Every system has its flaws. The question really should be, "Which method has the most net benefit?". Here's where it gets even more interesting.

The current BCS model gives us the number 1 versus number 2 every year (sort of). Occasionally, it gets it right and we get games like this year's Rose Bowl. Before this time, there is no way to ensure that the best two teams play each other due to conference tie-ins to the bowls. Still, I'd say that this is not so bad, except for the fact that there can be so much argument over who really are the number 1 and number 2 teams in the country. Last year, with SEC power Auburn undefeated at year's end, they were given the shaft by the BCS and locked out of the title game. When Oklahoma got humiliated by USC, USC was declared the undisputed national champion. How fair is that? Does this not just seem like a reincarnation of the old poll system? Two conference powers are still undefeated and have not played. Mere opinion is what separates these two teams, and yet everyone (outside of the South) looks the other way and says, "Yep, the BCS got it right." You have to be joking.

The old bowl system left open the possibility of split national titles because number one and number two were not necessarily going to play each other (not that numbers 3-5 were out of the running for the title). But there were other problems with the subjective nature of it too. If an undisputed championship was awarded, some team would have a serious beef. For example, in 1983, here is how it all went down. Number one and undefeated Nebraska was playing number four Miami (FL) who was 10-1-0. Undefeated and number two Texas was playing number seven Georgia (9-1-1). Number three Auburn (10-1-0) was playing number eight Michigan (9-2). Miami defeated Nebraska 31-30. Georgia defeated Texas 10-9. Auburn beat Michigan 9-7. Rather than Auburn jump to number one, Miami leap-frogged them for the national championship. The records were indentical. Auburn's one loss was to Texas (11-1-0) in the second week of the season. Miami's one loss was to Florida (9-2-1) in the first week of the season. The interesting thing to note is that Auburn beat Florida later in the year 28-21. Nothing can account for the jump other than the subjective nature of the polls and the "what-have-you-done-for-me-lately" attitude of the poll voters. Hence the problem with this system.

Now we'll look at the old, old bowl system where championships were awarded prior to the bowl games, which were viewed solely as exhibitions. When I first heard that this was the way it was done in the past, I laughed at how ridiculous it sounded. There are finally non-conference match-ups to compare teams and they are not taken into account? What's the point? Clearly this could not be a fair way to decide the national champion. And then, along came this year. I have never had a problem with bowl games factoring into the polls. It seems fair, doesn't it? But then let's consider the fact that there is at least a two-week layoff between a team's last game of the season and bowl game. Two weeks doesn't seem so bad, right? An extra week to heal and gameplan will not be turned down. However, with the addition of more and more bowl games, a team's final game can get pushed back to where the lay-off may be over a month. I find it hard to believe that this does not affect the way a team plays. The team that shows up to the bowl game is not usually the same team that finished the regular season. When you throw in the disruptions of final exams, Christmas, and New Year's, it's tough to ensure the right team shows up. Perfect examples from this bowl season are Miami, Auburn, Georgia Tech, and Georgia (in the first quarter). Anyone that watched these games realized that these teams were as flat as they came. This is not the Miami team that beat VaTech, nor the Auburn team that beat Georgia and Alabama, nor the Georgia Tech team that beat Miami and Auburn. What can account for this? It's my theory that the layoff, factored in with not being in a game that truly matters, explains this behavior.

In a rough attempt to test this theory with the limited information I had available, I began to compile the records of the favored teams and categorized them according to when the bowl game was played--pre-Chrismas, post-Christmas/before New Year's, and after New Year's. I also looked at whether the game was a BCS match-up or not, hypothesizing that the favored team would have a better winning percentage in these high-profile games. I was unable to track down the official lines for the games as I went all the way back to 2001. However, I used a line generated by a computer at www.collegefootballpoll.com/2001_archive_bowls.html. Though it uses a computer formula, I figured this variable would stay constant since it was not subjective and would hence provide a stable line to measure the games against from year to year. Here's what I found.

Since 2001, in games played prior to Christmas, and therefore with the shortest layoff, the favored team had a winning percentage of .731. In games after Christmas, the percentage dropped dramatically to .536. Even more interesting is that in the "also-ran" bowls between Christmas and New Year's Day, the winning percentage was a mere .478. This seems to lend credibility to my theory as there is a longer layoff and the games are not high-profile. The period with the longest layoff, on or after New Year's Day games, had a winning percentage of .634, (perhaps because the games are fairly high-profile?). And to break this down even further, the non-BCS, post-New Year's games boasted a winning percentage of .619, while the even later, but extremely high-profile BCS games recorded a .650. To put this all in perspective, the favored team had a winning percentage of .574 in all bowl games since 2001.

This is a relatively small sample, I know, but it did include 136 games. I'm not going to claim that my methodology is flawless, but it clearly seems I am onto something. So do with that information what you will. I will, though, continue on in telling you why I think this information is relevant and how it serves to answer the question regarding how to determine a national champion.

In light of this information, I feel one has to ask, "Did the old, old poll system have it right?". Should we pick our champion prior to bowl games, which were meant only to be exhibitions anyways? The answer, in my opinion, is a resounding "NO!" but that does not mean that we cannot learn from this. Going back to the old, old system would be relying solely on subjective opinions of how teams played during the regular season. There is room for this somewhere, but when eight of twelve games are played in-conference and four split between the MAC, Sun Belt, and Conference USA (or some other rather dull collection), voters learn very little. We do have notable match-ups during the year, but they are few and far between. It is tough to stack one conference up against another. I think there is a way to deal with this problem.

First, a playoff system would work. There are still the problems that I mentioned above, but I want to focus on a new one: selecting the eight or sixteen teams who would play. Even with a knowledgeable selection committee, this in an almost entirely subjective process. There needs to be a way to eliminate some of the subjectivity without letting an algorithm determine who gets to play for the national championship. An alternative to eliminating subjectivity would be to create more non-conference match-ups between the BCS conferences. Even with a twelve-game schedule next year, not every team will play a team from another BCS conference. That's just the way it is. Here is my radical suggestion, that I will be the first to admit has some serious short-comings, but remember, I am trying to provoke some new thought here, not fix the world. In my world, though, we have a playoff and not the BCS, so I will now refer to the current BCS conferences by their rightful names. The Pac-10, SEC, ACC, Big East, Big 12, and Big Ten should all limit membership to eight teams. This would leave seventeen teams out in the cold, but these teams (or at least sixteen of them) can create two more legitimate conferences, leaving one team (Duke?), to join a lesser conference. Each team plays every other team in its conference (7 games). That determines who wins the conference championship, utilizing a series of tie-breakers, should it happen. I am not a huge fan of conference tie-breakers that are not based simply on head-to-head, but I am willing to make this sacrifice for the greater good. So we now have eight power conferences, each with eight teams. Hmm...what sounds familiar about that number....64. Interesting. We have five games remaining on the schedule. Four of those should be determined on a rotating schedule between other power conferences, and a rotating schedule between teams within the other conference. This means that each team will play four other schools from power conferences each year. The fifth game can be used however the school sees fit. By giving this freedom, smaller schools will not be robbed of the payday that keeps their women's lacrosse team from folding. At the end of the season, there is now a pretty good idea of who the good teams truly are, as they have had to prove themselves repeatedly in other conferences. Unfortunately, I think a selection committee may be the way to pick the teams from here on out. As convenient as it sounds to say we should treat the conference games as a playoff with the conference champion getting the invite, it is quite possible that one conference could produce several 11-1 teams, while another conference (cough cough, Big East) could have its conference champion be 7-5 or lower. Conference championships are nice, but they should not grant a free pass to playing for a national championship. At the conclusion of the regular season, the playoff begins--on the weekend usually reserved for conference championship games. Just like that, we are already down to either eight or four teams. I am in favor of the sixteen team bracket, so we'll say we have eight left. The week before finals, another round is played, cutting the field to four. The semifinals and finals are played on consecutive weekends, in my opinion. Though this does not allow a break for Christmas, current college bowls do not seem to care about it much anyways. If a modification must be made, I'd suggest a week off after the regular season, then two rounds prior to Christmas, a week off for Christmas, and then the semifinals and finals following that and ending at approximately the same time as they currently do. This means a possibility of playing sixteen games, but I do not think any college player still playing that late in the year would care.

Basically, there is my idea. Games that matter every week. Not only does this fix the postseason picture, but it creates a lot more interesting regular season games.

I am not under the delusion that my statistical methodology is flawless or that my playoff scenario is perfect, hence, I would truly appreciate comments disputing what I have said here and offering alternatives.