Thursday, January 08, 2009

The Gus Rankings

I don't believe I've posted about this before, so let me explain. On an annual basis, college football has a serious issue on its hands-- how to select the two teams that play for the national championship. As you know, the Bowl Championship Series generates a calculation that determines who will play. It's not perfect; I'm not going to talk about that.

Well, my uncle had an idea for an alternative way of ranking teams a few years back, and it's fun to see how they line up with what the BCS generates at the end of the year. (I don't know if it was originally his or not, but I haven't seen it elsewhere.) We assign points to every team in the following manner: For every win a team has, they get a number of points equal to wins of the team they beat. So, if Texas Tech beats Oklahoma State, they get 9 points since Oklahoma State won 9 games this year. For every loss a team has, they lose points equal to the number of losses of the team that beat them. So, if Oklahoma State loses to Texas Tech, they lose 1 point, since Texas Tech lost one game. You do this for every team on the schedule for the entire season. Since more games means more points (for the successful teams), dividing by the number of games played needs to be done and, viola, you have the Gus Rankings.

Some thoughts on the Gus Rankings:

- Clearly, the rankings are completely objective, which has its positives and negatives, but on net I think it's a positive. Margins of victory don't count, nor does playing at home or on the road, nor does the timing of your losses, nor does the conference you play in.

- It's tough to figure how to manage games against I-AA teams (I believe they are called FCS teams now). The general idea behind the Gus Rankings is to give more points for beating teams with better records. Better teams have better records (see next comment for more on this), but if you shift divisions, this isn't true any more. You clearly shouldn't get as many points for beating a 11-1 I-AA team as for a 11-1 I-A team. For the time being, I excluded those games against I-AA competition from team's schedules.

- Here is what I consider to be the best aspect of the Gus Rankings. As it sits now, teams in BCS conferences have no incentive to play teams from outside of BCS conferences. They are perceived as weaker competition, so even a win against one of these teams is taken with a grain of salt, and a team could get snake-bitten by a team that turned out to be pretty good. Take the argument made by the BCS conferences as to why they shouldn't make it into the title games-- they run up gaudy records against poorer schedules, and therefore aren't actually that good themselves. If they believe that to be the case, then those are exactly the types of teams that you want to schedule if the Gus Rankings were used. You'd love nothing more than to play a 11-1 that has supposedly run up its record against bad competition; after all, you get a lot of points from a poor team. As the years go on, teams that persistently run up good records against poor compeition will end up on the schedule of better teams, and their true value will be ascertained on the field. This is a very good thing.

- There is a huge upside and a small downside to playing a very good team. Let's say you anticipate that USC is going to be good, so you schedule them, and they end up going 11-1. If you lost to them, you only lose a point, and if you were the team that beat them, you get 11 points. Conversely, it's not good to play a bad team-- you don't get a lot for winning and you could lose a lot if you lost the game. Insofar that good teams playing other good teams is an attractive proposition, the Gus Rankings are a big plus in this regard.

- The rankings really only make a lot of sense at the end of the year. Of course, with the BCS, the final one is the only one that ends up mattering, but it is fun to follow along through the season.

- Conferences with a title game will give a bump to the winner; it adds a game against good competition to the overall score. Thus, conferences with title games should produce champions with a better score. (Remember, it's an average, so as long as the points gained were more than the average points gained in a game throughout the season, the ranking goes up.)

- Good conferences end up with a bump for all their teams. Consider within conference games; they're a wash since teams will play a range of good to poor teams within the conference schedule. But consider now a conference that has performed well outside of its league games-- there is a bump up for points to be had from wins and a bump down for points to be taken away from losses. Unbalanced conference schedules mitigate this a little bit, but the overall idea is still the same-- better conferences get more points, but not by subjectivity. Their good teams get more points by beating teams that themselves have performed well. Rewarding strong conferences in an objective way is also a very good thing.

The BCS rankings are as follows:

1. Oklahoma
2. Florida
3. Texas
4. Alabama
5. USC
6. Utah
7. Texas Tech
8. Penn State
9. Boise State
10. Ohio State

The Gus Rankings are as follows, pre-bowl game results:

1. Oklahoma 6.417
2. Florida 6.250
3. Texas 6.167
4. Utah 5.727
4. Boise State 5.727
6. Texas Tech 5.500
7. Alabama 5.000
8. Ohio State 4.909
9. Penn State 4.818
10. USC 4.750

This is only the top 10 from above re-calculated-- it takes a decent amount of time to do this, since you have to find every team they played and then adjust THEIR record for bowl game results too. I realize now that I ignored the secondary I-AA games-- that is, I ignored them for those teams above, but if these teams played other teams that beat I-AA teams, then they got a bump up for that they shouldn't have received. Conferences that beat up on I-AA teams would be biased upwards here; take from that what you want. I really should write up an Excel sheet to do this for next season to take everything into consideration.