I have been going through some reflection on how exactly I rank teams. It started with a never-ending Twitter debate with Ramzy Nasrallah on Thursday that started as I was wrapping up my day at the office, continued as I was driving out of Midtown towards the Regal 24 at Shallowford, and only ended when the previews ended and Skyfall started. The argument was over how far Alabama should have dropped when they lost to Texas A&M and the main point of contention concerned my tendency to use computer rankings like Sagarin and SRS. The argument from Ramzy (bolstered at the outset by Brian Cook before Brian almost certainly got tired of the petty sniping) is that those rankings may be predictive, but they are not good at telling us who has actually had the best season. My counter is that the predictive power of the Sagarin Predictor comes from its interpretation of past results, so it is in effect a referendum on the season to date.
The debate led me to start thinking about my rules for rankings. Specifically, what are the circumstances under which Team A should jump Team B when Team B has a superior record. How much of a difference in strength of schedule is required to make that jump? How much of a difference in margin of victory is required? To me, those are the two relevant factors: whom did you beat and by how much? Ramzy argued at one point that college football produces relatively small sample sizes, so using a computer rating is a fool’s errand. My response would be that if we are dealing with a small sample size, then we need to use all available data, so casting aside margin of victory – a factor that supports Alabama because they were not forced to win a close game until November – is a poor decision. But how much should we use it? I still have Bama a spot behind Notre Dame, despite the fact that the Tide would be a double-digit favorite on a neutral field. So what are my rules for using Sagarin?
And how much should I use the eye test? I am a big believer in the idea that basing assessments on how a team looked to me is a terrible idea because there are all sorts of issues with human perception that cause us to make bad judgments. One pretty spiral or form tackle can become our mental representation of a team, causing us to overrate them. And yet, there is at least one instance in which I am letting the eye test affect my ranking of a team. Florida is 10-1, they have played a brutal schedule, and according to both the Sagarin Predictor and SRS, they would be a one- to two-point favorite over Georgia on a neutral field. As a friend pointed out to me yesterday, Florida’s yards per play margin has been better than Georgia’s in each of their five games against common opponents. Still, I have Florida behind Georgia because every time I see Florida, their offense is appalling. I cannot fathom the idea that a team could need a fourth quarter rally against Louisiana-Lafayette, culminating in a walk-off blocked punt touchdown, and then play for the national title a few weeks later. Florida’s yards-per-play margin is worse than Georgia’s overall, but how much of that is down to playing a much tougher SEC schedule and not bullying overmatched non-conference opponents in the same way? How much value do I give to the fact that Georgia beat Buffalo worse than Florida beat Bowling Green?
To sum up, I’m trying to think of rules to govern my irrational preferences, but I’m finding it hard to tie myself to the mast in a rational way.