What About Football?

by Maine Basketball Rankings

I get a couple of questions on a regular basis. The first is whether or not the model could be used for something like college basketball? Generally these people are thinking about gambling and/or their bracket, which is understandable. But also, the model out-performs every college basketball model that’s available to the public. All of them. Naturally, people are wondering if the model is just that good or maybe high school basketball is easier to predict. It’s a fair question and I’m pretty sure it’s the second one. I’m not as smart as Ken Pomeroy. I promise. The second question is if I can do other sports? Because of how the MPA’s website operates, there’s no easy way to import scores, so the model requires a LOT of basic data input and that’s very time-intensive, so I’ve been hesitant to add other sports. Soccer? Softball? Baseball? Lacrosse? I’m only one person. But it does occur to me that maybe football is do-able as a test. They only play once a week. There’s existing research to lean on. Plus (and this is important) there’s enough attention on football that the hours (and hours and hours) could potentially be worth it.

So…let’s try football.

I reserve the right to completely scrap this and pretend it never happened.

I’ve started adapting the model. The basic framework should do the trick, but I guess we’ll see. I have no previous data to lean on, so it’ll be pretty bare-bones until I’m able to add more things and test assumptions.

For the time being, we’ll use the basketball home court advantage data until proven otherwise, which is exactly what I did for basketball. The thinking is to use all the basketball framework until it’s proven to not work and then go from there.

I’ll caution that in starting from zero, it should take a little while for the model to catch up. 8 games isn’t a lot, so we’re dealing with a very small sample size. How quickly the model is able to catch up is an open question. Don’t expect it to be as scary-accurate as the basketball model is. Eventually, maybe.

But, more importantly, I’m not a football fanatic. I watch a lot of football, but not a TON of football. And I don’t know the Maine high school football scene all that well. Those co-operative teams? I have no idea what to call them. I don’t know a really easy way to put “Houlton/Hodgdon/Greater Houlton” in the data visualization without it being way too long. I’m leaning toward “Houlton, etc.” So please tell me things I should know that I don’t. Put it in the comments. Tell me on Twitter or Facebook.

What do I need to know from the jump?

Leave a Comment