We were hoping to have a NASCAR model rolled out today. Friday, actually. Or this past Monday. Didn’t happen. NASCAR’s hard for us to model. But we’re making progress.
Anyway, since we’ve written before about our models and our ambitions for future modeling (it’s a big part of The Barking Crow’s future prospects), it seemed like a good time to give an update. We’ll go through blow by blow:
NASCAR
This is going on the back burner. We’re hoping to have it out there in the next month or so, but with college basketball modeling still the focus (and a headache and a half right now because of scheduling uncertainty), it might not happen until April.
NHL
This is the priority behind NASCAR, so April’s probably a reasonable ETA for it as well. It’s easier to model than NASCAR because of the binary nature of team sports, but I don’t personally know a lot about hockey, which poses challenges.
FCS
We’d like to bring back our FCS Bracketology, and we’re hoping to pump out a quick model this week based around our 2018 model that’ll allow us to do that. We wouldn’t have playoff probabilities, because we aren’t confident in our ability to model the scheduling uncertainty, but we’d have a bracket. We’re hopeful this can be quick, but if it isn’t, it too will go to the back burner.
College Basketball
Our one live model right now. It’s been a tough year to use it because there’s so much uncertainty, including uncertainty about uncertainty, which in technical terms means that we don’t know how wide our error margins should be with the committees working with different sample sizes from those of prior years, and with different sample sizes between teams, even. When we’ve settled the FCS model, either by publishing it or shelving it, we’ll turn our attention to getting the college basketball schedule, tiebreakers, and conference tournaments hammered in for each league. For the time being, we’re finagling the schedule we input to our model with assumptions that all postponed conference games will be made up, that leagues will reschedule to get all teams the same number of games, that conference tournaments will be on neutral sites, and that conference tournaments will operate with the same formats within which they operated last year. This isn’t accurate, of course, but for the purpose of simplicity and within the time constraints we have, it’s the best we can currently do.
Future Models
With the NCAA announcing championship dates for basically all sports, the prospect of modeling college hockey/baseball/softball/volleyball/soccer/lacrosse is back on the table. Again, though, given the scheduling uncertainty, this might not be the best year to try those. We may be better served taking the time we would spend on those and dedicating it to hard skills so that we can improve our models’ quality and user experience.
With a part-time website with limited resources, there are always tradeoffs. These are the ones we’re weighing and making right now. Hopefully by this time next year we’ve got one or two dozen models finished and functional. That’s the goal. We’ll see how close we get to it.
