Our bracketologies are live (NIT, NCAAT), and with that, we wanted to give a quick rundown of how they’re currently working.
At the moment, we aren’t running our full model. We aren’t simulating the rest of the season 10,000 times. We’ll resume that at some point, I’d imagine, but for this point in the season, we’re running a very scaled back model that just looks at what teams have accomplished so far (using the committee’s metrics of SOR, KPI, and NET), what teams are generally expected to accomplish the rest of the way (using KenPom, a measure of how good each team is, and making the large but fine at this point assumption that their performance will roughly mirror how good they are), which team is the favorite in each conference tournament (to give us our NCAA Tournament automatic bids), and the rough probabilities of bid thieves and NIT automatic bids in each conference.
What does this mean for you, the reader?
Realistically, this isn’t giving us outputs all that different from what the full-scale model would be giving us, and it’s doing it without forcing us to jump through all the hoops we jumped through last year to adjust to constant rescheduling. Our hope is that this Covid surge will peak soon, cancelations will recede, and in a few weeks we’ll be able to roll out the full model with more confidence (we’re also trying to figure out whether there are ways to account for the scheduling uncertainty within our model, but we’re hoping to avoid having to resort to that).
In the meantime, if you see something wonky, it really might be wonky. But it also might not be. We’ll try to dig through the oddest oddities in our daily notes, and as always, we’re happy to field questions. Fingers crossed that the full model is back soon, and better than ever.
