A fascinating tournament. OK, my picks are totally blown. But beyond that. I have built a Bradley-Terry class model that gives the probability of each team beating every other that fits the regular season as well as possible. Given this belief structure (a set of 604 parameters) I can simulate hundreds of thousands of tournaments. This tournament (or one even more extreme) has like a 1 in 500 probability. (There are different ways to measure weirdness, but they are all similar). I’ve been doing this modeling for like 10 years now. So, a multiple comparison correction gets me to like 1 in 50 decadal probability. Makes me think teams either do, or are starting to, play differently in the tournament than in the regular season. You basketball hotshots out there–do underdogs play a higher variance game–like more 3-point attempts or such in this single elimination structure? Shees, am I going to actually have to learn something about basketball to build a sensible model? Fe