NP: John Cage, 4'33

Not crazy about what I was able to come up with for the NCAA tournament, mostly because I can't find a way to get Georgetown to beat Davidson without overweighting team seed as a variable.

I made one bad tactical mistake in grabbing data for this, in that I cherry-picked stuff instead of grabbing the whole two lines of data from ESPN. If I had more to work with, I could have tuned the model a bit more to avoid unreasonable upsets.

Plus, I could have created a better index based purely on what I think is important. The original notion was to go with rebounding, but there's not a lot of differentiation there, and it has Cornell making it to the Final Four. If I could find opponents' rebounds per game, that might be different. But the other key element would be free throw attempts per game, in that I want to reward teams that drive to the basket. Plus maybe steals and blocks for a stronger measure of defense.

Anyway, it is what it is, and if Boise St. knocks off Louisville, I'll know that I didn't trust the math enough, because that was sort of my "smell test" for any model.

UPDATE: I had minimized the seed weight while I was messing around, and hadn't set it back. Still can't flip the Davidson over Georgetown upset, but Texas now beats St. Mary's, which was the other one that had me a little worried.