The 2017 NCAA season ended with a win by UNC over Gonzaga in a title game marred by both poor shooting and excessive officiating. I didn't have much time/enthusiasm for college basketball this year, but I did want to take a moment to congratulate the winner of the Machine Madness competition... me. :-)
My entry barely squeaked out a win over Erik Forseth's entry. The Net Prophet entry to tournament pools takes the base prediction out of the prediction model and then picks some number of upsets (usually 8). Two of those upsets this year were Oregon to the Final Four, and UNC over Gonzaga in the final game. (Like most machine models, I had Gonzaga stronger than UNC at the start of the tournament.) Over at the ESPN Tournament Challenge, the Net Prophet entry finished with 1500 points, good enough for 19,407th out of 13+ million brackets. Not bad.
Lest my head swell too much, I have to point out that over in the Kaggle competition I finished in 416th place. This year I didn't have much time to spend on the contest, so I ran the code for last year with some quick hacks in place to get it working and although it generated a legal entry, I had no faith in the results, so I'm not surprised to see the poor showing. But my model has always done worse on Kaggle than in bracket competitions, so perhaps this was not a surprising result.
More notably, Monte McNair and the aforementioned Erik Forseth finished 3rd and 4th in the Kaggle contest. That's tremendously good results for both of them, congratulations!
