March Madness and, by extension, March Mathness are now over. This project was first and foremost a way to teach ratings and rankings at Davidson College and other schools. Now that it is all over, we’re checking back in with a few of the students to see if they will draw on their new math skills when they fill out their brackets next year.
Overall, I was fairly happy with how my brackets performed this year. None of the brackets performed particularly well in the early stages, but both the Piecewise Massey bracket and my own observation bracket came on strong towards the end of the tournament. In the end, the Piecewise Massey bracket finished in the 94th percentile with 1350 out of 1920 possible points. This was a much better result than I was expecting, and it was largely due to Kentucky’s victory (apparently predicted in 35% of all brackets on ESPN) and getting three of the final four teams correct. This bracket stayed between the 80th and 98th percentile every round, and generally did well throughout, picking several notable first round upsets (NC State and USF) while only really missing on a few teams (the bracket liked Duke and Missouri and had Louisville losing to New Mexico in the round of 32). If a couple very close matchups had gone differently, notably Ohio State-Kansas in the Final Four, this bracket could really have been an all-star.
As far as my other mathematical bracket, based on the adjusted “4 Factors of Winning”, it sadly did not perform up to expectations, scoring only 620 points and finishing in the 23rd percentile overall. This bracket missed badly on a number of picks, among them Missouri beating Kentucky in the Final Four (whoops), UNLV in the Elite 8, and Michigan State losing in the round of 32. In all fairness, however, the bracket got fairly unlucky with Syracuse and UNC, two of its Final Four picks, each losing critical members of their rotation during the tournament at some point. The model obviously cannot take these factors into account, and with the later rounds being worth so much, a few bad breaks really cost the bracket a shot at a good final position. However, this bracket did pick a number of interesting upsets, namely Lehigh over Duke in round 1, USF over Temple, and Ohio into the Sweet 16.
I think I will definitely use math to fill in a bracket next year, although my “competition bracket” will probably be based largely on my intuition, only using math to fill in games I am unsure about. I definitely want to refine my “4 Factors” bracket to account at least a little bit for strength of schedule. I feel as though this bracket had potential, but since I did not find the data until very late in the process, I was unable to really optimize what I was trying to do with my methodology. Beyond that, I will keep watching college basketball when it starts again in November, and keep cheering for the Davidson Wildcats!
This NCAA tournament was full of many fun, surprising, upsetting, and mind-blowing games. It was interesting to see how the mathematical ranking method (Colley Ranking method) held up this March. After all of the madness, my bracket ended up finishing in the 87th percentile–11th in our ESPN group. That was pretty amazing! With this bracket, I used math to rank teams, and I basically went in and edited some of the games based on new team information and my intuition. I knew Syracuse wouldn’t be playing with one of their best players, so although mathematically they were ranked the highest, I predicted they would lose in the Elite Eight. I knew UNC would possibly struggle, but because of personally reasons, I was gunning for them to still make it to the championship game. Their loss to Kansas was the most upsetting for me. The most surprising game was Kansas also beating Ohio St. in the Final Four. But in all, my bracket was pretty successful.
Next year, I will definitely continue to use math to rank teams and to help me determine winners of each round. This year I played it safe. But next year, I may do more research on teams so that I could go through and make changes, and hopefully predict a few upsets!