On March 15 many of us scrambled to complete our brackets and 64 of us placed them in the PUP March Mathness group of the ESPN Tourney Challenge. Almost everyone was in the top 50th percentile and 10 of our group did better than 90% of the 6.5 million people who entered a bracket. What we had in common was that we used a particular math algorithm or some combination to fill out our brackets. Quite a few of us also tossed in a bit of the human element because everything can and will happen in a tournament.
Math editor, Vickie Kearn, asked Tim Chartier, Professor of Mathematics at Davidson College for his take on how we did.
“I think one of the things to note is that such methods can clearly be effective. However, that personal modeling decision as to how to weight the season or even what part of the statistics to use is VERY important and makes a big difference. Even so, there is that inevitable “Madness” of human endeavor that keeps us ever watching and, frankly, enjoying! What will happen? We will never find a method to always know and as such, we keep trying and keep watching.”
And the Winner is Travis McElroy
Travis had the top rank in the PUP March Mathness Group, finishing with a total of 1470 points, in the 91.8 percentile. He is the recipient of several Princeton University Press books.
Travis is a senior math major and Spanish minor at Davidson College. Next year he will be working on his master’s in applied mathematics at the University of North Carolina at Chapel-Hill. Following he tells us about his experience in completing his bracket.
This bracketology experience has been fun as I normally don’t pay attention to the tournament as baseball, football, and tennis are the sports that I watch. I used the Colley Method and tried all different weights. I split the season into 42 different parts but really only put weights on 3 different sections, pre-conference play, conference play, and conference tournaments. The conference tournaments part was mainly for major conferences. So I tried a range of .3 to 1.5 for how much each game was worth and also for some brackets I added a bonus of a win or 3 for a road win.
For my winning bracket I decided to do something crazy and pick the underdogs. I used a colley method of .3 wins for a pre-conference play win, 1 win for conference play, and 1.5 for a tournament win. Also, a road win counted as 2 wins. After the rankings were compiled, I decided that if the 2 teams were in the top 10 and the lower ranked team was within 5 rankings of the higher team, the lower ranked team would win. After the top 10, if the lower ranked team was within 10 of the higher ranked, then the lower ranked team won. The only exception I made was I said Kentucky would win the whole thing.
Next year I would create more sections of the season and make the rankings vary more. It was very difficult to get any ranking that did not have Kentucky as number one and Syracuse as number 2. I also want to compare the actual ratings to see if there was a reasonable pattern to when the “underdog” won. I am very surprised I did as well as I did as I do not follow college basketball closely (except my Davidson Wildcats) and this bracket was my fun bracket. It shows that the underdog is not to be discounted. The only big surprises for me was the same surprises as everyone had with 2 #2 teams losing in the first round. Next year I also want to combine methods to see if that helps. My roommate (Greg Newman) used different methods than I did and we want to see if we can make a better combination for next year.
The last word(s)
Now that you have 12 months until the next tournament, you have lots of time to decide what method you will use next year. Get your copy of Who’s #1? by Amy Langville and Carl Meyer and start reading. You can also use it to predict the outcome of any sport as well as be able to rate and rank just about anything.