A bracket — a very nerdy one — where even your college can win.
Every year filling out my March Madness bracket I always wondered if there was a model where my alma mater Bucknell could squeeze out a championship. Although that seemed unlikely, this created the inspiration for NerdyBracket.com, where I use mostly meaningless datasets to determine winners of the tournament in hopes that a mid-major can pull out a title. Nerds across the internet seemed to share my interest as the site was picked up online by the NY Times and NBC Sports.
This year I relaunched the site with some even more bizarre determinants for bracket winners including acceptance rate (won by Yale), quality of campus food (Dayton), proximity to the equator (Hawaii), campus acreage (Duke), and endowment size (Yale, yet again). Armed with a massive spreadsheet and with the help of a freelance web scraper, I collected as much data on every Division 1 school as I could before Selection Sunday.
I’m curious to see if any of these ridiculous brackets can beat predictions by experts Nate Silver at FiveThirtyEight, Joe Lunardi (ESPN’s resident bracketologist), or President Obama. I think the best bets may be Largest Campus (Duke over Michigan St in the finals) or Worst Parking (Cal over Kentucky), as these brackets tend to skew towards larger colleges.
And if one of my brackets does do better than these experts (which I highly doubt), what does that mean? Is FiveThirtyEight not to be trusted? Well, it means absolutely nothing. And if you don’t believe me, you should probably read Nate Siver’s The Signal and the Noise where he’ll really hammer it home that a good model should be probabilistic, not deterministic.
But you never know, a Nerdy Bracket may be just be the perfect bracket. The odds of that? 1 in 9.2 quintillion.