Of all the years for Warren Buffet to offer $1 billion for a perfect bracket, he made a really good choice. As hard as it is to go perfect any year, 2014 proved to be the most difficult of all time.
There are many ways to calculate how surprising a March Madness tournament is. The simplest way is to count the number of times a lower seed beat a higher seed. However, that counts #9 Pittsburgh crushing #8 Colorado the exact same as #14 Mercer shocking #3 Duke. On the more complex end, it is possible to use advanced statistics to calculate the likelihoods of each team winning the tournament, and then compare the model to what actually occurred.
I will settle for something in between -- a method simple enough that it's easy to follow and grasp intuitively, but slightly more complex than the simple "counting upsets" idea. In this post, I will define how "surprising" a tournament is as the sum of all seeds of the winners of all 63 games.
For example, this year's Final Four and Championship game saw wins by #7 Connecticut, #8 Kentucky, and then another win by #7 Connecticut. So, the three games in Arlington add 7 + 8 + 7 = 22 "surprise points" to this year's total. Do the same thing for the remaining 60 games, and we get a grand total of 345. This method, while easy enough to follow, has the advantages of a sliding scale by seed (a 15-over-2 contributes more than an 11-over-6) and a sliding scale by round (a 14-seed winning two games increases the total more than a 14-seed winning just once).
Another way to think about it is that the average seed of the winning team in any game this year was 345 / 63 = 5.48. The higher the average seed of the winner, the more surprising a tournament is, since in general we expect lower-numbered seeds to win games. Note that this data excludes all play-in games.
That score of 345 for the tournament that just concluded is the highest score of all time, going back to the tournament's expansion to 64 teams in 1985. Furthermore, all of the top three scores came within the last four years:
|Rank||Year||Surprise Points||Champion||Final Four Seeds|
|T7||2001||319||#1 Michigan State||1,1,2,3|
This year's tournament was very solid for upsets from start to finish -- three #12s and a #14 all advanced; a pair of #11s made it to the Sweet 16 and Elite 8; a #10 made it to the Sweet 16; and then the 7-8-7 championship weekend really put 2014 over the top.
2011 might be among the most famous tournament for upsets, thanks to its Final Four featuring #8 Butler and #11 VCU, but the last two years had slightly more upsets top-to-bottom by this measure.
For what it's worth, though it's a small sample size, UConn and Louisville seem to thrive on bracket chaos.
On the flip side, here are the bottom ten, meaning the least surprising years in tournament history:
Click here to see the full article on the Harvard Sports Analysis Collective.