The appellation 'March Madness' didn't just come from nowhere. It's called that for a reason. A 68-team single elimination tournament over three weeks is bound to spit out some surprises. Even the 2008 tournament, when all four number one seeds made the Final Four, was considered historic. All of this is just to say that any exact outcome is exceedingly unlikely, which is precisely why Warren Buffett's billion dollars were so safe.
Nonetheless, it feels like this year there has been a lot of crowing about how many upsets we have had (e.g. “we almost had 4 12-seeds in the field of 32,” “Dayton versus Stanford in the Regional Semis,” “a 7-seed and an 8-seed in the Final Four!”)
But actually, the results thus far have not been terribly chaotic. In fact, this year's tournament has been a little less “mad” than expected. We analyzed the actual results this year by running 100,000 simulations of the tournament and calculating the likelihood that any one of those results would actually play out, (that likelihood being the joint probability of the simulated winner of each game actually winning). We then compared the actual result odds to these simulations, and found that the actual results from this year rank in the 45th percentile, meaning that 55 percent of the time we would have expected a more chaotic outcome. So, despite the excitement over upsets, this tournament has actually been slightly less “mad” than average.
By comparison, last year's tournament, perhaps made most notable by Florida Gulf Coast's run to the Sweet 16 as a 15-seed (2.5% chance of happening according to our model) and Wichita State's unexpected run to the Final Four as a 9-seed (1.2% chance of happening), was in the 74th percentile through the Final Four.
What accounts for this difference? Well, compared to last year, this year's tournament has been characterized by several series of minor upsets from teams like Kentucky, UConn, and Dayton, as opposed to major shockers a la Florida Gulf Coast. What do we mean by minor upsets? Consider that in getting to the Final Four, neither UConn nor Kentucky has been more than a 5.5-point underdog in any game. Mercer over Duke was in fact the only major upset of this tournament (for our purposes here, meaning the only win by a team that was more than an 8.5-point underdog). Dayton over Syracuse and Stanford over Kansas were pretty significant, but still, both of those teams had about a 30% chance of winning those games, according to our calculations.
If we narrow our focus to the teams still standing, this tournament has been a little more chaotic than expected, with only one favorite making the Final Four. That team is Florida, which had a 38% chance of being here, and was helped considerably by a weak region, injuries, and upsets. The other three representatives are no pushovers, Wisconsin (14.7% chance of making the Final Four) was the third-most likely representative from the West, Connecticut (5.2%) the sixth-most likely from the East, and Kentucky (4.8%) the fifth-most likely from the Midwest. This combination ranked in the 74th percentile compared to our simulated outcomes, right in line with last year's bunch.
OK, so if the Madness hasn't been that mad, then your next question is likely, what the $%^& happened to my bracket? We said from the beginning that this was a tough year to predict, with no team having a greater than 40% chance of making the Final Four. The average bracket has about one team left, and that one team is probably Florida. This is about what we expected when the tourney started. Other favorites have fallen by the wayside, most notably Arizona and Louisville. But we expected these teams to have near toss-up matchups against the likes of Wisconsin and Kentucky. And when this happened, the favorites came out on the wrong end. Only Florida has had a relatively easy road so far. And come next weekend, their toughest matchup could come against the last team that gave them a real test, Kentucky.