Not long after 14th-seeded UAB defeated 3rd-seeded Iowa State in the South Regional, another No. 14 seed toppled another No. 3 seed — this time, Georgia State over Baylor in the West Regional. It’s the first time that two No. 14 seeds have won in the same NCAA tournament since 1995.
As far as 14-over-3 upsets go, Georgia State’s victory wasn’t anywhere near as unlikely as UAB’s. While the Blazers had a mere 9.2 percent pre-game probability of advancing to the round of 32, the Panthers’ chances of pulling the upset were a more reasonable 24.2 percent. But in many ways, the manner in which Georgia State won makes its upset even more remarkable.
According to FiveThirtyEight contributor Stephen Pettigrew’s research about the likelihood of an upset at any given stage of an NCAA tournament game, only 2 percent of major underdogs5 that trailed by any margin with 2:39 left went on to win the game, and when those teams were down by more than 6 points with that much time on the clock, they only won 0.4 percent of the time.
And Georgia State was down by even more than that — it trailed by 12 points with 2:39 remaining. Of the 262 NCAA tournament games since 2004 in which an underdog trailed by 10 or more points with 2:39 to go, only one (0.38 percent) went on to win the game. … Until Thursday, that is. And that one upset was a No. 4 seed over a No. 1 seed — so this is unprecedented in the past 11 years.
As for the ripple effect of the Panthers’ victory, Xavier’s Sweet 16 probability jumped 11.5 percentage points with Baylor out of the picture, while Ole Miss saw its chances of a spot in the Sweet 16 improve by 8.8 percentage points. And Arizona’s path to the Final Four is now clearer; its odds of an Elite Eight berth increased by 5.2 percentage points, and its odds of making the Final Four improved by 3.2 percentage points.
Stephen Pettigrew contributed to this post in a major way.
CORRECTION (March 19, 4:55 p.m.): An earlier version of this post misstated the last year two No. 14 seeds won in the same NCAA tournament. It was 1995, not 1997. It also misstated the percentage of underdogs that trailed by 10 or more with 2:39 to go and still won. It was 0.38 percent, not 0.0038 percent.
Check out FiveThirtyEight’s March Madness predictions.