Four rounds into March Madness and the Crowd is continuing to prove a robust predictor of the tournament outcome. Here's the low-down:
Sweet 16
The Crowd picked 6 of 8 correctly (75%), missing only Texas A & M going down to Memphis and Wisconsin's loss to Oregon. The latter of these mis-calls was bittersweet for me. Oregon was my favorite team of the tournament, but their loss sealed my spot as #2 in the office pool.
Elite 8
In a repeat on a smaller scale, the Crowd correctly picked 3 of 4 (75% again). Georgetown over UNC was doubly rewarding: the Crowd picked the only Elite 8 upset and my boss is a UNC alum. Oregon broke my heart a second time, though. The least they could do after busting my brackets was to win the tourney. Instead, they stayed in just long enough to stick it to me. Love-hating those Ducks has never been easier.
Overall
So far, the Crowd has maintained its impressive record: 46 of 60 games. That beats my nearest competition by 7 games, (and my weakest rival by 13). Unfortunately, given my pool's scoring rules those numbers haven't been enough to give me the highest number of points (I'm down by 11 and can only make up 7 of them in the next two rounds). Next year I'll know a little more about properly weighting the point value of later games.
One of the most impressive characteristics about the Crowd's performance is how consistent it has been. It's percentage accuracy in the first four rounds have not deviated by more than 3 percentage points: 78% (Round 1), 75% (Round 2), 75% (Sweet 16), 75% (Elite 8).
I'll provide the next update after the finals. In a subsequent post I'll consider some of the weaknesses of my approach this go around, and examine some ways of improving (ex ante) on the Crowd's performance this year.
Comments