13 Ohio Bobcats logo. Schools from Ohio are 8-0 so far. Image credit: Wikipedia |
Once again, chalk rules despite the prevalence of upsets in the 2012 NCAA Tournament (the AP Postseason bracket is tied with Pure Chalk in terms of total points, but trails in possible points remaining). It's not that the other systems didn't pick any upsets, it's that they picked the wrong upsets.
Turns out, it's very hard to pick the big upsets. A few systems will get the 8-9 and 7-10 match-ups right. And as we saw a couple days ago, quite a few of the brackets successfully picked NC State, an 11 seed now headed to the Sweet Sixteen.
But only one system (Nolan Power Index) had 13 Ohio beating 4 Michigan, and that system is still dead last. No systems predicted the success of 15 Lehigh, 13 Norfolk State or 12 USF.
Moreover, most of the upsets that the least chalky brackets picked didn't come through. Nolan Power and AP Preseason put 14 Belmont in the Sweet Sixteen. Nolan Power also picked 12 Harvard to upset 5 Vanderbilt and 13 Davidson to oust 4 Louisville while sending Murray State to the Final Four. ESPN/USA Today's Preseason Poll put 12 VCU in the Sweet Sixteen.
Those were the big upset picks of our three least chalky systems, and none of these predictions came true (although ESPN/USA Today's Preseason Poll almost hit paydirt with VCU). Only the AP Preseason poll hasn't suffered for it, so far.
Rank | System | R64 | R32 | S16 | E8 | FF | NCG | Champion | PPR | Total | Pct |
1 | Pure Chalk | 220 | 220 | 0 | 0 | 0 | 0 | Kentucky | 1200 | 440 | 92.4 |
2 | AP Postseason | 220 | 220 | 0 | 0 | 0 | 0 | Kentucky | 1120 | 440 | 92.4 |
3 | AP Preseason | 210 | 220 | 0 | 0 | 0 | 0 | UNC | 1240 | 430 | 86.1 |
3 | ESPN/USA Today Preseason | 230 | 200 | 0 | 0 | 0 | 0 | UNC | 1240 | 430 | 86.1 |
5 | Jeff Sagarin | 230 | 200 | 0 | 0 | 0 | 0 | Kentucky | 1200 | 430 | 86.1 |
5 | Sonny Moore | 230 | 200 | 0 | 0 | 0 | 0 | Ohio St | 1200 | 430 | 86.1 |
5 | Vegas | 210 | 220 | 0 | 0 | 0 | 0 | Kentucky | 1200 | 430 | 86.1 |
8 | ESPN Decision Tree | 210 | 220 | 0 | 0 | 0 | 0 | Kentucky | 1120 | 430 | 86.1 |
9 | ESPN CBPI | 220 | 200 | 0 | 0 | 0 | 0 | Kentucky | 1200 | 420 | 79.0 |
9 | ESPN National Bracket | 220 | 200 | 0 | 0 | 0 | 0 | Kentucky | 1120 | 420 | 79.0 |
11 | ESPN/USA Today Postseason | 220 | 200 | 0 | 0 | 0 | 0 | Kentucky | 1120 | 420 | 79.0 |
11 | Lunardi RPI | 200 | 200 | 0 | 0 | 0 | 0 | UNC | 1240 | 400 | 61.0 |
13 | Pomeroy | 220 | 180 | 0 | 0 | 0 | 0 | Kentucky | 1240 | 400 | 61.0 |
14 | ESPN Simulation Model | 220 | 180 | 0 | 0 | 0 | 0 | Kentucky | 1200 | 400 | 61.0 |
15 | FiveThirtyEight | 220 | 180 | 0 | 0 | 0 | 0 | Kentucky | 1120 | 400 | 61.0 |
16 | ESPN InsideRPI | 200 | 180 | 0 | 0 | 0 | 0 | Syracuse | 1240 | 380 | 41.8 |
17 | LRMC Bayesian | 200 | 180 | 0 | 0 | 0 | 0 | Kentucky | 1200 | 380 | 41.8 |
18 | Nolan Power Index | 190 | 160 | 0 | 0 | 0 | 0 | Kentucky | 1120 | 350 | 19.5 |
*Rank sorted by total points (Total) and possible points remaining (PPR). "Pct" denotes where the brackets rank among the rest in the ESPN Tournament Challenge.
Considering that, last year, the only system to outperform chalk was the ESPN National Bracket, the evidence seems to indicate that bracket entrants shouldn't rely on the advanced systems alone to pick their upsets for them.
No comments:
Post a Comment