When I fill out a bracket for an NCAA tournament pool (entered for pride only, no $$ exchanged of course), I’m obviously hoping in vain to get every game right. Not gonna happen. I have two other main goals (besides winning the pool): picking the Final Four correctly, and calling the first round upsets. Most of the time the stats favor the favorite (duh), but sometimes they at least indicate that the underdog has a chance. With that in mind, I’m going to go seed by seed and show you which higher seeds have the best chance of getting toppled (at least according to the stats).

I’m picking the games solely based on the trendline score prediction. But I’m also breaking out a new toy that I haven’t used before. For the games listed, I’ve taken the last 15 adjusted game ratings for each team and calculated the probability that a randomly selected game from one team would beat a randomly selected game from the other team. Like I said, I’ve never done this before, so I have no idea if it’s useful or not, but I find it interesting, and that’s the main thing.

8-SEED vs. 9-SEED

Purdue over Arizona — I know, a 9 over an 8 doesn’t really count as an upset, but I’ve got to include it for completeness. Arizona’s offense has been on a downward spiral, as has Purdue’s defense. But … down is good for defense.

• Trendline Prediction … Purdue 70, Arizona 56
• Chance of a Purdue win: 65%

7-SEED vs. 10-SEED

Creighton over Nevada — Nevada is a little upset about their seed, and may be playing with a chip on their shoulder, so I don’t expect this game to play out quite like the stats say, but I still think Creighton is as good or better than Nevada. They played superbly down the stretch in the MVC, and anyone who watched the tournament championship game knows they were watching a very good team. Time for them to get (Nate) Funk-y.

• Trendline Prediction … Creighton 74, Nevada 65
• Chance of a Creighton win: 71%

6-SEED vs. 11-SEED

George Washington over Vanderbilt — You were expecting maybe Winthrop over Notre Dame? Winthrop is the 11-seed getting all the attention, and they deserve it, but Notre Dame is one of the strongest 6 seeds, and not one that I think will get upset. As you can see below, the trendline prediction doesn’t agree with the game-probability prediction. That’s because GW has had a lot of bad games, but their last few were very good. So I checked the other games to see if there was one with a better upset chance. Surprisingly, no - the other chances were 29% for Stanford, 20% for Winthrop, and 20% for VCU. Perhaps this year all the 6-seeds advance? Probably not - even with those low percentages, that works out to a 70% chance that at least one pulls the upset.

• Trendline Prediction … GW 73, Vanderbilt 67
• Chance of a George Washington win: 30%

5-SEED vs. 12-SEED

Arkansas over USC — This isn’t just a case of a lower seed who happens to be hot at the right time, like a few of these are. Arkansas is the better team over the course of the whole year and has been playing better recently. Honorable mention goes to Illinois over Virginia Tech, another 5-12 game between two major conference teams, also with a predicted upset chance of 54%

• Trendline Prediction … Arkansas 78, USC 68
• Chance of an Arkansas win: 54%

4-SEED vs. 13-SEED

Albany over Virginia — Now we’re getting into real upset territory. Albany played UConn close last year as a 16-seed, leading by double digits at one point. Virginia might be the worst ACC regular season co-champ ever, thanks to the new imbalanced schedule. While I don’t expect Albany to win, it wouldn’t surprise me.

• Trendline Prediction … Virginia 60, Albany 59
• Chance of an Albany win: 18%

3-SEED vs. 14-SEED

Wright St. over Pittsburgh — This is more a statement about the quality of Wright St. than the weakness of Pittsburgh (although they have been on a slight downturn recently). Wright St. beat Butler 2 out of 3 times this season, including the last 2 meeting, and made probably the most dramatic turnaround from the beginning to the end of the year as any team. They’re obviously on the rise, and trendlines suggest this will be a toss-up. All other indications say it’ll be a reasonably close victory for Pitt.

• Trendline Prediction … Pittsburgh 61, Wright St. 61
• Chance of a Wright St. win: 32%

2-SEED vs. 15-SEED

Belmont over Georgetown — What are the classic ingredients for an upset? Low scoring game, hot 3-pt shooting by the underdog, and turnovers by the favorite. Well, out of all top-4 seeds, Georgetown plays at at the slowest pace and turns the ball over at the highest rate. Belmont shoots the most 3’s out of any team in the tourney except Butler, and has Andy Wicke, who is 30 of 57 from deep in the last 7 games. I’m not saying it’s going to happen. I am saying I’m going to try and keep my schedule free unless I need to catch the 2nd half of this game.

• Trendline Prediction … Georgetown 76, Belmont 64
• Chance of a Belmont win: 8%

1-SEED vs. 16-SEED

Eastern Kentucky over North Carolina — Just for the sake of completeness. I’m surprised the Kansas-Niagara game didn’t go here, simply because Niagara is far and away the best of the 16-seeds. Doesn’t matter.

• Trendline Prediction … North Carolina 66, Eastern Kentucky 56
• Chance of a Eastern Kentucky win: 1%

And I still think those chance for Eastern Kentucky are a little high.