<
>

Mass: Sowing the seeds of upsets

3/18/2008

Which team will be this year's George Mason? How can we tell if the Cinderella slipper will fit?

The most important round in your tournament bracket is the first round. Because there are 32 games, there always will be a few upsets. If you pick the right ones (the ones nobody else does) and avoid selecting against eventual Final Four teams, you've got a huge leg up on the rest of the competition. But basketball is not an exact science, and Hampton, Richmond and Santa Clara didn't exactly leap off the page when the brackets were printed in the years of their upsets. Still, there must be a way to red-flag certain teams that are asking for an unexpected whupping, and those that are champing at the bit to give one. And there is … it's all in the seeds.

Simply put, seeding the NCAA tournament is a very inexact science. Many times, the selection committee severely underestimates or overestimates teams. This is especially true when it needs to wait until the very last minute just to figure out which teams will be in the field, as what happened Sunday. In addition, the committee has a history of favoring teams from the power conferences at the expense of those from the mid-majors and below. That's why it takes the Gonzagas of the world several years of making the tournament before they get assigned anything higher than a 7-seed, even though they probably deserve far higher. A team like Butler, which was nationally ranked as high as No. 8 this season, still was given only a 7-seed. If the name on the front of the jerseys reads D-U-K-E for a team with the same resume, that team probably would be staring at no worse than a 3-seed. But that's OK. It is instances such as those when a 13-seed probably deserved to be an 8-seed, or when a 6-seed looks an awful lot like an 11-seed in which the upsets occur. Identifying these irregularities in seeding is the key to filling out a successful bracket.

I have taken the 65 teams in the NCAA tournament and ranked them according to a statistical formula that takes into account the same categories that the committee used to select the at-large teams in the first place. These categories include: strength of overall schedule, record in the last third of the season, record in games played on the road and at neutral sites, strength of the out-of-conference schedule, and wins against teams in the RPI Top 50. By comparing each team's strength in these areas in relation to the other teams in the tournament, we can assign each school a single value, which I call a team's Atomic Mass. The larger the number, the "heavier" a team is, the more it sinks to the bottom of the list, and the closer to 16 we should expect its seed to be. The following is this year's "expected seed" list.

The teams listed in green are our "money teams": teams I would have seeded at least three spots higher than they were seeded by the committee. Conversely, the teams listed in red are the "teams on alert": teams I would have seeded at least three spots lower than they were by the committee. If any upsets happen, they likely will be in the games that involve these teams. This doesn't mean we're actually going to pick all the money teams to win, nor will we necessarily go with all the teams on alert to lose. Portland State may well be more worthy of a 13-seed than a 16-seed, but we're still not picking them to beat Kansas, for example.

After careful examination of all 32 of the first-round matchups, we'll target just seven matchups for upsets:

East Regional

George Mason over Notre Dame: My head is telling me that George Mason doesn't have an answer for Luke Harangody. My head says Notre Dame actually will win this one going away. However, because I also think Washington State could take out the Irish in Round 2, I'm willing to trust my gut (and the "actual" seeding in my chart) and see if lightning can strike twice for the Patriots.

Midwest Regional

Kent State over UNLV: I'm not really going out on a limb predicting a 9-seed to win, but UNLV is exactly the kind of team that should fall to pieces against the MAC standouts. Kent State versus Kansas could be the best game of the first weekend.

Davidson over Gonzaga: The Davidson bandwagon is probably one that many people will jump on. Last year's team gave Maryland a scare and then some. I hate that these two teams have to face each other, because both have the talent to make a deep run. I just think Stephen Curry will get hot and the Wildcats will win.

South Regional

St. Mary's over Miami: The Hurricanes simply have been rated a few spots too high and draw a team that plays a defensive style that should be able to shut them down. St. Mary's already lost by 19 at Texas this year, so we won't be picking them any further, however.

West Regional

Baylor over Purdue: While Baylor itself is a team on alert, its Atomic Mass is almost identical to that of its opponent, Purdue. Since the Boilermakers, as a 6-seed, have been over-seeded by a far greater amount, we're going to go with the Bears in a close game.

Arizona over West Virginia: Forget all the complaining by the Arizona State fans that they should be in over the Wildcats. Arizona wasn't even close to being the last team on the bubble. This is another game that should play very close, but in the end, we're going with the more battle-tested squad.

Belmont over Duke: This is the one crazy pick in my bracket. Do I really think Duke is going to lose? Probably not. But I like Xavier to beat Duke later on anyway, and Belmont is a tournament-tested squad whose players, if they hit their 3s, could shock the world. Nothing ventured, nothing gained.

So there you have it. My first-round upset specials, courtesy of the committee's rushed seeding process. Only time will tell whether I'm right, or if the madness has caused me to bow out of the Bracket Challenge before the Dance even gets started. Either way, I can't wait to see what happens.

AJ Mass is a fantasy football, baseball and college basketball analyst for ESPN.com. You can e-mail him here.