Posted: Tuesday March 15, 2005 6:46PM; Updated: Tuesday March 15, 2005 7:03PM
First-round upsets: None.
Second-round upset: No. 5 Villanova over No. 4 Florida. I know the Gators aren't the same softies of recent vintage, but they're still not as tough as these Wildcats.
Sweet 16 upset: No. 3 Kansas over No. 2 UConn. The Huskies may have won a title just last year, but their experience pales in comparison to the wise old Jayhawks stars.
General thoughts: You can't argue with North Carolina's talent, which is why I'm going with the Tar Heels. As difficult as meeting Kansas would be for Roy Williams, I don't think his players will feel the same way.
Grant Wahl will periodically answer questions from SI.com users in his mailbag.
First-round upsets: No. 9 Mississippi State over No. 8 Stanford. This is the same Stanford team that needed a home win against Washington State two weeks ago and didn't get it.
No. 12 Old Dominion over No. 5 Michigan State. This will be a tight one, but my gut says Alex Loughton is going to come up big while MSU's hard-luck seniors won't -- again.
No. 11 UTEP over No. 6 Utah. The Utes haven't had much challenge in the Mountain West, and I'd argue the top tier of the WAC is better than the top of the MWC.
No. 10 Iowa over No. 7 Cincinnati. With or without Pierre Pierce, the Hawkeyes have shown they can do big things. The Bearcats haven't.
Sweet 16 upsets: No. 4 Syracuse over No. 1 Duke. Toughest game to call in the entire bracket. I went one way, then the other, then finally decided the Devils were too thin -- and the Orange would finally find some weapons besides Hakim Warrick and Gerry McNamara.
No. 3 Oklahoma over No. 2 Kentucky. Not a lot of difference here, but I like the Sooners' size and skill inside.
Regional final upset: No. 4 Syracuse over No. 3 Oklahoma: I see the zone causing problems for OU.
General thoughts: Syracuse seems primed for another deep tourney run, and this bracket isn't nearly as hard as the others.
FINAL FOUR: Illinois, North Carolina, Wake Forest, Syracuse.
CHAMPION: Illinois. The Illini have been far and away the best team in college basketball this season. I don't expect that to change when they play in front of their adoring fans in St. Louis on April 4.
The office pool I'm in at work is an "upset-incentive" contest, meaning it goes out of its way to reward entrants who pick bracket surprises. There are any number of scoring systems available -- some more complicated than others -- but the one we use awards a point total for each victory that is a product of the winning team's seed times the round in which the victory occurred.
In other words, if you correctly pick an eight-seed to upset a one-seed in the second round, you'll get 16 points. Correctly pick the one-seed to beat the eight-seed, and you'll get two points.
If your pool is formatted in a similar way, you might want to check out Poologic.com, a free Website that has one goal: to use expert data and high-level mathematics to maximize potential score in an upset-incentive bracket.
Or, put in simpler terms, to help you finish at, or near, the top of your office pool.
Edward H. Kaplan is the Yale School of Management professor who co-developed the algorithm on which most of Poologic.com is based. Using his formula, Kaplan finished 28th out of more than 90,000 entries in the 2000 upset-incentive bracket contest run by CBS Sportsline.
"If you're just trying to pick as many winners as possible, it's not that easy to beat the seeding committee," Kaplan says. "But if you get eight times as much credit for picking an eight-seed to win as a one-seed, that's different. How do you figure out the tradeoff in your head? That's where the computer really helps."
Poologic is hosted by Tom Adams, a systems analyst and hoops fan in Chapel Hill, N.C., who constructed the site in 2000 after examining the research of Kaplan and Brad Carlin, a biostatistics professor at the University of Minnesota. "Carlin figured out how to maximize the score for any kind of upset-incentive pool," says Adams. "But he did it with a Monte Carlo simulation, which took too many hours on a computer, and even then it was only an estimation. The [Kaplan] algorithm Poologic uses only takes a fraction of a second."
Adams' Poologic calculator will spit out its bracket picks for you if you feed it the scoring system of your office pool. Not only does ours do "seed-times-round," but it also awards 15 bonus points for each correct Final Four team, 20 points for each NCAA finalist and 25 points for picking the national champ.
That means I'll first select 1-2-3-4-5-6 on the six columns of the "Seed" row (i.e., the second row) on the calculator page. Then I'll select 0-0-0-15-20-25 on the six columns of the "Standard" row (i.e., the first row). The next step is to choose one of four data models, including Sagarin, Vegas odds and Futures (which Adams explains on the site).
Once you're done, hit "Calculate Now" and voila: a filled bracket appears.
Like the bracket itself, what the computer has just done is deceptively complicated.
"A tournament bracket has two to the power of 63 possible tournament outcomes," says Kaplan. "That's a huge number: nine billion billion. A nine with 18 zeroes after it. If there's six billion people on the planet, you could tell every single person to put in one billion distinctly different sets of picks each, and exactly one person would get every game right.
"This algorithm is the mathematical equivalent of looking at all nine billion billion possibilities and picking one which will give you the biggest average number of victories."
One caveat, though: Use Poologic.com at your own risk. It may improve your chances in an upset-incentive pool, but it's no guarantee of victory. In fact, in a one-off situation it may not help you much at all.
Take my example. For educational purposes only, I entered five brackets in our office pool last year: one that included my own picks and four that used each of the respective data models from the Poologic site. Not only did my own picks finish in the highest position out of the five entries (24th out of 110), but my Poologic picks finished dead last (110th out of 110), nearly dead last (108th), 89th and 67th -- all in the lower half.
Who knows what went wrong? Maybe I entered our scoring scheme incorrectly. Maybe the Final Four bonus points in our pool canceled out the upset incentives. Or maybe it was just a bad year for the algorithm.
But at least now I can explain to everyone at work why I finished dead last in our 2004 office pool -- and respond by doing what anyone in my situation would do.