BRACKET STRATEGY #9: Seed Match-ups

Risk: High

Results: 2008 ? 30th percentile in the ESPN Tourney Challenge; 2009 ? 77th percentile; 2010 ? 72nd percentile; 2011 ? 39th percentile. OVERALL AVERAGE ? 54.5 percentile.

Strategy: I did this strategy five years ago and it picked third-seeded Oklahoma State to win it all. They reached the Final Four, but fell short of the prize. In 2008, at member requests, I resurrected the model?and it flamed out. Over the last two years, the results have been decent. I go matchup-by-matchup, only taking the lower seed when the guidance yields better than even odds?and a higher success rate than the high-seed winning rate. In situations where there were countervailing factors?in other words, conditions were ripe for a Cinderella but not for a victim?I advanced the higher seed.

Outlook: In 2008, I felt good about the bracket this strategy produced. In 2009, I was very skeptical about it, because it advanced North Carolina (good) and Michigan State (weird) to the finals. Lo and behold, that?s exactly what happened. Just goes to show how reliable emotions are. Of course, this model also had the Spartans upsetting the Tar Heels?and that?s why it settled back to the 77th percentile. In 2010, this model got West Virginia and Duke right, but liked the Mountaineers over the Blue Devils. Last year, the model was one year late on Duke--and suffered like most stats-based models.