The primary issue with Pomeroy's stats is that is wildly overvalues teams that are inconsistent, like Marquette, Kansas State, Texas A&M, Nebraska, Oklahoma and Arizona State. The true rating of those teams is somewhere in the middle, because an extremely good showing in a single game will shoot any team up the charts (heck, look at Illinois). It's a very valuable tool, and at least as valuable as the RPI if not more, though. It is a purely predictive rating, though, and says very little about how good a team's season has "been". In essence, Pomeroy's ratings are a good measure of who the "best" teams are, and the RPI is more of a measure of how much a team has "earned" a tournament spot. There's definitely somethign to be said for actually winning the damn game. I don't want to go all Meph here on this, but there's also something to be said for counting Kentucky's loss to Vandy by 85 as much of a loss as A&M's 5 OT loss to Baylor. As I said I use it as a baseline, and then compare schedule to schedule from there. It does allow some teams to sneak by in my subconscious as much better than they actually are(A&M, though they still should be in), but I like it better as a quick and dirty tool than RPI. Another driving force behind my picks is that the Pac 10 is really really good, and I don't think any other conferences are particularly close. This of course means no Pac 10 teams will make the final 4 and some crap team like Mississippi St. runs to the elite 8, but such is the tourney.