Undefeated Texas Tech blew away a top-10 team by more than 30 points. Undefeated Alabama squeaked by a top-20 team in overtime. And yet Alabama remains in first place in both human polls and BCS standings. Granted, Alabama played on the road at a bitter rival. Yet, even if we spot Tech 15 points, they would have still looked better. The computers have Tech a fairly clear #1. (4 of 6 have them first, 2 have them second. For Alabama it's 2 first, 3 second, and one third.) The computers are not permitted to look at score differentials. So, for all they care, Tech could have won by 1 point, and 'bama could have blown LSU away by 40. It looks like the only rationale for keeping Alabama first is "they were first last week, and didn't lose". Great job rankings.
It's also interesting to look at the difference between human polls and computers. The humans love Florida, but the computers are pretty that Texas is 3rd. (Interesting - the computers don't realize that UT's only loss was a last minute squeaker to the now number 1 team - while Florida's loss was to a currently unranked team.) Utah is also not getting much love, showing up as 8 and 7 in the human polls, though 4th in the computers. (Sure they did seem to get a little lucky against TCU, but a win is a win. Penn State on the other hand seems to be overranked by the humans - Is the loss really that bad of a loss? On the other hand, Oklahoma State seemed to fall too much. Sure they were mauled - but it was one of the top ranked teams, not a lowly Iowa.
USC is a couple notches higher by the humans. This seems somewhat justified. They have been doing some serious mauling of some of their opponents - something that would not show up in the computers. (Though the most recent game against Cal did not look that great.) The computers also seem to like Georgia and the ACC much better than the humans. Perhaps they should allow some score-differential algorithms in, to balance out the rankings. (Even better yet - some that penalize for home blowouts of IAA schools.)
No comments:
Post a Comment