ESPN hasn't started beating us over the head with this debate yet (give it a week), so I thought I'd get a head start, if from a slightly different perspective.
Pomeroy currently has Duke as 3.3 points better, but I think that's a tough argument to make. Here are the spreads on each of the Blue Devils' ACC games, along with what the current Pomeroy rating predicts the final score to be:
In each of their eight conference games, including Wednesday's visit to Clemson, Pomeroy's current ratings have Duke covering the spread (in reality they are 4-3 ATS thus far). The wagering market clearly thinks KenPom is overvaluing them, by about 2.8 points per game. For the purposes of this exercise, I'm going to weight the more recent games more heavily (11/10/9/8 on down the line); for Duke, this makes their average Vegas spread -12.2, and their average Pomeroy spread -15.0. So either way, we get to a difference of 2.8 points.
Here's the same chart for UNC:
Here we see a less dramatic difference, and in the opposite direction. The Wake line was way "off", but you have to remember that it was three weeks ago, and Wake's reputation has changed dramatically since then (in no small part because they beat UNC). Since that loss, the lines have been very close to the predictions, with only the Florida St. game having more than a one point discrepancy.
Unweighted, Pomeroy looks to be underrating UNC by 1.6 points. But that decreases with the weighting, since the Wake game was so long ago. Weighted, the average spread is -15.6, and Pomeroy's average margin is -14.4. So Pomeroy is underrating the Tar Heels by an average of 1.2 points.
That means we should close the gap between the two teams by four points, which actually puts UNC ahead by 0.7. The line for next Wednesday's showdown in Durham will shed some light on the situation; my guess as of now would be Duke -3.5.
Photo: Daylife.
Original Story:
http://vegaswatch.net/2009/02/duke-or...
