Comments on: Analyzing Advanced Metrics: Does Florida Have What It Takes to be a Championship-Level Team?

By: greyCat

greyCat — Wed, 28 Nov 2012 19:23:28 +0000

There is a reason why Sports Illustrated’s Luke Wynn holds off on his “frauds and overlooked” column using the same metrics to compare with the AP Top 10 list until late December. Pomeroy’s system needs at least 4 weeks to “connect” the 330+ Division I teams. On December 5 of last season Pomeroy’s top 10 (yes, I have collected these irregularly going back to 2005) were:
1. Ohio State
2. Kentucky
3. Wisconsin
4. Syracuse
5. North Carolina
6. Marquette
7. Duke
8. Missouri
9. Florida
10. Kansas

Note that both teams in the Championship game were “top 10” that week and a third Final Four team was ranked #1.

On November 26, 2010 Pomeroy had none of the Final Four teams in his Top 10.
On November 29, 2009 Pomeroy had West Virginia and Duke ranked (#2 and #3 respectively), but not Michigan State or Butler.
On November 29, 2008 Pomeroy had North Carolina and Connecticut ranked (#6 and #5), but not Michigan State or Villanova.
On November 28, 2007 Pomeroy had UCLA and Kansas ranked (#6 and #7), but not Memphis or North Carolina.
On December 10, 2006 Pomeroy had Ohio State and Florida ranked (#7 and #1), but not UCLA or Georgetown.
On December 23, 2005 Pomeroy had none of the Final Four teams in his Top 10.

By: Bjoyce

Bjoyce — Wed, 28 Nov 2012 17:59:11 +0000

Great points… Will be interesting to revisit this again later in the year. And agreed… Archives would be great to check off at different points in the year.

By: rtmsf

rtmsf — Wed, 28 Nov 2012 17:34:30 +0000

In reply to Matt B. This is a really good point, I really wish KenPom would provide at least a weekly archive b/c I'm sure he has it. At a minimum, preseason and post-regular season ratings would be good to see.

By: Matt B

Matt B — Wed, 28 Nov 2012 16:59:42 +0000

There are a few other things to note about KenPom’s numbers right now:

1) KenPom’s offensive, defensive, and pythag number are still influenced by his preseason projections. For example, Duke, Kentucky, and Wisconsin were all narrowly eliminated from title contention based on defense. Depending on what Pomeroy’s projections were for those teams, they might already be in the to 20 based on raw data alone.

2) This one is more important, KenPom continues to calculate data during the postseason, so while UConn and Syracuse appear to have been top ten teams, that was only after their title run. Given that most champions will play 4 or.5 highly ranked teams in the tournament, the title run itself can have a big impact on pythag and adjusted offense and defense. KenPom does not archive midseason numbers on his site, so I can’t say how far back they came from, but when it’s time to fill out brackets, don’t eliminate teams who miss these metrics by small to moderately small margins, they can (and have) made up the difference during the tourney.