Pomeroy: How it works and how MU jumped from 49th to 36th in a week

In addition to the more important developments, yesterday was a reminder of why it is important to root for Marquette's opponents. I admit that after the University of Arizona's web page noted that I had rated Nick Johnson as the 14th best player in the NCAA (read the story on the University site here), I was happy to see Nick Johnson light up Florida with 4 steals, 3 assists and 3 of 4 shooting from beyond the arc to hand the Gators their first loss.

However, in general good performances by MU's past opponents (Butler and USC) yesterday are a big reason MU shot up to 36th in today's ratings at www.kenpom.com.

Those performances were part of a great day for Marquette basketball.  Most importantly, despite my past bias against a "Catholic League," MU helped lay out a clear vision that could result in a great conference. During MU's win the most important development was Steve Taylor forcing his way to the basket against one of the best shooting defenses in the country for 10 points, while grabbing 6 rebounds and hitting all four free throws after looking shaky from the line earlier.

When MU moved to 36th place, a discussion did start up on MUScoop regarding how MU's Pomeroy rating had fluctuated in between games, so I wanted to lay out an oversimplified how Pomeroy (and the other services) measure impacts and why every time a past MU opponent plays a game it may impact Marquette.

First let's look at the most important factor in Marquette's rating, the score of each game:

MU opponent	Their opponent yesterday	Favored	Won by
Marquette	Savannah State	13	20
	Marquette 7 points better than expected		7
	Weighted as 1/9th of games played		0.78

Marquette was favored by 13 points yesterday and won by 20.  Basically the system says, "Marquette's average performance over the first 8 games was a '50,' and if they are a '50' then they will beat Savannah State by 13 points.  Wait, they won by 20, so they have a '57' for this game and when I average this weekend's performance with the eight '50s' from the previous games then the new average is '50.78', so I need to move that up.  Now actually Pomeroy is much more sophisticated (and better) than this approach because he uses the Pythagorean Theorem first applied to baseball by Bill James which realizes that a 50-40 point win is actually more impressive than a 80-70 win, but let's keep the basic math here.

However, Marquette's rating is based on the rating of every team they have played, and since their ratings change every game that also impacts the '50' rating on the past games.  Yesterday was a very good day for MU opponents overall, as four played:

MU opponent	Their opponent yesterday	Favored	Won by
Florida	Arizona	4	-1
USC	UC Riverside	15	34
Missisippi State	Loyola Chicago	-4	-8
Butler	Indiana	-15	2
	4 opponents 27 pts better than expected	0	27
	27 pts divided by 8 previous opponents		3.78

Kevin O'Neill's defense was dominant, as it often is, and USC held UC Riverside to 11 of 58 shooting to win by 34 points.  That is 19 points better than Pomeroy thought USC was before the game.  Potential future Big East member Butler was expected to lose by 15 points to Tom Crean's Indiana so they were 17 points better than he thought.  Florida and Mississippi State underperformed by 5 and 4 points, respectively, but overall MU's opponents were 27 points better than Pomeroy thought coming in, and even when you take the 4 previous opponents that did not play, the average of MU's opponents yesterday was 3.78 better than their Pomeroy ratings coming into their games.

You take MU being 0.78 better than expected and their opponents' being moved 3.78 better than expected combined to jump MU from 42nd to 36th.

Southeastern LA	at North Texas	-17	-5
	1 opponent 12 pts better
	12 pts divided by 9 opponents		1.33

The only opponent that played today was SELA, which kept it close at North Texas today - only losing by 5 instead of the 17 their ratings would have dictated, which improves the overall opponents ranking by 1.33.

The morale is, always root for all of MU's past opponents and especially pull for Georgetown, South Florida, Seton Hall and Pitt since they will all count twice by the end of the season.