KenPom 2014-15

[Taj79]

To be fair, I feel a need to note this from a Ben W. tweet: KenPom has published his 2014-15 preseason statistical rankings and we come in at #81.

The rest: VCU @ #17; Dayton @ #50; Richmond @ #51; George Washington @ #62; Umass @ #74; St. Joe's @ #90; La Salle @ #101; Rhode Island @ #105; the Bonnies @ #107; Duquesne @ #124; Masona @ #138; Davidson @ #139; and Fordham @ #154.

Such a rating would rank us at #6 in the conference, aligning with "our friend" Matt Norlander and his "projection." So we go from a high ranking of six, down to other projections at either 9 or 10, and not a mention of us in the Top 144. Collectively, the body of work seems relatively consistent. The good thing is Pomeroy keeps updating and changing based on actual results. No one poll is perfect. Again, just food for thought and discussion. This is no ore a jumping off point than any of the others are. This one does reenergize after each game as the seaso goes on. The evolving rankings will be more valid.

If anybody can clue me in on how Pomeroy does this first preseason poll, I'd be interested in understanding.

[slu72]

I wonder if Pomeroy rankings take into consideration our relatively easy schedule. The guy who came up with his projection probably never checkied who were going to play. Instead he looked at the lack of stats for our returnees and said, they're gonna suck this year. No noteworthy returnees, no high impact transfers, no reputable FR. Just a guess.

[slu72]

By the guy, I meant Mr 144 in 144 days.

[Zink]

I wonder if Pomeroy rankings take into consideration our relatively easy schedule. The guy who came up with his projection probably never checkied who were going to play. Instead he looked at the lack of stats for our returnees and said, they're gonna suck this year. No noteworthy returnees, no high impact transfers, no reputable FR. Just a guess.

Pomeroy weights his margins based on the opponent's rank in each category, so strength of schedule is automatically factored into the final rating. No clue how he generates this ranking with regard to SOS. I would guess that his initial model is agnostic, as his rankings attempt to control for it anyway. He likely performs some transform of the past few years' ranking that includes a "returning contributions" factor. I would also guess that he weights our freshman based on some sort of "caliber of program/typical recruit" metric.

Our team loses a lot, hence the drop from the top 30, but has been ranked highly for the past few years and is in a good conference, therefore the drop only to 81. Note that Fordham is the only a-10 team outside of his top 150. That would be a precipitous fall.

[Taj79]

I would say Mr. 144 has little in terms of deep analysis, and his "analysis" can be as short in "research" as your description is in words. Now KenPom has something that translates into some sort of "Pyth" stat and that's how he ranks them. Of course, I don't know what that is either right now and would love an education as to how he arrives at that and what it means.

Doing simple math (the only kind I know), our OOC opponets are ranked/numbered 171 (out of 351), 132, 242, 337, 83, 211, 130, 10, 178, 287, 294, 111, and 91 in this preseason ranking. The average team then on our schedule is #175; if you throw out Wichita State (10), the average ranking for the rest becomes 189 (this adds in both Bradley and TCU from the Corpus Christi tournament, not knowing which one we'll get).

One must believe that even with a bunch of unknowns, the only team we will not be "higher" than in this OOC is Wichita State and therefore woudl you classify it as in a "favored" position? I think so. Vermont actually has a higher start (111) than Indiana State does (132). Based on this, what, 12 and 1 going into the conference slate?

Can't wait.

[slu72]

12-1 going into Conf play? Hope so, and I'll gladly eat my crow dinner. I know that's just numbers, and you don't foresee that happening. There will be surprise L's as in, are you kidding me? My guess is more like 9-4.

[kshoe]

As a subscriber to Pomeroy, I can tell you that his model predicts us to go 9-3 in non-con and 18-12 overall. He doesn't count the Rockhurst game. I'd have to go back and look but I think that lines up exactly with my prediction in the prediction thread.

When thinking about his predicted records its important to remember that if we play 10 games and are expected to win each of them 60% of the time, he'd expect a 6-4 record. Others might look at it and say we are expected to win every game so the prediction should be 10-0.

[kwyjibo]

Pomeroy's method is based on several factors: the efficiency stats of the last 5 years of the program, the efficiency stats of returning players, recruiting rankings of freshman in the top 100.

For the record, Pomeroy is forecasting that SLU is and will be a better offensive team than last year but worse defensively. So,, in this method he probably overrates some returnees who were efficient in limited minutes, but underrates the potential of Ash and any freshman outside of the top 100 (but still going to contribute). Basically, though SLU is 81 because they have done pretty well in the last 5 years.

It is also true that Pomeroy uses these "preseason ranking" as a seed for his rankings in the early season (so beating Chicago St. by 40 does not make any team #1) but it IS HIS best guess at how good the teams are (and "will be" for that matter).

[Cowboy]

Pomeroy's method is based on several factors: the efficiency stats of the last 5 years of the program, the efficiency stats of returning players, recruiting rankings of freshman in the top 100.

For the record, Pomeroy is forecasting that SLU is and will be a better offensive team than last year but worse defensively. So,, in this method he probably overrates some returnees who were efficient in limited minutes, but underrates the potential of Ash and any freshman outside of the top 100 (but still going to contribute). Basically, though SLU is 81 because they have done pretty well in the last 5 years.

It is also true that Pomeroy uses these "preseason ranking" as a seed for his rankings in the early season (so beating Chicago St. by 40 does not make any team #1) but it IS HIS best guess at how good the teams are (and "will be" for that matter).

-I think I read an interview with KenPom or perhaps it was in a blog of his where his preseason estimates are eliminated from the calculation by conf play

[kwyjibo]

-I think I read an interview with KenPom or perhaps it was in a blog of his where his preseason estimates are eliminated from the calculation by conf play

Yes. That is what I meant by seed in the early season. His preseason ratings factor in less and less as the season goes on and is at zero by mid to late January (technically after conference play has started but roughly conference is correct). I say this only to oppose it to other people like Sagarin who removes the "preseason seed" factor about a month earlier (often before confernce play starts).