CAC conference preview/summary:

Here's the output of a new toy I'm playing with that predicts RPI by inputting # of seniors, juniors, overall rating, top 3 rebounders and players >500 and >600.

It doesn't account for SOS and the weightings need work (I think I'll also add average defense rating), but I'll start honing and tracking this. Personally, I think it underrates York and overrates VJ, giving it undeserved credit for its terrible junior class.

Oh well, more later.

COY: andrewfrance

POY: Junior Bennett, CUA

FOY: Don't know if Lewis (SMM) qualifies; I'll go Ross (Gallaudet)

DOY: R. Washington, Goucher

1st team:

PG: Ryder (York)

SG: Campbell (CUA) edging Chen (Marymount); SG much stronger than PG

SF: Barela (Marymount)

PF: Turner (Goucher)

C : Bennett (CUA); several other strong C's

Your first five variables are related. I would suggest putting in variables that are more independent of one another. Instead of # of Jr and Sr's for instance, you could put in avg knowledge of OFF - and assign some score from one to five for this knowledge but with very high weights (since most teams will have a high avg. knowledge). I'd probably drop overall rating since this is affected by extremes such as walk-ons that are not a factor, and is already captured by the of use players above 500 and above 600. The issue becomes that the more categories you add like "rebounding" the more overlap you have with categories "above 500" and "above 600." In otherwords, you are measuring the same characteristic twice-- although I suppose you could make an argument for some type of interaction effect of having players over 500 and 600. Anyway, perhaps more feedback than you want, and with what you've done, you've got a pretty accurate picture of a horrible Villa Julie team.

This may not be possible, but a category of "# of years w/ same coach" might be worth measuringas well.

FRANCE,

Like the idea of your predictor-- maybe you could use end-of-season results and work backwards? Adding D ratings helps, but doesn't consider the variables of zone vs man D, which can be significant. IQs make some sense, too, but how to weigh them...

Wrono

I think # of Seniors/Juniors is a decent proxy for IQ; there will be anamolous teams, but I think would help. I'm not sure m2m v. zone should predict a different RPI. I'll track how my predictions did this year and tinker for next year's preview.