John wrote, in part..
>         Actually, I checked last year's PWR ratings, and I found a
> discrepancy win whether they're even a Team Under Consideration
> (defined as a team with a winning record in Division I games).  Army's
> record in all DI games last year was 11-8-1, over .500 and exactly the
> minimum number of games (20) to be eligible for the NC$$s.  However,
> six of their wins came against DI teams (Canisius, Holy Cross,
> Fairfield (x2), Villanoca and Iona) which didn't meet the 20-game
> limit.  So their record against "major" DI teams was only 5-8-1, which
> was the record listed in the PWR rundown
> <http://www.uscollegehockey.com/articles/PWH2H>.  But if the NC$$ is
> counting games against "minor" DI teams towards the 20-game limit,
> surely they must also count those games when calculating Pairwise
> Comparisons.  (If not, consider the disaster ...
 
I haven't tried to analyze PWR, but most ranking systems will go
"bonkers" if some of the teams play just a few "eligible" games.  This
may be a serious problem without solution.
 
cheers,
 
Wayne T. Smith                 mailto:[log in to unmask]
Systems Group -- CAPS          University of Maine System
Co-owner of the College Hockey mailing lists
 
HOCKEY-L is for discussion of college ice hockey;  send information to
[log in to unmask], The College Hockey Information List.