>   Now, I've felt all along that the current system in place in college
>hockey to select/seed teams for NC$$ Tournament play is a good one, especially
>now that it's totally out in the open so that industrious fans like John
>Whelan can make the whole process available for infinite "what-if" scenarios
>on his www site (thanks John :-)  I don't expect this process to change any
>time soon (other than invoking minor tweaking such as reducing the "what
>have you done lately" component from 20 games to 16) but this *IS* the NC$$
>after all - who knows :-)
>
>  So, in the spirit of pre-season discussion generation, I'd be interested
>to hear what others think of such a scheme being put in place for college
>hockey.  All the components are already in place:
>
>  Polls: Use the USCHO + USA Today Polls ....
>  Computer Rankings: Use CHODR, HEAL, and PWR ranking ....
>  Strength of Schedule: Already computed using RPICH, which would have to
>     be adjusted to split out OWP and OOWP from each team's individual WP
>     (i.e, the leading term in the current 0.35, 0.50, 0.15 formulation).
>
>  Is there anyone (Charlie ??  John ??  Robin ??) out there with time on their
>hands who's looking for another nifty project ??  Here you go: programmers -
>man your keyboards !!
>
>  Can you tell it's a slow day here in Maryland ?? <grin> ....
>
>  Cheers from the Chesapeake - Jim
>
 
        How would you deal with the losses factor that is the last element
of the BCSR?  In college hockey, where the best teams are expected to have
lost some games, and where the total number of games played can vary
somewhat widely, adding the total losses is ridiculous.  But, using RPI
rankings instead of just schedule strength rankings might work, though that
might be a little too duplicative of a major portion of the PWR.  I like
the diversity of computer rankings chosen, each calculates very
differently, but the rankings are thus assured to violate the deviation
standard on nearly every team outside the top 5, and perhaps some inside it
as well.  Perhaps a more liberal deviation standard should be created?
This would be intriguing if someone put together a reasonably adjusted form
of this for hockey.  Then we can draw conclusions about how much better the
PWR system really is (for those who still complain that it isn't a terribly
good system.)
 
Larry Weintraub
 
HOCKEY-L is for discussion of college ice hockey;  send information to
[log in to unmask], The College Hockey Information List.