Steve G writes...
>My biggest problem with RPI is that it blindly rewards a team for having a
>tough schedule (the 0.50 of opponent's record) regardless of whether they
>actually win any of the games versus tough opponents.  That's a big flaw...
 
But that makes very little difference in the RPI value relative
to a team's actual winning percentage.  I used to feel like
Steve, but I have "seen the light."  Looking at last season's
OppWinPct entering the tournament, the highest among the
42 conference teams was Minnesota's 55.76% (or .5576).  The
lowest was Union's 46.52%, a difference of 9.24%.  Multiply
that by the .5 factor/weight and WIDEST RPI margin that
can be attributed to OppWinPct is .0462.  Most teams will
have an OppWinPct in the .4800-.5200 range.
 
ASIDE: The same argument can be made for OppOppWinPct,
where the distribution of values is much tighter and the weight
applied (formerly .25, now .15) is considerably lower.  Last season,
the high and low among the 42 conference teams was Denver
at 51.72% and Alaska Fairbanks at 48.81%.  All other components
being equal, that would lead to an RPI difference of .007275
given a weight of .25, even less this season.
 
If you're not winning these games, the fact that your OppWinPct
and OppOppWinPct is higher than other teams isn't going to
help you.
 
I do agree with Steve that balanced conference schedules
is HIGHLY DESIRABLE.  The imbalance in the CCHA is about
as far as I would deviate.  (Three games versus every team,
which means the site of the games isn't balanced.)  The
imbalance in the WCHA is something I personally would like
to see changed.  Especially if the conference schedule were
reduced to 30 games or less, thus allowing for more non-
conference action with WCHA clubs.
 
 
John H ([log in to unmask])
 
HOCKEY-L is for discussion of college ice hockey;  send information to
[log in to unmask], The College Hockey Information List.