Honestly Dick, it seems that you are grasping at straws here in order to
cast some doubt on KRACH.
 
  Ken's passage you relayed about jumpy results early in the season is just
another way of saying that the ratings can change significantly early in
the season due to the addition of relatively few games.  The same can be
said for RPI/PWR and it is merely due to sample size (i.e. adding a weekend
of results in October/November may be doubling the current year's data for
that team whereas adding the same results in March is only increasing the
number of scores for that team by 3-6%).
 
  Perhaps you should direct some of your critical analysis at the RPI
itself.  It does not take a Rhodes Scholar to look at the algorithms for
RPI, HEAL (which you mentioned earlier), and KRACH and see that the RPI is
rather arbitrary in its' construction and weightings, HEAL is overly
influenced by Opponents' Winning % yet treats all losses equally regardless
of opponent, and that KRACH is a pretty sound way of ranking teams when
only the w-l-t outcome of games can be considered.
 
HOCKEY-L is for discussion of college ice hockey;  send information to
[log in to unmask], The College Hockey Information List.