> Of course the argument is not statistically perfect --- because with two
> more weak teams to feed on -- the higher teams get more wins than they
> would otherwise have.  ...
 
And with a perfect RPI formula (the NCAA formula is flawed), those wins
are balanced by the low "opponents" and "opponents-opponents"
contributions.  I'm unconvinced that having weak sisters in a
conference helps a team RPI ranking. Likewise, having everyone in a
conference very good probably isn't measured well.
 
Also, as Erik suggested, RPI may be a poor method for measuring poor
teams. A poor team rating will be more influenced by their competition
than than the rating of a very good or good team. RPI correctly
discounts all games played by poor teams, so that the occasional upset
of a very good team by a poor team means little in the RPI for the
very good team, but a lot for the poor team.
 
cheers,
wayne
 
Wayne Smith -- [log in to unmask] -- Old Town Landing
 
HOCKEY-L is for discussion of college ice hockey;  send information to
[log in to unmask], The College Hockey Information List.