Deron wrote, in part..
>Here's an opinion, does anyone else think that everything, power-ratings,
>winning percentage, division finish and head to head competition should ALL
>be taken into effect by the NCAA?  I realize winning % is included in the
>power-ratings, but what about these other factors?  ...
 
Let's ignore my contention (shared by a *few* others) that RPI uses an
incorrect and indefensible formula ...
 
The question becomes how do choose and seed the teams.  The NCAA has
done an admirable job, IMHO, in that opinion and prejudice have nearly
been eliminated from the process!  So let's not beat ourselves to death
by dumping on a process that is very good (unless you have an
alternative that is somehow measurably better).
 
The NCAA has said that league regular season champ and tournament champ
get into the tournament.  It doesn't say where they will be seeded.
 
It uses the RPI (similar to RPICH we see on HOCKEY-L), until the
differences in RPI ratings are "small" (0.01 or less).  The 0.01 is
defensible, but IMHO another value might be used.  RPI includes winning%
as well as a measure of the strength of opponents.  Why should division
finish mean anything to RPI or any ranking?  Maybe because it measures
an ability to play "in a big game" (but this is implied in RPI).  Maybe
because a tournament win shows an improving or peaking team (this is not
part of RPI, but is part of the tie-breaking procedures used when RPI
rankings are close).
 
So that leaves head-to-head.  If the RPI rankings are close, then the
NCAA procedures will use head-to-head as part of the tie-breaking
process.  Should it be used before then?  I don't think so.  If RPI is a
good measure of a team's success relative all other teams, then if
significant differences exist in RPI, how can we call the head-to-head
team better if lower in the rankings?  better against that one team,
but not necessarily better overall (now there's a concept for you!).
 
Of course, I've just muddied the waters.  RPICH, TCHCR, KRACH, YAM2,
CHODR and HEAL are all approaches with some validity but different
characteristics.  The authors of each have given us glimpses into their
characteristics over the past few years.  RPI has seemingly had the most
analysis done on it and has the advantage of being used in several
sports.  I liked TCHCR better than all others, but it was so complicated
people may have dismissed it for that.  I dislike CHODR because it
*seems* to prefer high scoring teams to low scoring teams "that win the
close ones".  CC fans should like HEAL, for it would have placed them
8th last year :-)
 
Should consistency count?  Should a team be "burdened" by a weak
schedule?
 
Right now it appears Maine falls into this latter category.  But the
season isn't over yet.  Both Maine and BU play relatively lowly ranked
teams between now and tournament time, except that Maine and UNH plays
twice.  IMHO, the Maine/UNH series will be huge in all the rankings.
Ignoring other games that will be played, a sweep by UNH would place
Maine and UNH in a virtual tie for (HEAL) 2nd in the East. (I haven't
done the all-important RPI calculation).
 
Yours in college hockey,
 
Wayne Smith
The College Hockey Discussion List administrator
Systems Group - CAPS              BITNET/CREN:  wts@maine
University of Maine System        internet:     [log in to unmask]