Dick Tuthill characterizes my previous posts as follows:
 
> Last week he argued for subtraction of all the MAAC league games from their
> PWR calculations.
 
No, I performed that calculation to illustrate a point (anyone who's
still listening has already heard the argument).  Of course I don't
think the NCAA should modify their selection criteria in this way, but
if they want to judge whether the MAAC has reached competitive equity
or instead has an artificially inflated RPI, this would be one way to
do it.
 
> This week he is pushing for the use of the KRACH ranking
> to have some influence.
>
>         Gosh, how convenient. While the PWR had Quinnipiac at 8th (I think I
> recall) going into this weekend, KRACH had QC at 41st. Yeah, that's where
> intuitively they must belong. Let's use the KRACH.
>
> Well, in that case, why not use the HEAL ranking? That ranking had QC at
> 10th. But of course, using the HEAL would not be consistent with the
> unstated agenda.
 
I have no agenda.  I chose KRACH because I know it is designed with
the same concepts in mind as RPI, but in a more rigorous way.  I don't
know anything about HEAL.
 
>         Actually, I am not sure that I am comfortable with any of the computer
> rankings. In particular, there is the issue of the subjective biases built
> into them: KRACH, HEAL, and I forgot to mention CHODR as well. Each starts
> off with a set of historically based team strength influence coefficients
> (or similar) to get them started. Thus, the results they produce are path
> dependent. Yet another catch twenty-two. The new teams don't have a history
> so they are arbitrarily rated low initially by an arbitrary amount.
 
I can't speak for HEAL and CHODR (which differs from the apparent
philosophy of NCAA selection by including margin of victory, BTW), but
that is simply NOT TRUE of KRACH.  The only ingredients are the number
of games played by each team against each other team and each team's
total number of points.
 
>    But let's get back from the forest to see the trees here, folks. All we
> are doing is playing with numbers. The more we do it, the more the counter
> argument (that playing with numbers can be done ad naseum with no meanin
> gful result) is reinforced.
 
[anecdotal comparison involving Quinnipiac, UConn and Mankato deleted]
 
Anecdotal evidence would seem to be even more susceptible to this
problem; here's a semi-anecdotal argument:  no MAAC team has taken a
point (close losses are still losses) from Niagara or Mankato, who are
off the bubble in anyone's book and have a total of 27 points in 39
games against major conference competition, while no team from an
established conference has dropped a point to Army or Air Force, who
have 8 points in 9 games against the MAAC.
 
                                          John Whelan, Cornell '91
                                                  [log in to unmask]
                                     http://www.amurgsval.org/joe/
 
HOCKEY-L is for discussion of college ice hockey;  send information to
[log in to unmask], The College Hockey Information List.