Replying to a couple of messages from Eric Carlson in one go:
>>> Actually, I was pretty impressed by the Wayne State team here in
>>> Fairbanks.
[...]
>>> Still you'd be hard pressed to put them
>>> any better than about #45 or #46 among D1 teams this year
>> Considering they were not a D1 team at all this year, I'd say so!
> I meant IF they were compared to the other D1 teams!
Sorry, I should have realized that you for one would know who was
Division I and who was not.
> The point is most of the rating systems all bring something to the table
> worth looking at. I doubt if I'd ever advocate solely using the main CCHP
> ratings--and perhaps not any single system--to pick a tournament field even
> though I'm very partial to it. But CCHP will often show you something that
> the systems based only on wins and losses instead of scores cannot.
This is true, although I agree with the NCAA policy of considering
won-lost-tied outcomes and not margin of victory for several reasons:
first, I'm leery of a situation where a team would be encouraged to
run up the score for tournament selection purposes (just look what
happens in international competitions where net goals are used as a
tiebreaker); second, in hockey some two-goal games are actually
one-goal games with an empty-netter at the end, and that goal really
shouldn't be given the same significance as a gamewinning goal; third,
team sports are as a rule quantized into individual games, and there
is something philosphically appealing about a win being a win no
matter what the score.
> CCHP
> does have a separate rating using only wins and losses instead of scores
> otherwise using the exact calculation method.
Do you post that anywhere? It would be useful to have another system
for comparison that used the same input as RPI (and KRACH, HEAL and
RHEAL). Having skimmed the definition of CCHP, I suspect that the
system would be like KRACH and RPI in ranking teams in the same order
as winning percentage if they've all played balanced schedules.
> Moreover it's a crazy system used by the NCAA when Minnesota coach Don Lucia
> knew it would be better to lose to Wisconsin in the semifinals of the WCHA
> tournament and then defeat St. Cloud in the consolation game than to win and
> then lose to North Dakota in the finals. They would have won a critical
> pairwise comparison by doing that. Try to defend that one!
It should be noted that this sort of problem exists independent of the
way strength of schedule is handled, or not handled, in the present
pairwise comparisons. That situation has the potential to arise as
long as head-to-head competition is one of the selection criteria,
since a team may benefit from the opportunity to win another
head-to-head game against a team with whom they have an important
pairwise comparison.
The same thing can also occur with record vs common opponents: if
Clarkson had won the ECAC play-in game, Rensselaer would have been
better off losing to Colgate in the semis and beating Clarkson in the
consolation game, since it would have given them the pairwise
comparison with Mankato on the strength of common opponents.
With respect to RPI, KRACH, or whatever measure you want to use for
overall record and strength of schedule, losing a semi and winning a
consolation should be worse than winning the semi and losing the title
game, since in principle your opponent in the consolation game should
be weaker, therefore you will end up going 1-1 against weaker
opposition than if you went to the final and lost. (Of course, if the
other semi has already been won by the weaker team, the situation is
reversed, but in that case you would expect to have better odds just
to win the final and get an automatic bid!) Perhaps this is really an
argument against consolation games!
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.
|
