Keep those Computer Ratings questions coming! I am happy to explain
whatever you want to know about my system.
Keith
==========================================================================
>From: Erik Biever <[log in to unmask]>
>Subject: Instone ratings
>
>It is interesting to note that no WCHA team is ranked above #9 in the
>latest Instone ratings. Is this a direct result of Maine sweeping
>Northern Michigan last week? As the WCHA expanded to nine teams this
>sesason, the opportunities for inter-conference play are reduced as
>each team now has 32 conference games. As we get into December, though,
>there should be enough inter-conference action to give the ratings a
>better base.
Absolutely right on all counts, Erik.
The WCHA as a whole is rated low in the Computer Ratings because of
their poor showing against other conference teams, including NMU's defeats
at Maine.
vs other vs Total
3 lgs. Indeps non-conf
CCHA 4-4 6-0 10-4
ECAC 11-11 0-1-2 11-12-2
HE 21-14 13-1-2 34-15-2
WCHA *0-7* 6-1-1 6-8-1
If that 0-7 isn't bad enough, the WCHA has been outscored 10 goals
to 45 goals in those 7 losses, never scoring more than twice in any
game.
So, even though Minnesota itself did not lose any of those 7 games,
their opponents Denver, Colorado College and Northern Michigan did,
and that makes the Gophers schedule an easy one.
I am sure the WCHA will come around and win their share of non-conference
games. But with only 15 scheduled so far, there haven't been enough to
offset Denver's 0-4 non-league lambasting.
------------------------------------------------------------------------
>From: "Douglas J. DeAngelis" <[log in to unmask]>
>Subject: The infamous "Instone Ratings"
>
>I find myself putting much more stock in this rating system than in
>the innumerable polls which seem far too influenced by "gut feelings" and
>hopes for the future rather than the simple facts of how teams have
>done (although I believe both methods serve a function...). I have a
>couple of questions, however (please pardon if this has been hashed about
>in the list before and I missed it.):
(There are 6 different Div. I polls. According to my math theory,
this isn't innumerable. ;^)
>1) Is there a concept of the "relative worth" of a team other than just the
> ranking? AND
I am not sure what you mean by "relative worth" but a team is rated
solely on its performance on the ice this season. A win over Lake Superior
is only better than a win over Kent State because LSSU has played better
than Kent this year.
I don't seed the teams based on pre-season predictions, recruits, last
year's record or personal preferences. I assume that Jeff Sagarin (USA Today)
does this somehow for basketball because UNLV was rated 5th or so and they
hadn't even played a game yet.
>2) If so, does this mean that a highly ranked team A playing a lowly ranked
> team B and having beaten them twice earlier in the year, could then
> throw the third meeting and RISE in the rankings as a result due to
> the greater value thus attributed to team B?
With this system, you don't want to throw a game. If you lose to a team,
your rating will tend to drop below the team you lost to. The worse you
lose, the farther below. In all circumstances, you will be rated higher
if you win than if you lose. The only possible benefit to throwing a game
is that your schedule strength may rise a bit, but since your schedule
strength is calculated AFTER your rating, why bother?
>The reason I ask this is that it used to be the property of a similar "points"
>system used for high school basketball in Maine that this could occur.
Sounds like Ohio's HS football 'computer' ratings also.
>-Doug
|
