A couple of things to cover here...you'll want to read through till
the end, since that is the best part (analysis of last year's RPI with
& without 2 CC-AF games).
To address a few things:
1) I think Wayne's <[log in to unmask]> points were good ones.
We don't really disagree...when I said that the sole purpose of the
RPI was in selecting teams at the end, I was looking at it from the
NCAA's point of view. I do support people trying to understand it by
looking at it earlier on, so they can provide a better analysis and
come up with ideas to improve it.
2) Sean Pickett <[log in to unmask]> writes:
> CC could stop scheduling Air Force and replace the game(s) with ones
> against stronger opponents to get a better SoS.
Charlie Shub <[log in to unmask]> responds:
>No, Sean, thee only sheet of ice within 75 miles of their campus that
>has stands for more than 250 people is at the Air Force Academy.
>Until the money brokers in town get the new arena built, CC has no
>choice but to play at the academy.
Why? Did CC agree to play USAFA twice each year in return for using
their rink? (maybe...I just hadn't heard about this if so. And, it
also doesn't explain CC playing USAFA before last year.)
3) On my comment about waiting till the end of the year to see what
happens:
I decided to look into this to find out once and for all. Here's what
I did:
* Used last year's scores up through 3/19/95, the last games played
before selection.
* Used the RPI weights of last season (25-50-25; this year they are
35-50-15, thus putting an added weight on Win%).
* Did two runs: one with all games, the other without the two games in
which CC played and beat USAFA. The latter is as if the two games had
been canceled and just not played. The games in question:
11/25/94 Colorado College 4 at Air Force 2 NC
11/26/94 Colorado College 10 at Air Force 1 NC
Remember that the RPI only cares about who wins and who loses a game.
Margin of victory, site, etc. are not factors.
So...
Basic results: when CC's 2 wins over USAFA were added in, CC dropped
and USAFA rose by amounts that were more significant than I expected.
More detail:
>This report includes the 761 games through 03/19/95 between teams eligible
>for the Division 1 tournament. [WITHOUT 2 CC WINS OVER USAFA]
>
>RPI Opp OppOpp
>RK G W L T Win% RK Win% RK Win% RK RPI
>...
>4 Minnesota 41 23-13- 5 62.20 11 55.61 1 50.87 13 56.07
>5 COLORADO COLLEGE 40 28-11- 1 71.25 4 50.45 18 51.29 4 55.86
>6 Michigan State 38 24-11- 3 67.11 7 50.78 15 49.53 31 54.55
>...
>42 Mass Amherst 34 6-27- 1 19.12 43 48.69 36 49.84 22 41.59
>43 AIR FORCE 19 5-13- 1 28.95 39 37.41 44 48.59 43 38.09
>44 Army 11 1-10- 0 9.091 44 37.57 43 47.32 44 32.89
>This report includes the 763 games through 03/19/95 between teams eligible
>for the Division 1 tournament. [WITH 2 CC WINS OVER USAFA]
>
>RPI Opp OppOpp
>RK G W L T Win% RK Win% RK Win% RK RPI
>...
>4 Minnesota 41 23-13- 5 62.20 11 55.79 1 50.86 11 56.16
>5 COLORADO COLLEGE 42 30-11- 1 72.62 4 49.43 28 50.93 9 55.60
>6 Michigan State 38 24-11- 3 67.11 7 50.78 15 49.53 31 54.55
>...
>42 Mass Amherst 34 6-27- 1 19.12 43 48.55 37 49.96 17 41.54
>43 AIR FORCE 21 5-15- 1 26.19 40 40.63 43 48.51 43 38.99
>44 Army 11 1-10- 0 9.091 44 37.18 44 47.83 44 32.82
NOTES (all are when going from 761->763 games, i.e. adding in the 2
games in question):
* CC dropped 0.26 in RPI, USAFA rose 0.90 (in the above representation
in which the final values are multiplied by 100). Neither team's rank
changed, although the distance between them and the surrounding
opponents did change.
* CC's OppWin% dropped 1.02 and went from being ranked 18th to being
ranked 28th. USAFA's OppWin% rose 3.22, but that was only enough to
move it from 44th to 43rd.
* CC's OppOppWin% dropped 1.36 and went from 4th to 9th. USAFA's
OppOppWin% rose 0.08 and stayed 43rd.
* There was no change in RPI rank among the top 20, but below that,
there were several cases where teams shuffled spots. Some teams had
their ratings change by very small amounts (i.e. 0.05).
Thus, I would say that the effects may be diluted as Wayne put it, but
they could still be significant enough to be a factor and cause teams
to change spots...and this is more important now since, I believe, the
.01 threshold (1.00 above) may have been eliminated and straight RPI
rank will be the factor w/o comparison of teams close to each other.
So go ahead and criticize the rating. :-)
--- ---
Mike Machnik [log in to unmask] [log in to unmask]
Cabletron Systems, Inc. *HMM* 11/13/93
***** Unofficial Merrimack Hockey home page located at: *****
***** http://www.tiac.net/users/machnik/MChockey/MChockey.html *****
HOCKEY-L is for discussion of college ice hockey; send information to
[log in to unmask], The College Hockey Information List.
|
