Here are the two important questions, both dealing with the WCHA First
Round Series played by #1 Colorado College against #10 Alaska-Anchorage
and by #4 Minnesota against #7 Minnesota-Duluth.  Both higher seeds
swept their series, but in the RPI, MN moved ahead of CC, presumably
because CC had a weaker opponent.
 
(Note that the NC schedules played by these teams DO NOT MATTER in the
below comparison, because we are only looking at what happens when the
playoff games CC & MN played last weekend are included or excluded.)
 
Q1: What happens when the CC-UAA games are not included, and then they
    are?  (MN games are not included in either.)
A1: The effect of the CC-UAA series on CC was that CC's rating dropped,
    a net change of -0.0004.  MN's rating went up, a net change of
    +0.0001.
    CC's OppWin% dropped by -0.0083 and went from 22nd overall to 33rd.
    MN's OppWin% rose by +0.0005 and stayed 1st overall.
 
Q2: What happens when the MN-UMD games are not included, and then they
    are?  (CC games are not included in either.)
A2: The effect of the MN-UMD series on MN was that MN's rating rose,
    a net change of +0.0053.  CC's rating went up, a net change of
    +0.0004.
    CC's OppWin% rose by +0.0010 and stayed 22nd overall.
    MN's OppWin% rose by +0.0004 and stayed 1st overall.
 
The hard numbers and some thoughts:
 
After 3/9/95
RPI                                             Opp         OppOpp
RK                     G  W  L  T   Win% RK     Win% RK     Win% RK      RPI
5  Colorado College   38 27-10- 1  72.37 4     49.94 21    50.98 10    55.81
6  Minnesota          36 19-12- 5  59.72 10    55.30 1     51.21 5     55.38
 
After 3/13/95 WITH ALL GAMES
5  Minnesota          38 21-12- 5  61.84 10    55.49 1     50.79 12    55.90
6  Colorado College   40 29-10- 1  73.75 4     49.14 32    50.89 7     55.73
 
MN overtakes CC in the RPI despite both teams winning their first
round series, which led to the questions posed above.  But we don't
know what effect other games played between 3/9/95 and 3/13/95 had on
the ratings, so the following is needed.
 
After 3/13/95 WITHOUT BOTH CC & MN SERIES
5  Colorado College   38 27-10- 1  72.37 4     49.88 22    50.84 11    55.74
6  Minnesota          36 19-12- 5  59.72 11    55.40 1     50.95 8     55.37
 
After 3/13/95 WITH MN SERIES, WITHOUT CC SERIES
5  Minnesota          38 21-12- 5  61.84 10    55.44 1     50.86 10    55.90
6  Colorado College   38 27-10- 1  72.37 4     49.98 22    50.81 12    55.78
 
After 3/13/95 WITH CC SERIES, WITHOUT MN SERIES
5  Colorado College   40 29-10- 1  73.75 4     49.05 33    50.93 7     55.70
6  Minnesota          36 19-12- 5  59.72 11    55.45 1     50.89 8     55.38
 
One very big question remaining is what effect the unbalanced WCHA
schedule has on all of this.  In the regular season, CC played UAA 4x
and MN played them 2x, and CC played UMD 2x to MN's 4x.  This would
seem to magnify the effect of MN's having played UMD upon MN as
compared to CC, and likewise for the effect of CC playing UAA upon CC
as compared to MN.
 
It was quite ironic that MN's opponent (UMD) had only played CC 2x as
opposed to playing MN 4x, and that CC's opponent (UAA) had only played
MN 2x as opposed to playing CC 4x.  I would bet that helped make things
worse.
 
If the schedule does have an effect, should the NC$$ be required to
adjust the RPI to account for the WCHA's decision to play an unbalanced
schedule?  I say no.
---                                                                   ---
Mike Machnik                                            [log in to unmask]
Cabletron Systems, Inc.                                    *HMM* 11/13/93