Hockey-L seems quite dead this time of year, so hopefully no one will
mind that I am forwarding the following article which Jayson Moy posted
on the RPISPORT list in early April with Jayson's permission.
 
The reason that I am sending this message is that there has been
little or no controversy over who has made the NCAA tourney in the
couple of years that the PWR ratings have been used -- controversy
over the seedings, yes, but not the teams that were taken.  As the
following example shows, it is quite possible that changing the outcome
of one game can change which teams would have the highest ratings.
This is almost always the case and shouldn't be a surprise. The game
looked at is the CCHA tourney game between Miami and MSU.  As shown
by the PWR calculation performed by Tim Brule of USCHO, had MSU not
won this game (and Miami was the higher seed), RPI would have finished
above MSU.  As Jayson states, I too suspect that MSU would still have
been chosen over RPI, but there certainly would have been some
controversy in either case.
 
I am not finding fault in the PWR ratings -- any system would have such
problems, and having a system is better than not having a system.  I am
just saying that it is quite possible for there to be a controversy.
 
Ralph Baer
RPI '68, '70, '74
 
 
>Date:    Tue, 8 Apr 1997 23:00:49 +0000
>From:    Jayson Moy <[log in to unmask]>
>Subject: What if? A different scenario.....
 
>Here is something real interesting for you folks out there to take a
>look at.
 
>Courtesy of US College Hockey Online, here is what the Pairwise
>Rankings would have looked like if:
 
>All things remained the same around the country EXCEPT:
>Miami(OH) defeated Michigan State in the CCHA semifinal, and then
>Michigan defeated Miami(OH) in the CCHA Final.
 
>So here's what the PWR would have looked like:
 
>Rk Team                GP  W- L- T  Win%  Rk     RPI  Rk  PWR
> 1 Michigan            40 33- 3- 4 0.8750  1 | 0.6295  1 | 20
> 2 Clarkson            36 27- 9- 0 0.7500  2 | 0.6002  2 | 19
> 3 North Dakota        40 28-10- 2 0.7250  4 | 0.5882  4 | 18
> 4 Miami               38 27-10- 1 0.7237  5 | 0.5856  5 | 16
> 5 Minnesota           40 27-12- 1 0.6875  7 | 0.5809  6 | 15
> 6 New Hampshire       38 28-10- 0 0.7368  3 | 0.5908  3 | 14
> 7 Maine               32 21-10- 1 0.6719  9 | 0.5639 10 | 14
> 8 Vermont             35 22-10- 3 0.6714 10 | 0.5788  7 | 13
> 9 Boston University   38 24- 8- 6 0.7105  6 | 0.5772  8 | 13
>10 Cornell             33 20- 8- 5 0.6818  8 | 0.5702  9 | 13
>11 Colorado College    41 23-14- 4 0.6098 15 | 0.5599 11 |  9
>12 Denver              39 23-12- 4 0.6410 11 | 0.5543 12 |  9
>13 Rensselaer          36 20-12- 4 0.6111 14 | 0.5384 14 |  8
>14 Michigan State      37 21-12- 4 0.6216 13 | 0.5374 15 |  8
>15 St Cloud            40 23-13- 4 0.6250 12 | 0.5404 13 |  6
>16 Union               32 16-13- 3 0.5469 18 | 0.5232 18 |  5
>17 Princeton           34 18-12- 4 0.5882 16 | 0.5357 16 |  4
>18 Colgate             33 16-14- 3 0.5303 19 | 0.5142 19 |  2
>19 Minnesota-Duluth    38 18-16- 4 0.5263 20 | 0.4995 20 |  2
>20 Lake Superior       38 19-14- 5 0.5658 17 | 0.5253 17 |  1
>21 Bowling Green       37 16-16- 5 0.5000 21 | 0.4941 22 |  1
 
>So according to pure numbers, RPI gets the 12th and final seed in the
>NCAA tourney.  But is it that simple?
 
>Here is some analysis done by the gurus at USCHO:
>--------------------------------------
 
>Note: This does not mean that RPI is in and MSU is out. Remember the
>committee did not use the PWR they used LOCAL comparisons. I saved the
>result in the following files. My guess the committee would look at the
>comparisons of RPI, MSU and SCSU. These work out to be
 
>RPI   2-0
>MSU   1-1
>SCSU  0-2
 
>For this reason RPI would make the tournament. Now if the committee did
>something real strange like put Denver and CC on the bubble we would have.
 
>CC    3-1
>DU    3-1
>MSU   2-2
>RPI   2-2
>SCSU  0-4
 
>Since Rennselaer has the RPI on MSU by .001 the committee may give the nod
>to Rensselaer. If the committee chose to throw out the RPI because it
>was too close (they have done this in the past) they would pick MSU for
>obvious reasons (they beat Michigan twice and are awesome vs TUC). The
>only reason Rensselaer is above MSU is the RPI and its too close to use.
 
>My guess is that the committee would somehow find away to put MSU in the
>tournament ($$). The latter analysis could give them the ground to stand
>on. But by the criteria RPI has MSU by an eyelash.
 
>---------------------------------------
 
>So the one thing we'll never know is would RPI have made the tourney
>if the above scenario happened.  Remember MSU beat Miami in OT in the
>CCHA semifinal.
 
>Of course it's all water under the bridge, but at the same time it's
>sort of interesting to wonder looking at these numbers huh?
 
HOCKEY-L is for discussion of college ice hockey;  send information to
[log in to unmask], The College Hockey Information List.