I sent this to someone in response to some questions he asked me.  It
might be interesting to some of the rest of you.  I have edited it
some to take out some stuff that was not relevant.  I will also ask
Keith or anyone else to correct me if it needs to be done.  And, thank
you to Keith for enlightening me as well as everyone else on some of
the insides of the process that were not as clear.  - mike
 
...
As the committee seeded the teams, MSU came out 12th.  We know this
from the seeds.  How?
 
First, they said the regions were originally:
E BU, Harv, UNH, UML, RPI, NU
W Mich, Minn, LSSU, Wisc, WMU, MSU
 
RPI & MSU stayed at home; UML/NU and Wisc/WMU had to cross over.  They
re-seeded the new regionals according to the original 1-12 ranking and
we got:
 
E BU, Harv, UNH, Wisc, WMU, RPI
W Mich, Minn, UML, LSSU, NU, MSU
 
Note the position of MSU in both sets of seeds compared to the other
teams.  By the first set, MSU was the low seed in the West and NU the
low seed in the East.  When NU crossed over, they were re-seeded above
MSU.  So MSU had to have been 12th.
 
Who was 11th?  NU - they were below RPI, and WMU got put above RPI.
 
10th?  RPI.  They ended up below WMU.
 
9th?  WMU.
 
8th?  Wisconsin.  Wisc was definitely above the four teams I already
mentioned.  Next question is whether they would have been above the
original 4E team, Lowell.  But Lowell was re-seeded in the West above
LSSU, and LSSU was above Wisconsin.  That also makes it easy to
determine...
 
7th? LSSU.
 
6th? Lowell.
 
BTW, I believe 1-5 in order were BU, Mich, Harv, Minn, UNH.
 
Anyway, MSU-WMU wasn't even the real battle.
 
But just to compare, let's see how the two teams compared.  Remember,
the ONLY 5 factors that matter are:
 
* RPI
* H2H
* Last20
* CommOpp
* TUC
 
First, look at RPI.  RPICH closely approximates this.  WMU leads, but
the difference is .0055 - less than the .01 threshold.  So let's look
at the other factors, of which RPI is still one (and a point in WMU's
favor).
 
* RPI: WMU wins, .5366 to .5311.
* H2H: WMU wins, 2-1-0.  (If WMU had swept, it would have been 3
  points in WMU's favor - one for each difference in the series.)
* Last20: WMU wins, .725 (14-5-1) to .650 (12-6-2).
* CommOpp: MSU wins, .652 (19-9-5) to .645 (19-10-2).
* TUC: MSU wins, .548 (10-8-3) to .450 (8-10-2).
 
So WMU wins, 3 factors to 2.  The only factor I had to guess how to
calculate was TUC - I used 21 teams while Keith suggested 18 may have
been used.  Either way, since that was a point in MSU's favor, it
would not have helped MSU if I had used 18 or any other number of
teams.
 
And as I said, WMU was seeded 9th and MSU 12th.  It was really down to
CC vs MSU.  How does MSU compare with CC?
 
* RPI: MSU, .5311 to .5292.  (Less than .01. Look at other factors too.)
* H2H: none.
* Last20: MSU, .650 (12-6-2) to .575 (10-7-3)
* CommOpp: MSU, .722 (6-2-1) to .600 (8-5-2)
* TUC: CC, .568 (11-8-3) to .545 (10-8-3)
 
MSU wins, 3 factors to 1.  That's why MSU got in.  The RPI determined that
the teams were closely related, but since MSU was ahead by 2 factors,
the rating was not the determining factor.  HOWEVER, if CC had an RPI
of .5411 or greater (.01 greater than MSU), then the factors that hurt
CC in a comparison with MSU would not have mattered.  .5411 would have
put CC 10th in RPI (or RPICH).  They could have achieved this rating
by winning more games, or having a better strength of schedule.
---                                                                 ---
Mike Machnik                                          [log in to unmask]
Cabletron Systems, Inc.                                  *HMM* 11/13/93