Subtracting a team's record from the total record seems like it
would work, but it changes the stat a bit.  Let me explain...
 
Method I:  Two teams are involved.  One has a record of 6-2-0 and
   the other a record of 1-3-0.  The win% are .750 and .250 which
   average to .500
 
Method II:  Again, the same two teams.  Combined record of 7-5-0
   which gives a win% of .636, giving more weight to the team that has
   played more games.
 
With method II, subtracting a team's record would work, but not with
   method I, which I believe to be a better method because
   it doesn't weight teams which have played more games.
 
The reason why so much more programming would be involved is that in
the PHOENIX system, who wins an individual game isn't kept track of.
One array keeps track of the opponents a team has played, and another
array keeps track of wins, losses, ties, GF, and GA.  I use the
information from 9394conf stand (thanx Mike! :-) and discard
non-DivI games as shown in the 9394comp schedule.
 
Once winter break gets here, I may try method II to check out the
differences.
 
As for counting games twice by including them in both Rec and OpRec,
all games for all teams are counted twice in this fashion so
nothing is unfair.
 
Please continue with the comments.  I am more than willing to
improve this system.
 
*******************************************
* Tim Harrison   *                        *
* R3TLH          *      GO TEAM USA!      *
* AkronVM        *                        *
*******************************************