+ (* Steve the Greek **) <[log in to unmask]> writes:
+
+          COMPUTER RANKINGS FOR THE WEEK OF NOVEMBER 7TH, 1993
+
+ KEY: POS(Position in rankings); %(Win percentage-tie is 1/2 win); GF(Goals
+ For); GA(Goals Against); CONF POS PTS(Standing in conference-first place is
+ 12 points, second is 11, and so on); OVERALL POS PTS(Like CONF POS
+ PTS, but using overall record); TOTAL POINTS( % + GF - GA + CONF_POS_PTS +
+ OVERALL_POS_PTS + CONF WINS*2)
 
Hmm... is this "yet another" ratings system?  By looking at the
formula, it looks as though it will be very heavily weighted by games
within the conference.  Combining this with the fact that 90% of the
games will be within the conference, it might not be an accurate way
to compare teams that are not in the same conference.
 
A rankings system that only really compares within the conference
might not be very useful since you only need compare conference
records to align the teams in such a way.
 
Also, I think the various conferences play a different number of games
among their own teams.  Your system would automatically give a higher
rating to the teams playing more games within their own conference due
to the "CONF WINS * 2" element, throwing off any connectivity between
the various conferences even further.
 
Well... I thought I'd throw this back to the list to get other's
impressions as well.  [I hope it won't lead into a "dead horse" thread
regarding computer ratings...]
 
adam
BU '89