Subject: | |
From: | |
Reply To: | Craig A. McGowan |
Date: | Mon, 27 Jan 1992 19:47:43 EST |
Content-Type: | text/plain |
Parts/Attachments: |
|
|
Keith's analysis of connectivity helps put the NC$$ schedule reductions in
one perspective. I would like to see a different perspective: inter-division
connectivity. To me the numbers showing average connectivity "fuzz" the
issue of ranking top teams in each division relative to each other.
What I am still struggling with is: how much less sure are we of the
relative standings of the top 12 teams this year vs. last year? Let me
illustrate the point I am to make:
Div X Div Y
------- -------
team x1 team y1
team x2 team y2
team x3 team y3
team x4 team y4
team x5 team y5
team x6 team y6
In case #1, each team plays each other team in it's division 2 times, plus
the teams in the other division once. Therefore, it plays 16 games (5x2
intra-division, 6 inter-division). Of these 16 games, 11 are "new
connections" and 5 are "repeats".
In case #2, each team plays the same intra-divisional schedule, but plays
only 1 inter-divisional game. Thus, there are 6 "new connections" and 5
"repeats".
Comparing the two scenarios:
case #1 case#2 %change
------- ------ ------
total new 11 6 ~45%
interdiv new 5 1 ~400%
While this is a crude analysis, one might quickly see that by separating the
out the "inters", we see a large magnitude change. If we don't separate them,
the change looks smaller.
Conclusion: ranking is an art, not a science :-).
--
Craig McGowan
[log in to unmask]
|
|
|