> So, if we can't have both a strong offense and defense, it appears that
> a strong offense is slightly more likely to give a good winning pct.
>
> BTW, the (rank) correlation between CHODR offensive and defensive ratings
> is r=0.66, probably mostly a reflection of overall stronger talent vs.
> weaker talent.
 
These kind of studies are somewhat mathematically flawed in one distinct way
(as years of reading Bill James will tell me).
 
Good offensive teams win games by wider margins  i.e.  8-4, 6-3, etc...
Defensive teams win games 3-2, 2-0, 4-2.
 
The latter is no less effective, but in doing analysis based on total goals, it
skewers things.  It's possible to correct for this.  I'm not sure if Robin did.
 If so, I apologize.
 
Just wanted to put that on the table.
 
Adam Wodon - AC Productions
Host: "Around the Rinks" - The only National College Hockey talk show
Listen 24 hours/day, 7 days/week at:
http://www.audionet.com/sports/shows/rinks/
 
HOCKEY-L is for discussion of college ice hockey;  send information to
[log in to unmask], The College Hockey Information List.