This is an early-bird special, just for HOCKEY-L. I'd appreciate comments and/or
questions (email me, don't ask the list) from you all. I'll summarize the
 responses
in mid-December when the first official TCHCR is released.
 
For the many of you who have been following and supporting TCHCR, here is what
 has
changed from last year. I simplified the game outcome measure as much as
 possible,
so only winning and losing count--goal margin, site, OT, etc are ignored. I
 sacrificed
some accuracy for simplicity. The other major change is that the basis team has
 been
eliminated. This year, the weighted sum of the ratings is zero (and a BIG thanks
 to
Tim Danehy for doing the math for me to make this possible!). I hope no one
 suffers
from a fear of negative numbers (nega-numa-phobia?), because TCHCR has them for
 the first
time.
 
For those of you who have never heard of TCHCR before, it is the longest-running
mathematical ranking of Division I college hockey teams (yes, a grand-daddy at
 age 4).
Computing TCHCR involves collecting all of the Division I results, building a
 big
graph that represents who has played whom, and then a bunch of number crunching
 to
get the ratings. It is not a poll, but a RATING.
 
A disclaimer: I may be changing the algorithm for schedule computation in the
 future.
We (Tim, Mike and I) are studying the "connectedness" of college hockey.
 Depending on who
meets whom in the Christmas tournaments, it may become necessary to reduce or
 alter
the way the schedule aspect of the rating is computed. For example, no ECAC team
 will
play a WCHA team unless Minnesota & St. Lawrence meet in the Mariucci. Tim
 already has a plan
prepared; we just have to decide how much of it to implement, based on how the
 match-ups
fall.
 
Again, comments and questions welcome, as always.
 
Keith [log in to unmask]
(Obviously, this does not include the Tuesday night results.)