Quotation Bill says:
>The criteria the NCAA uses are:  head-to-head competition between teams,
>record among common opponents, strength of schedule, and record against
>other teams "under consideration" for NCAA tournament berths.  Let's see how
>Cornell and St. Lawrence stacked up last year:
 
And don't forget the most important criteria: record/winning percentage! A very
minor criteria is "eligibility of players," which means if your star goalie is
hurt, the NCAA may not select you.
 
>3.   Strength of schedule
>     I'm not sure how the NCAA determined this, but Keith Instone did some
>     work for them last year in compiling stats and working up some kind of
>     power rating based on winning percentages for the selection committee.
 
The NCAA's strength of schedule is based on the winning percentage of your
opponents (let's call it OWP). It is ALSO based on the OWP's of your opponents.
Thus, the stats go 2 levels deep to consider the opponents that your opponents
played.
 
>     As a ballpark estimate, I looked at the schedule strength ratings from
>     The College Hockey Computer Rating for the week before the NCAAs last
 
That may be a good estimate, but please remember that the NCAA DOES NOT use
TCHCR to select the teams. The strength of schedule on TCHCR is based on the
ratings, so if you don't buy into the rating method, then the schedule stuff
is worthless to you.
 
>     year.  (which reminds me, we haven't seen a TCHCR for this season, have
>     we?  Or did I miss it?  More polls!  More polls!)  Anyway, Cornell had
 
It's coming; have patience. I am waiting til January for the first *official*
release. If I came out with a rating now, I would be falling into the same
trap as the polls: evaluating the teams on very little information.
 
>     different.  I wish to hell they would release the numbers they use when
>     they decide on tournament seedings!
 
So do I. Call your local NCAA selection committee member...
 
Keith