Before the season starts in earnes, and we have no time for
such frivolities, I thought I'd describe a couple of ideas on which
I've been ruminating for a while.  They both involve changes to the
current NC$$ tournament structure.  One is fairly serious, the other a
bit unrealistic.  Both reflect my bias towards the importance of
conference matchups and conference playoffs.
 
        First, the more serious one.  I'd like to see two new
selection criteria (to join the present list of Ratings Percentage
Index, head-to-head, common opponents, last 20 and record vs. teams
with winning records), to be used when the two teams being compared
are in the same conference.  One is regular season conference record,
and the other conference playoff finish.  So, for example in last
year's ECAC, the ranking would be 1) Cornell (winner of conf tourney);
2) Harvard (loser of ECAC championship game; 3) Vermont (winner of
consolation game); 4) Clarkson (loser of consolation game); 5-8) RPI,
St. Lawrence, Brown and Colgate (losers of quarterfinal series); 9-10)
Princeton and Dartmouth (losers of preliminary round games); 11-12)
Union and Yale (didn't make the playoffs).  Teams eliminated in the
same round would be counted equally for this criterion--neither one
would get the advantage.
        A few notes about the impact of these:
        a) In the case of upsets in the conference playoffs, adding
the two criteria would cancel out.  For example, Cornell wins one of
the new criteria (playoff finish) over Vermont but Vermont wins the
other one (regular season finish).
        b) The conference record is a bit redundant with the common
opponents and head-to-head criteria, but there's no principle
forbidding that.  As was pointed out last year, RPI (the ratings
method) gets used several times in the current system.  Still, if
there seems to be too much double jeopardy here, the different
criteria could be weighted unequally, or conference record used
*instead* of common opponents for members of the same conference.  (As
discussed below, conference record is more evenhanded than common
opponents.)
        c) Common opponents tries to compare how the teams fared
against the same competition, but doesn't take into consideration how
often each team plays.  As an extreme example, imagine that Lowell is
being compared to DU for common opponents; suppose DU played Merrimack
in the Denver Cup and Lowell faced Colorado College in the Burlington
Coat Factory Challenge[;-)].  That means that those two games, along
with DU's four conference games against CC and Lowell's three against
Merrimack, are counted towards the common opponents.  Now the
criterion which is supposed to compare the overlap between the teams'
schedules is weighing DU's four CC games and one Merrimack game
against Lowell's one CC game and three Merrimack games.  Hardly fair
to DU, if CC and Merrimack have seasons similar to this past one.
Comparing the conference records of teams in the same conference
provides a balanced version of common opponents; both teams play all
of the opponents the same number of times.  (The WCHA is an exception,
but the unbalanced schedule there also means that the final WCHA
standings are not neccessarily fair either.)
 
 
        The more far-fetched idea is a re-structuring of the NC$$
tournament.  The same twelve teams (the four conference [tournament]
champions, four regular season champions the four to eight "best"
other teams, as ranked by pairwise comparisons) would make the
tourney, but there would be two tournaments to determine the Eastern
and Western champions, each of which would be a double-elimination
tournament in which everyone but the conference (tournament) champions
started with one loss.  In detail, this would mean:
 
        On the weekend after the conference tournaments, the four
conference champions would host two games each.  In the "ideal"
scenario, where two teams advance from each conference, these would be
the two "other" qualifying teams from the other conference in the same
region, but if only two teams qualified from a conference, the extra
slot would be filled with the "extra" team from the other conference.
The higher seed of the two visitors would get to play on the second
night (when the host team was more tired).  So, in last year's
tournament, that would mean Cornell (ECAC Champion) hosted Lowell (HE
qualifier #2) one night and BU (HE Q1) the next; Providence (HE
Champion) hosts Clarkson (ECAC Q2) followed by Vermont (ECAC Q1).  Out
west, Minnesota (WCHA champ) would host Western Michigan (CCHA Q2),
then Lake State (CCHA Q1) while Michigan (CCHA champ) would host
Michigan State (CCHA Q3) and Colorado College (WCHA Q1).  Any team
which won a game would advance.  So if the host team swept, they'd be
the only one from their bracket to go on, if they split, they and the
team that beat them would go, and if they lost both games, the two
visitors would advance.  For example, suppose Cornell beats Lowell and
loses to BU, Providence loses to Vermont and Clarkson, Michigan
sweeps, and Minnesota beats WMU but loses to Lake State.  Then
Cornell, BU, Vermont and Clarkson advance in the East and Michigan,
Minnesota and Lake State go on in the West.
 
        The following weekend, each region would hold a regional
tournament at a neutral site to crown a regional champion.  Depending
on the events of the preceding weekend, the regional could feature
two, three, or four teams.
        If there are two teams (i.e., both hosts swept the previous
weekend), they play a best of three series.
        If there are three teams (one host sweeps and the other
doesn't, as in the West in the above example), the undefeated team
again takes on the other two on consecutive nights.  If they sweep,
they are the regional champion; if they do not, the winners of the
first two games play on the third night to decide the championship.
So for example, suppose Michigan beats Lake State, but loses to
Minnesota.  Then they play Minnesota again for the West championship.
        If there are four teams (as in the East above), they play a
standard four-team single-elimination tournament that weekend to
determine the regional champion.  Of course, teams are paired to avoid
rematches and conference matchups.
 
        On the third weekend, the east and west champions face off on
a neutral site to determine a national champion.  This could be either
a single game or a best-of-three.  (I think it's better not to count
the regional champions' losses in the regionals against them.)
 
        Not that I expect the NC$$ to do anything like this, but I
think this has a lot of nice features.  By going to three weeks, it
avoids the conflict with basketball without adding a week off.  A team
can't be knocked out of the competition by a single upset (in effect,
the non-conference-champions have the loss that knocked them out of
their conference playoffs counted against them).  This scheme would
combine the campus sites and neutral sites of prior tourneys.  Teams
would no longer be sent out of their regions, which the NC$$ has in
the past based on things other than performance on the ice.  In fact,
the whole system would have cut-and-dried rules, eliminating the
judgement calls made by the selection committee.  Regionals would
involve local teams, boosting attendence, while still pitting teams
from different conferences.
 
        Little bits of tinkering can still be done, such as declaring
that teams in the same conference be seeded according to regular
season finish rather than pairwise comparisons.  Also, there's some
question as to the best arrangement of the two or three games of the
two regionals; two or three teams would have to play one game a night
for two or three nights, while a four-team tournament would be more
fair with two games one night and one the second.
 
        And of course, there are down sides.  Only two teams in the
finals could bring down attendence.  Fans of east-west matchups would
be dissapointed on the whole (heck, even major league baseball has
added inter-league play).  But like I said, I think it's cute.  And
I'd say it's more reasonable than other alternative schemes which have
been put forth.  [Like the scheme where the RPI Engineers get the top
seed every year. ;-)]
                                        John Whelan, Cornell '91
                                        <[log in to unmask]>
        <http://www.as.ucsb.edu/kcsb/tmss/jphock.html>
 
1996 Cornell Hockey: Ivy League Women's Champions
Ivy League Men's Champions/ECAC Men's Champions
 
HOCKEY-L is for discussion of college ice hockey;  send information to
[log in to unmask], The College Hockey Information List.