I just finished reading Jayson Moy's article
http://www.uscollegehockey.com/news/2000/02/01_argument.html on the
continuing dissatisfaction over the ECAC's decision to rank its final
standings by winning percentage.  It seems to me like a lot of people
are over-reacting.  Given that
 
1) The only difference in teams' schedule will be the presence of
absence of one game against Vermont, and
 
2) Vermont was a middle-of-the-pack team before they folded,
 
there will actually be very little imbalance in strength of schedule,
certainly less than in the WCHA and CCHA.
 
The ECAC's solution, while somewhat unexpected, was really the fairest
of the simple methods.  In the end, ranking the teams in the final
standings comes down to comparing pairs of teams.  There are three
different possible situations: 1) neither team has played UVM, 2) both
teams have played UVM and 3) one teams has played UVM and the other
has not.
 
        In the first case, any conceivable method will give the same
results; the team with the better won-lost-tied record is better.
 
        In the second case, the question is whether to compare records
including the teams' respective results against Vermont or not.  Since
the teams thought at the time that their Vermont games would count,
and obtained their results in good faith, it is reasonable to insist
that they get credit for their results, and thus compare their records
against their entire schedules.  (The one subtlety would be whether to
account for the fact that some teams got to play UVM at home, and
others on the road.)
 
        This brings us to the third situation, which is the trickiest
one.  One might argue that since the team not playing Vermont never
had an opportunity to obtain a result, the league should ignore the
other team's game as well.  But considering UVM games when comparing
teams that both played them and ignoring them otherwise would get very
complicated, and once again there is the counterargument that the team
that actually played UVM should get credit for their result.  So in
effect the league needs to project a would-be result for the unplayed
game against Vermont and compare the two teams after a 21-game
schedule (real in once case, hypothetical in the other).  Forfeiting
Vermont's unplayed games would have been a logical matter of protocol,
but would have effectively assigned two points to those unplayed
games, which seems as unfair as assigning them no points (the result
if point totals had been used).  Nothing should be considered certain,
in either direction.  Going by winning percentage projects that teams
not playing UVM would have gotten the same proportion of the points in
the unplayed games as the rest of the league.  Again, since UVM was
3-2-2, that's not a bad projection.  Also, since it's only one game,
all that matters is whether the projection is more or less than one
point (having argued it must be less than two and more than zero).
The only other straightforward option would be to project one point
for each unplayed game (assuming effectively that anyone would have
tied the extra game), which hurts teams with winning records and
benefits those with losing records.  This method would be the same as
ranking the standings by wins-minus-losses.
 
So, if you want to give teams credit for all the games they've played,
the only simple options are winning percentage or wins-minus-losses,
and the former gives a better picture of what a team would have done
with that unplayed game.
 
Kevin Sneddon complains that the league didn't provide the coaches
with possible scenarios.  Well, there are not that many scenarios
possible.  If a team plays one more game and finishes with two more
points, they will finish higher in the standings.  (They effectively
played an extra game and won it.)  If they play one more and finish
with the same number of points, they will finish lower.  (They played
one more game and lost it.)  If a teams finishes with one more point
in one more game than another, they effectively played an extra game
and tied it.  If the teams in question have winning records, that was
a worse-than-average result and the team with the extra game finishes
below.  If they have losing records, the opposite is true.
 
People complain about being edged out of playoff spots by percentage
points, which I think bothers them because they're not used to seeing
things that way.  In a full 22-game season, each point in the
standings corresponds to .023 worth of winning percentage.  The
closest two teams can end up in winning percentage this year is if
they are each one game above (or below) .500 and play different
numbers of games.  A team with 21 points in 20 games has a winning
percentage of .525, while one with 22 points in 42 games has one of
.524, for a difference of .001 (actually around .0012).  So that is
indeed a considerably smaller difference, but consider that a
one-point difference in a 70-game NHL season amounts to only a
difference of .007 in winning percentage, and I don't see anyone
complaining that that is too close to call.  It is actually possible
(although phenomenally unlikely) for this 21/20 vs 22/21 situation to
arise in the battle for the regular season title and associated NCAA
bid, but a more reasonable question is how close that could be in
winning percentage if one of the teams with an extra game ended up
with an extra win.  The answer is if RPI ends up with 33 points and a
.786 winning percentage, and SLU or Colgate with 31 points and a .775
percentage, in which case the difference will be .011, or about half a
point's worth in an ordinary season.
 
At any rate, I'm amazed that anyone would talk about the league
re-opening this issue once the decision has already been made.  It was
unfortunate enough that the decision had to be made in the middle of
the season, when decisions could be influenced by which teams had
beaten UVM and which hadn't, but to try and revise the decisions late
in the season, when it's clear who benefits from what details, not to
mention to rescind a decision already made, would really reflect
poorly on the league.  (I'd like to see the league drop the Final
Five, but I'd be opposed to their making that decision in the middle
of the season.)
                                          John Whelan, Cornell '91
                                                 [log in to unmask]
                                     http://www.amurgsval.org/joe/
 
It's playoff possibilities time!
        http://www.slack.net/~whelan/cgi-bin/tbrw.cgi?ecac.cgi
 
HOCKEY-L is for discussion of college ice hockey;  send information to
[log in to unmask], The College Hockey Information List.