Okay, it's the start of the college hockey season, so now I
want to ask a question about the end of the season. What criteria
will the NCAA selection committee use to determine the seedings for
the national tournament this year? If possible, what will be the
algorithm?
I ask this because every year it seems that the committee uses
a different set of rules, and every year those rules are revealed only
after the seeds are announced. Yet I've heard assertions that the
process was determined in advance, but no one asked. Well, I'm
asking. If we all know what the process is ahead of time, no one will
be surprised by the seedings come March (while they might disagree
with the priorities in the algorithm itself). I'm glad that there is
an algorithm now, I'd just like to know it ahead of time.
Last year in particular, we all thought the Pairwise Rankings
(PWR) of the teams would be used to order the teams as a starting
point. (To calculate the PWR, you compare a team to every other team
with a winning record according to a set of criteria. The PWR is the
number of teams which a team defeats in these Pairwise Comparisons.)
Dan Olsen and I attempted to predict the would-be seedings based on
each week's PWR, using as a guide the reasoning used for seedings the
previous year, namely to avoid intra-conference matchups and increase
attendance by shifting teams from one regional to the other, while
making sure that the 3-6 seeds in each regional ended up in order by
PWR. Then the seedings came out, and bore little resemblance to our
expectations, with the principal surprise being that the seeds were
shuffled within each regional, so that for instance Minnesota, with 16
PWCs won, was seeded below Miami, with 12 PWCs won. Several
Hockey-Lers with media connections posted detailed explanations of the
algorithm this year, and noted that it was completely cut and dry,
according to procedures laid out in the championships manual. Then, a
couple of days later, an interview of committee chair Joe Marsh by
Adam Wodon appeared on US College Hockey Online
<http://www.uscollegehockey.com/tournament/032097.html>. This article
explained the actual process used by the committee, and it differed in
its details from the analysis confidently posted a few days earlier
(after all, with only twelve teams, it's not too difficult to get the
same set of seeds for different reasons). The most important detail
was that the total number of PWCs won was not used to distinguish
teams, only the PWCs among teams competing for a particular seed.
Thus Cornell was seeded below Miami, despite winning one more total
PWC, because Miami won the individual PWC with Cornell. So I think
it's fair to say that no one on Hockey-L (who include numerous media
types, as well as knowledgeable fans) knew the seeding algorithm
before the seeds were announced.
Can someone ask the powers that be what changes are being made
to the process for this season? Here's my understanding of last
year's algorithm for comparison:
Selection:
Winners of conference tournament and regular season titles
gain automatic berths; the remaining 4-8 at large bids are filled by
the Pairwise Comparison process (i.e., compare relevant teams
head-to-head), with the stipulation that each conference must receive
at least two total bids.
Byes:
Any team winning both the regular season and tournament title
in its conference receives an automatic bye. The remaining 0-2 byes
in each region are given to the best teams in in that region according
to PWCs. (According to Adam's article last year, Eastern byes are
given to Eastern teams and vice-versa "unless there is an
overwhelmingly compelling reason do so" but Vermont receiving a bye in
1996 despite being ranked below four Western teams would seem to
indicate that it will never happen.) The two bye teams in each
region are seeding according to a pairwise comparison.
Okay, now we get to the judgment calls; from here on, the
ordering of teams by PWCs may be over-ridden so that the final
pairings produce higher attendance and/or avoid conference matchups.
The order of distaste for conference matchups, from least desirable to
most, is
1. Conference matchups in the first round
2. Second-round conference matchups involving a victory by
the higher-seeded team (1/4 or 2/3)
3. Second-round conference matchups involving an upset (1/5 or 2/6)
If seeds are changed to avoid a matchup, the first round pairings
should be preserved if possible (i.e., if you swap the 3 and 4 seeds,
you should also swap 5 and 6).
Regions:
There will be eight non-by teams at this point, 2-6 in each
region. They are first split into four Eastern and four Western
teams, with the lowest one or two seeds from the more well-represented
region are considered to belong to the other region. [BTW, I think
this is an absurd way to do things, since these teams are about to be
swapped back into their own region while higher-ranked teams are
shipped out.] Then the two lowest-ranked Eastern teams (again,
determined by the individual PWCs) are shifted to the West and vice
versa, unless attendance or conference match-up considerations
indicate that other teams should go, as specified above; in addition,
the host school, if it qualifies for the tournament, is guaranteed to
play in its own region.
Seedings:
Once the four non-bye teams in each regional are determined,
they are seeded relative to one another according to the pairwise
comparisons; again, this may be changed to avoid conference matchups,
as described previously.
Reading Adam's article, I believe those were last year's rules. Can
someone tell us (or ask the appropriate people) what, if anything, is
different this year?
John Whelan, Cornell '91
<[log in to unmask]>
<http://www.cc.utah.edu/~jtw16960/jshock.html>
Cornell Men's Ice Hockey: Back-to-back ECAC and Ivy League Champions
HOCKEY-L is for discussion of college ice hockey; send information to
[log in to unmask], The College Hockey Information List.
|
