I feel like Puxatawney [sic] Phil, only popping his head up on 1 day
out of the year. I have been reading up the last week, so hopefully I
am not too clueless. Given all of the people here that have been
immersed in the process and numbers, it is actually pretty easy to
catch up. (Thanks!)
Two points:
1. Seeding is ALL guesswork this year
2. What about the last 2 at large bids?
1. Seeding is even more up in the air than in the past.
Given this quote, from Lee's article at USCHO:
"There hasn't been any decision yet if they will order the teams by
region or not. We will have to see how the comparisons turn out
before any decision is reached."
and the fact that the conference make up of the 12 teams is VERY
lopsided (5 teams from 1 conference and 2 conferences with only 1
team), we might get some new seeding criteria.
The committee has never really committed to any set way of seeding,
but in general they have gone by things like:
* Keep teams in their natural regions when possible
* Reward higher ranked teams by letting them stay in their natural region
* Avoid first round conference match-ups
* Avoid 2nd round possible conference match-ups
* Reward higher ranked teams with easier paths to the finals
(including byes) and/or try to balance the strength of each bracket
* Just "go by the numbers", plop teams into slots (with almost no
tweaking based on fuzzy factors like attendance), and let's play
hockey
The problem is that these ALWAYS conflict. So, which one will be the
determining factor for questions already raised here, like does CC or
Minnesota go in the east, does Providence go west simply to avoid BC
and Maine, is there even any effort to have WCHA teams miss each
other (it is easy to put MSU/Michigan and BC/Maine in brackets so
they can only meet in the finals; after that, maybe we do not bother
any more).
I expect to see some brand new seeding criteria this year. Could be a
whole new process.
For example, instead of dealing with teams in groups ("swap 2 eastern
teams into the west"), this year I would probably just start at the
top of the ranking and place each team 1 at a time into a slot that
is best for them.
MSU bye in west
BC bye in east
SCSU bye in west
UND bye in east
4 teams, 1 in each bracket
Michigan in SCSU bracket (stay in west, avoid MSU, SCSU is lower ranked team)
CC in BC bracket (avoid MSU and WCHA foes)
Minnesota in MSU bracket (sorry, only bracket without WCHA team so far)
Maine in UND bracket (stay east, avoid BC)
8 teams, 2 in each bracket
Wisconsin in SCSU bracket (might as well keep in west)
Providence in MSU bracket (again, only non-HEA bracket open)
Probably have SLU next, put in UND bracket maybe
Mercyhurst, put in BC bracket, these last 2 are coin flips
So that would be, with overall rankings included:
#1 MSU - #7 Minn / #10 Providence
#3 SCSU - #5 Michigan / #9 Wisconsin
#2 BC - #6 CC / #11/12 Mercyhurst/SLU
#4 UND - #8 Maine / #11/12 SLU/SLU
(I would not even think in terms of 1W and 4E and such this year:
will probably be totally meaningless, even more than in past years. I
hope the ESPN show does NOT include them and only gives us who plays
who, when and where.)
Looks like the UND bracket is the "easiest", if that means anything.
Overall, western brackets have higher ranked teams, if that means
anything.
Anyway, I'd say seeding is about 80% guesswork this year (20%
predictable), where in the past is might have been as high as 50%
predictable.
2. Nobody here seems to care about *the last 2 at-large bids*. PWR is
based on the 5 criteria, and I am far from convinced these are any
good. The details get hidden when you just declare 1 team the winner
of each comparison, but the devil is in the details. (Thanks to USCHO
for these details, I did not compile them myself this year.)
The most interesting comparisons to look at are with Wisconsin &
Providence (2 lowest ranked at-large teams) and UNH & Clarkson
(highest ranked teams left out).
Wisconsin
* Ties with Clarkson 2-2, tiebreaker goes to Wisconsin on RPI
* UNH only wins TUC
Providence
* Ties with Clarkson 2-2, tiebreaker goes to Providence on RPI
* Loses to UNH on head to head (2 games to 1) and TUC, but wins RPI,
Last 16 and common opponents
Yes, Clarkson could be out only because they lose the RPI tiebreaker *twice*.
(Part of this depends on whether you are looking at Clarkson vs
Mercyhurst to determine of CU is in. Clarkson loses this comparison,
too, but sometimes the committee throws away the teams that are
already in and only looks at the "local" criteria comparisons. John's
script call this IPWR.)
So, I think this is a new record in being close but not selected. I
recall Bowling Green tieing on the criteria and losing on the RPI to
one other team, so they were left out. But Clarkson loses 2 of these
RPI tiebreakers. Ouch, that is close.
I'd have to spend a lot more time looking at the details to find out
how the criteria themselves determined the outcome. Like what
teams(s) did Wisconsin go 3-0-0 against that Clarkson only went 2-1-1
against. And what if it was Last 20 games instead of Last 16. How
close are some of these Last16, TUC, and Common opponent records were
are comparing: maybe there is just the difference between a win and a
tie that is awarding one key criteria to Providence or Wisconsin. How
about different weights for RPI. And of course, RPI is the deciding
factor here, and anyone who ever considered a college degree in math
can tell you how crappy the ratings percentage index is.
So, my conclusion, I guess: the new-fangled seeding will get the
attention this year, it seems, but I hope the college hockey
community does not stop trying to improve the rest of the selection
process. It is "straight-forward", which is not all bad (ie, better
than other sports), but I am STILL hoping for better than "not bad".
Keith
--
Keith Instone - [log in to unmask] - http://keith.instone.org/
|
