Charlie Shub wrote:
>can anybody explain what factors lead to differences among the
>rankings by the 4 systems given below.
>
>
Not me! I can say, completely unscientifically, ...
1. HEAL and RPI are almost brother and sister ... quite close
algorithm-wise. RPI includes 2nd-order effects, but those are almost
always quite small; HEAL treats a loss to a powerhouse the same as a
loss to a doormat.
2. RHEAL is an experiment of mine to try to makeup for missing 2nd-order
effects and better represent strength of schedule. (RPI & HEAL tend to
over-value good teams of a weak conference and undervalue good teams of
a strong conference). RHEAL succeeds at that, but fails because it has
VERY strong conference leanings. This year, for example, it seems to
think the HEA sucks and the CCHA is head and shoulders above everyone else.
3. KRACH seems to have a consistancy that the other systems lack. Over
the past 4 weeks only a few teams have, from week to week, moved more
than 2 places, and none (that I've noticed) more than 4 or 5 places. I
looked at a few of these, and saw a probable "unexpected" game or games
that might explain the change. However, the other rankings tend to have
large adjustments now and again ... much more than KRACH in terms of
ranking. For example, I can't explain the HEAL/RHEAL treatment of
Miami, but I'm sure Miami fans like the it just fine (now 5th in
RHEAL/6th in HEAL, in teens of other rankings). Miami did win 2 last
week, but against barely-over-500 Notre Dame.
4. PWR tracks very closely with RPI. It basicly *IS* RPI, with a
bazillion silly rules that mostly cancel themselves out.
5. I'd include CHODR here, but my reading of it is that it is built to
explain and predict scores, not won-lost records. Maybe that sounds the
same, but I think it quite different. Maybe Robin can comment.
6. I haven't looked at the "polls". The TV still reports Polls, so they
must exist.
7. Even Dave Hendrickson can beat CHODR/RPI/HEAL/RHEAL at predicting
game outcomes, and so do many Hockey-Lers most years at the end of
regular season contests. Sorry, Dave ;-) How will/did KRACH? Perhaps
someone else can say, but ...
10. My opinion:
If season rankings are meant to be the best possible representation of
season performance, then KRACH wins the rankings battle, hands down. If
the purpose of selection for the NCAA Tournament is to include the 16
teams each with the best chance of winning the National Championship,
I'm sure I DON'T know how to do it! On second thought, I don't want
that latter ranking. Ranking by season performance is just fine with me.
Can you answer your question, Charlie?!?
cheers, wayne
>>Various rankings for Men's Division I College Hockey, through 2004-02-08.
>>Compiled by Wayne T. Smith
>>
>>
>>
>> School KRACH RHEAL HEAL RPI
>> ---------------------------------------------------
>> 4 Maine 4 11 3 3
>> 6 St. Cloud State 6 7 14 11
>> 9 Denver 9 12 15 12
>> 11 Colorado College 11 14 20 17
>> 12 New Hampshire 12 17 7 8
>> 13 Miami 13 5 6 13
>> 14 Ohio State 14 9 12 9
>> 15 Providence 15 25 21 14
>> 16 Michigan State 16 10 16 15
>> 17 Brown 17 23 13 16
>> 18 Alaska-Anchorage 18 22 29 21
>> 19 Notre Dame 19 13 19 23
>> 20 Colgate 20 21 11 18
>> 27 Ferris State 27 15 22 27
>>
>>
|