Too early for the PWR? Princeton and Brown say no!

[FlagDUDE08]

Re: Too early for the PWR? Princeton and Brown say no!

Last time checked this is a hockey sight. Yes, no

Yes, but association is the best method of understanding.

[MavRick]

Re: Too early for the PWR? Princeton and Brown say no!

I prefer "tumultuous conclusion" or "precipitous contest-termination score event."

[unh1982]

I prefer "tumultuous conclusion" or "precipitous contest-termination score event."

How about the great work Jim dahl did a few threads back with his pairwise assumptions

[burgie12]

Re: Too early for the PWR? Princeton and Brown say no!

I posted my Final PWR predictions for UND yesterday, here's what I came up with for remaining PWR possibilities:

Team	PWR	Overall
		Possible Final PWR	Likely (>1%) Final PWR
Rensselaer	16	10 to 19	11 to 16

I'm curious as to how you get Rensselaer down to 18th / 19th. The lowest I had been able to put them was 17th.

ETA: Not sure why the formatting got screwed up, but I can't figure out how to fix it.

[JimDahl]

Re: Too early for the PWR? Princeton and Brown say no!

I'm curious as to how you get Rensselaer down to 18th / 19th. The lowest I had been able to put them was 17th.

I should have clarified in my post here (as I did with the original publication of that table) that ties are broken with RPI. Rensselaer can end up in a 3-way tie for 17th, but with the lowest RPI of the three, hence 19 for seeding purposes.

[burgie12]

Re: Too early for the PWR? Princeton and Brown say no!

I should have clarified in my post here (as I did with the original publication of that table) that ties are broken with RPI. Rensselaer can end up in a 3-way tie for 17th, but with the lowest RPI of the three, hence 19 for seeding purposes.

Understood. I had been able to get them in a 2-way tie for 16th, hence 17. I hadn't yet found a way to get them any lower than that, though.

[JimDahl]

Re: Too early for the PWR? Princeton and Brown say no!

Understood. I had been able to get them in a 2-way tie for 16th, hence 17. I hadn't yet found a way to get them any lower than that, though.

Ah, no problem. One is a really obscure branch in which Alaska-Anchorage wins the WCHA tournament.

That's enough to flip the Alaska-Anchorage comparison and Colorado College comparisons (both on the basis of RPI). Set everything else up the same but have Minnesota-Duluth beat Alaska-Anchorage and Rensselaer ends up 14th and in.

There are a couple other similar setups, but all are equally delicate.

Atlantic Hockey

* Semifinal #2: Air Force defeats Holy Cross
* Semifinal #1: Connecticut defeats RIT
* Championship game: Air Force defeats Connecticut

CCHA

* Semifinal #2: Miami defeats Notre Dame
* Semifinal #1: Michigan defeats Western Michigan
* Championship game: Michigan defeats Miami
* Consolation game: Notre Dame defeats Western Michigan

ECAC

* Semifinal #2: Cornell defeats Dartmouth
* Semifinal #1: Yale defeats Colgate
* Championship game: Yale defeats Cornell
* Consolation game: Dartmouth defeats Colgate

Hockey East

* Semifinal #2: New Hampshire defeats Merrimack
* Semifinal #1: Northeastern defeats Boston College
* Championship game: New Hampshire defeats Northeastern

WCHA

* Play-in #2: Minnesota-Duluth defeats Bemidji State
* Play-in #1: Alaska-Anchorage defeats Colorado College
* Semifinal #2: Minnesota-Duluth defeats Denver
* Semifinal #1: Alaska-Anchorage defeats North Dakota
* Championship game: Alaska-Anchorage defeats Minnesota-Duluth

[Ralph Baer]

Re: Too early for the PWR? Princeton and Brown say no!

Ah, no problem. One is a really obscure branch in which Alaska-Anchorage wins the WCHA tournament.

That's enough to flip the Alaska-Anchorage comparison and Colorado College comparisons (both on the basis of RPI). Set everything else up the same but have Minnesota-Duluth beat Alaska-Anchorage and Rensselaer ends up 14th and in.

There are a couple other similar setups, but all are equally delicate.

The system certainly isn't stable to small perturbations.

[burgie12]

Re: Too early for the PWR? Princeton and Brown say no!

Having Holy Cross win the AHA title over UConn, instead of AFA over UConn, bumps Rensselaer up to 15th and in the tournament. Now that's close!

[Priceless]

Re: Too early for the PWR? Princeton and Brown say no!

The system certainly isn't stable to small perturbations.

When the sample size is as small as it is, one game can change so much - particularly among COp when the sample size may just be a game or two.

[Craig P.]

Re: Too early for the PWR? Princeton and Brown say no!

The system certainly isn't stable to small perturbations.

It's a combination of a hard off/on on TUC (the TUCliff (tm)) and the small sample sizes in COP. I'm not sure that anything should necessarily be done about the latter (although I'm not thrilled with seeing, e.g., 1 - 0 beat 1 - 1, maybe a floor of, say, five games could help there), but the former is crying out for some sort of switching function to ease teams across the line between TUC and not TUC as far as impact on PWR is concerned.

[redwings8831]

Re: Too early for the PWR? Princeton and Brown say no!

I posted my Final PWR predictions for UND yesterday, here's what I came up with for remaining PWR possibilities:

Team	PWR	Overall
		Possible Final PWR	Likely (>1%) Final PWR
Michigan	4	2 to 11	3 to 11

It would definitely be interesting to hear if anyone has found anything outside those ranges.

How did you get Michigan above 4? I couldn't get them over 4 even with a CCHA championship.

[slurpees]

Re: Too early for the PWR? Princeton and Brown say no!

How did you get Michigan above 4? I couldn't get them over 4 even with a CCHA championship.

And how did you get BU to 13, I can't get them above 14.

[burgie12]

Re: Too early for the PWR? Princeton and Brown say no!

How did you get Michigan above 4? I couldn't get them over 4 even with a CCHA championship.

I got Michigan to 2 with this:

Atlantic Hockey
Semifinal #2: Air Force defeats Holy Cross
Semifinal #1: RIT defeats Connecticut
Championship game: RIT defeats Air Force

CCHA
Semifinal #2: Notre Dame defeats Miami
Semifinal #1: Michigan defeats Western Michigan
Championship game: Michigan defeats Notre Dame
Consolation game: Miami defeats Western Michigan

ECAC
Semifinal #2: Dartmouth defeats Cornell
Semifinal #1: Colgate defeats Yale
Championship game: Colgate defeats Dartmouth
Consolation game: Cornell defeats Yale

Hockey East
Semifinal #2: New Hampshire defeats Merrimack
Semifinal #1: Boston College defeats Northeastern
Championship game: New Hampshire defeats Boston College

WCHA
Play-in #2: Bemidji State defeats Minnesota-Duluth
Play-in #1: Alaska-Anchorage defeats Colorado College
Semifinal #2: Denver defeats Bemidji State
Semifinal #1: North Dakota defeats Alaska-Anchorage
Championship game: North Dakota defeats Denver

(Yale loses two games to have their RPI drop below Michigan's and flip the Yale-Michigan / BC comparisons, BC loses to UNH and Michigan beats Notre Dame to flip the Common Opponents and TUC (Notre Dame gets 2 wins, BC gets 1 loss from their tournaments) to flip the Michigan / BC comparison)

[Ralph Baer]

Re: Too early for the PWR? Princeton and Brown say no!

It's a combination of a hard off/on on TUC (the TUCliff (tm)) and the small sample sizes in COP. I'm not sure that anything should necessarily be done about the latter (although I'm not thrilled with seeing, e.g., 1 - 0 beat 1 - 1, maybe a floor of, say, five games could help there), but the former is crying out for some sort of switching function to ease teams across the line between TUC and not TUC as far as impact on PWR is concerned.

Yes, I understand that. The topic of the cliff has come up many times. Somewhere I recall someone writing that switching from the top 25 being TUCs to everyone with .5 RPI would cure that. That made little sense since there still would be a cliff. With the total number of teams being TUCs likely to be more, (an expected 29 instead of 25), the effect of one entering or exiting would be a bit less -- but not much.