Bracketology Starts

[adamw]

If BC wins out, Minnesota is out, we get the CHA autobid and St. Cloud gets Penn State. But if St. Cloud finishes first, they get the CHA autobid and we get Penn State.

Atlantic Hockey - not CHA ... FYI. But your point stands.

[Swampy]

In a way, given how Cornell has played most games against teams that are highly regarded (H/D weekend, all 4 Q games, BU, 2nd K game) versus teams that might be regarded as meh (2nd Brown game, RPI, 1st Miami game), I'm beginning to think I'd prefer Cornell to play, say, Minnesota rather than a team our guys might subconsciously think is a cupcake, like one from AH. Remember Bemidji State.

[BearLover]

In a way, given how Cornell has played most games against teams that are highly regarded (H/D weekend, all 4 Q games, BU, 2nd K game) versus teams that might be regarded as meh (2nd Brown game, RPI, 1st Miami game), I'm beginning to think I'd prefer Cornell to play, say, Minnesota rather than a team our guys might subconsciously think is a cupcake, like one from AH. Remember Bemidji State.

Strongly disagree with this--i think you're cherrypicking. We also have games versus weak teams where we blew them out (first StL game, second Miami game, etc.), and you're not taking into account how those games played out (we won nailbiters versus H, BU, got thrashed by Clarkson the first time, badly outshot Miami in the first game, etc.)

This team will come ready to play for the NCAA tournament regardless of who our opponent is.

[KenP]

15 games left, one can be a tie. That means 2^15 + 1 = 32,769 possible outcomes.   I'd be curious to see a version of the CHN Pairwise Probability Matrix that ignores KRACH and simply shows haw many (or what percentage of) scenarios work out to who is in the tourney at what seeding.

[ursusminor]

15 games left, one can be a tie. That means 2^15 + 1 = 32,769 possible outcomes.   I'd be curious to see a version of the CHN Pairwise Probability Matrix that ignores KRACH and simply shows haw many (or what percentage of) scenarios work out to who is in the tourney at what seeding.

Isn't it 2^14 * 3 = 98,304?

[KenP]

15 games left, one can be a tie. That means 2^15 + 1 = 32,769 possible outcomes.   I'd be curious to see a version of the CHN Pairwise Probability Matrix that ignores KRACH and simply shows haw many (or what percentage of) scenarios work out to who is in the tourney at what seeding.

Isn't it 2^14 * 3 = 98,304?

ummmmmmmmmm... i think it's time for some random emoji.....  ::burnout:: ::crazy:: ::panic::
Actually, it's 49,152.

[ursusminor]

15 games left, one can be a tie. That means 2^15 + 1 = 32,769 possible outcomes.   I'd be curious to see a version of the CHN Pairwise Probability Matrix that ignores KRACH and simply shows haw many (or what percentage of) scenarios work out to who is in the tourney at what seeding.

Isn't it 2^14 * 3 = 98,304?

ummmmmmmmmm... i think it's time for some random emoji.....  ::burnout:: ::crazy:: ::panic::
Actually, it's 49,152.

:-/

[adamw]

15 games left, one can be a tie. That means 2^15 + 1 = 32,769 possible outcomes.   I'd be curious to see a version of the CHN Pairwise Probability Matrix that ignores KRACH and simply shows haw many (or what percentage of) scenarios work out to who is in the tourney at what seeding.

I tried this a bunch of times before giving up on figuring out a good way to code out, and going with the Monte Carlo method.

It's probably better with Monte Carlo anyway, since not all scenarios are equally likely, so it doesn't do much good to treat them as such.

But yes, I would love to be able to run something that goes through all scenarios. It's easier now than in years past, when everyone had semis this weekend and therefore there were over 3 million permutations. Now, with two conferences already to their finals, there's far fewer. So I could brute force it.  But it's still a programming struggle.  Someone with much more experience in this area could probably help me out - but I googled endlessly on it and came up empty.

[KenP]

15 games left, one can be a tie. That means 2^15 + 1 = 32,769 possible outcomes.   I'd be curious to see a version of the CHN Pairwise Probability Matrix that ignores KRACH and simply shows haw many (or what percentage of) scenarios work out to who is in the tourney at what seeding.

I tried this a bunch of times before giving up on figuring out a good way to code out, and going with the Monte Carlo method.

It's probably better with Monte Carlo anyway, since not all scenarios are equally likely, so it doesn't do much good to treat them as such.

But yes, I would love to be able to run something that goes through all scenarios. It's easier now than in years past, when everyone had semis this weekend and therefore there were over 3 million permutations. Now, with two conferences already to their finals, there's far fewer. So I could brute force it.  But it's still a programming struggle.  Someone with much more experience in this area could probably help me out - but I googled endlessly on it and came up empty.

Adam, you've done the hard part already -- you have the engine to run results and tally PWR.  I can create a list of 49,152 numbers where 0 means "first team wins", 1 means "second team wins" and 2 means "tie for NCHC consolation".  Run each one through your PWR engine, cut the brackets down based on lower seed AQs and you're set.  PM me if you want to discuss further.  It'll take me about 15 minutes to create values in a spreadsheet.

[jy3]

Cornell gets the #1 overall IF:

• Cornell wins out AND
• St. Cloud is held to one point or less (semi-final loss AND loss or tie in the consolation game)

The CHN Monte Carlo simulation shows a non-zero chance of Cornell at the #5 PWR.  Denver needs to beat St. Cloud in the finals AND Ohio State needs to lose AND all RPI factors work to bump Ohio State up to #4 despite a loss:

• ECAC Semifinal #1: Princeton def. Cornell
• ECAC Semifinal #2: Harvard def. Clarkson
• ECAC Championship: Harvard def. Princeton
• Hockey East Semifinal #1: Boston College def. Boston University
• Hockey East Semifinal #2: Northeastern def. Providence
• Hockey East Championship: Northeastern def. Boston College
• Atlantic Hockey Semifinal #1: Robert Morris def. Mercyhurst
• Atlantic Hockey Semifinal #2: Air Force def. Canisius
• Atlantic Hockey Championship: Robert Morris def. Air Force
• WCHA Championship: Michigan Tech def. Northern Michigan
• Big Ten Championship: Notre Dame def. Ohio State
• NCHC Semifinal #1: St. Cloud State def. North Dakota
• NCHC Semifinal #2: Denver def. Minnesota-Duluth
• NCHC Championship: Denver def. St. Cloud State
• NCHC Consolation: North Dakota ties Minnesota-Duluth

I was playing with this and did not have time to figure out these scenarios. Thank you for posting this. Always good to check out elynah this time of year. LGR!

[Jim Hyla]

This week's movement shows Northern Michigan in, Omaha out

This week's brackets

Midwest Regional (Allentown):
15 Northern Michigan vs. 4 Ohio State
12 Penn State vs. 5 Denver

East Regional (Bridgeport):
13 Minnesota vs. 3 Cornell
9 Providence vs. 6 Minnesota State

West Regional (Sioux Falls):
16 Mercyhurst vs. 1 St. Cloud State
10 Michigan vs. 8 Minnesota Duluth

Northeast Regional (Worcester):
14 Boston College vs. 2 Notre Dame
11 Clarkson vs. 7 Northeastern

Conference breakdowns

Big Ten — 5
NCHC — 3
Hockey East — 3
ECAC Hockey — 2
WCHA — 2
Atlantic Hockey – 1

This week's movement:

Out: Omaha

In: Northern Michigan

[BearLover]

I think it's looking very likely that we get Minnesota at this point.

[BearLover]

BU wins in OT to keep their season alive and end BC's. Harvard is up 2-0 vs Clarkson. If Harvard hangs on, we just need BU to win tomorrow to avoid Minnesota. Otherwise, we almost certainly get Minnesota. (I think.)

EDIT: Never mind. Didn't notice how close Minnesota and Penn State are. I think we are very likely to play one or the other.

[Dafatone]

If Sucks beats Clarkson, I believe North Dakota is out.

[Beeeej]

If Sucks beats Clarkson, I believe North Dakota is out.

Unless other math pushes them back up above #14, yeah, they'd have to be.