The 2015 Pairwise, Bracketology and History Thread

[Biddco]

Re: The 2015 Pairwise, Bracketology and History Thread

Would Tech be a one seed if they had not lost to Wisconsin? Remember how we thought that was going to be a PWR killer? Yet here we are and they are going to regionals.

 

[Dirty]

Re: The 2015 Pairwise, Bracketology and History Thread

Would Tech be a one seed if they had not lost to Wisconsin? Remember how we thought that was going to be a PWR killer? Yet here we are and they are going to regionals.

They'd be 3rd in the PWR right now.

 

[fr. joe]

Re: The 2015 Pairwise, Bracketology and History Thread

Probability inventories laden with covariates and poor internal validity are relatively redundant in "one and done" tournament formats.

How stochiastic and multivariate of you, and it's first thing in the morning, too. Just don't go boolian on us, that would be piling on.

 

[HarleyMC]

Re: The 2015 Pairwise, Bracketology and History Thread

The key word being relatively

For teams that are judged by face validity to be heavy favorites...relatively. But overall, the key word is redundant as covariate factors latent in the main interaction effects are too numerous for valid causality and conclusion validity, not to mention the PWR system and its parameters are embedded with selection bias. With probability models and maximum likelihood estimates, these procedures must typically be based upon data that demonstrates equality of variance as independent, normal distributions unless one implements a non-parametric model, which of course will decrease statistical power with smaller sample populations. It's noteworthy that in more advanced applications of the Bradley-Terry model, explicit information on covariates is included in the model based upon the principle of recursive partitioning which produces a tree structured regression in the covariate space.

How stochiastic and multivariate of you, and it's first thing in the morning, too. Just don't go boolian on us, that would be piling on.

Sorry about the above Joe. Just my .01 Have a good day!

 

[MagnessMan]

Re: The 2015 Pairwise, Bracketology and History Thread

For teams that are judged by face validity to be heavy favorites...relatively. But overall, the key word is redundant as covariate factors latent in the main interaction effects are too numerous for valid causality and conclusion validity, not to mention the PWR system and its parameters are embedded with selection bias. With probability models and maximum likelihood estimates, these procedures must typically be based upon data that demonstrates equality of variance as independent, normal distributions unless one implements a non-parametric model, which of course will decrease statistical power with smaller sample populations. It's noteworthy that in more advanced applications of the Bradley-Terry model, explicit information on covariates is included in the model based upon the principle of recursive partitioning which produces a tree structured regression in the covariate space.

So are you saying? It's not over till the fat lady sings. I grew up playing youth hockey in the 70's when the helmets were paper-thin and I banged my head a few times. When I read your post I get a little lost.

 

[vizoroo]

Re: The 2015 Pairwise, Bracketology and History Thread

Probability inventories laden with covariates and poor internal validity are relatively redundant in "one and done" tournament formats.

And don't forget the Dunning-Kruger effect.

 

[Priceless]

Re: The 2015 Pairwise, Bracketology and History Thread

Despite their win today, Lowell is in must-win territory for tomorrow. Any room for error went by the wayside with wins by Harvard and St Cloud. Also, Mankato now leads North Dakota for #1 overall and controls its own destiny. A loss either tonight or tomorrow coupled with a North Dakota win in 3rd-place game tomorrow puts the Whoix back on top.

 

[Critical Thinker]

Re: The 2015 Pairwise, Bracketology and History Thread

Is something wrong with the Pairwise predictor? When I run a scenario with no upsets, Yale jumps up 5 spots.

 

[LTsatch]

Re: The 2015 Pairwise, Bracketology and History Thread

Others are saying on the Yale Facebook page that the SCSU win has eliminated Yale, but some are still holding out including a very trusted source. As to the above, no way they jump up five spots.

 

[JimDahl]

Re: The 2015 Pairwise, Bracketology and History Thread

Others are saying on the Yale Facebook page that the SCSU win has eliminated Yale, but some are still holding out including a very trusted source. As to the above, no way they jump up five spots.

After the first round of games have finished... Yale is still alive and their outlook is actually unchanged (about 5% of scenarios get them in). BC seems a lock, SCSU seems a lock, BG is in much worse shape, Minnesota in much better shape.

 

[Shirtless Guy]

Re: The 2015 Pairwise, Bracketology and History Thread

After the first round of games have finished... Yale is still alive and their outlook is actually unchanged (about 5% of scenarios get them in). BC seems a lock, SCSU seems a lock, BG is in much worse shape, Minnesota in much better shape.

are you going to post a new projection tonight?

 

[JimDahl]

Re: The 2015 Pairwise, Bracketology and History Thread

are you going to post a new projection tonight?

I'll probably put together a summary of where we are after tonight's games are done/before going to bed, then if there's anything still unclear a forecast in the morning. So far, though, this is the most straightforward season in memory. Could all be pretty obvious tonight. I've mostly been posting real-time updates on Twitter https://twitter.com/chranked

 

[Numbers]

Re: The 2015 Pairwise, Bracketology and History Thread

In general, for those using the predictor, I have found what I believe are strange and bad results by not going "Start Over" to try a new scenario.

 

[Bonin21]

Re: The 2015 Pairwise, Bracketology and History Thread

I'll probably put together a summary of where we are after tonight's games are done/before going to bed, then if there's anything still unclear a forecast in the morning. So far, though, this is the most straightforward season in memory. Could all be pretty obvious tonight. I've mostly been posting real-time updates on Twitter https://twitter.com/chranked

You're the man.

 

[Numbers]

Re: The 2015 Pairwise, Bracketology and History Thread

Can someone help me? I am playing with Minnesota scenarios. Right now, Mankato leads Ferris State 3-0. Assuming that holds, I am getting Minnesota is a lock. Is that right?

Reasoning.... How many from each conference can pass them?
AHA - 1 (only the champ)
B10 - 1 (Michigan by winning. Doesn't really pass them, but takes a spot)
ECAC - 2 (Champion plus Quinn in certain scenarios)
HEA - 3 (BU, BC, and either Lowell or UNH. I can't get Providence that high, unless I glitch the Predictor by not doing a total Start Over)
NCHC - 6 no matter what
WCHA - 2 if Mankato wins - namely MSUM and MTU.

Add those up and you get Minny 16th, COUNTING the auto bids.

Can anyone get them out if Mankato wins tonight?

Thanks.

 

[Shirtless Guy]

Re: The 2015 Pairwise, Bracketology and History Thread

Can someone help me? I am playing with Minnesota scenarios. Right now, Mankato leads Ferris State 3-0. Assuming that holds, I am getting Minnesota is a lock. Is that right?

Reasoning.... How many from each conference can pass them?
AHA - 1 (only the champ)
B10 - 1 (Michigan by winning. Doesn't really pass them, but takes a spot)
ECAC - 2 (Champion plus Quinn in certain scenarios)
HEA - 3 (BU, BC, and either Lowell or UNH. I can't get Providence that high, unless I glitch the Predictor by not doing a total Start Over)
NCHC - 6 no matter what
WCHA - 2 if Mankato wins - namely MSUM and MTU.

Add those up and you get Minny 16th, COUNTING the auto bids.

Can anyone get them out if Mankato wins tonight?

Thanks.

i know Jim had them at in on 97% possibilities with a loss to Michigan tomorrow before the MSU win goes final.

 

[JimDahl]

Re: The 2015 Pairwise, Bracketology and History Thread

Can someone help me? I am playing with Minnesota scenarios. Right now, Mankato leads Ferris State 3-0. Assuming that holds, I am getting Minnesota is a lock. Is that right?

Close. A Mankato win gets them in 97% of the time. So, there's more than one niche branch to get them out. If that stays true at the end of the night I'll try to fine them.

 

[Numbers]

Re: The 2015 Pairwise, Bracketology and History Thread

Thanks. Since the NCHC 3rd place had 3 results, that probably means there are 2 scenarios, plus the NCHC game. Oh. And the NCHC final shouldn't matter either, so that leaves 1.

 

[Bonin21]

Re: The 2015 Pairwise, Bracketology and History Thread

Pretty sure I found the MN out scenario on my second try...

Atlantic Hockey
Semifinal #2: RIT defeats Canisius
Championship game: Mercyhurst defeats RIT
Big Ten
Championship game: Michigan defeats Minnesota
ECAC
Championship game: Colgate defeats Harvard
Hockey East
Championship game: Massachusetts-Lowell defeats Boston University
NCHC
Semifinal #2: Miami defeats Denver
Consolation game: North Dakota defeats Denver
Championship game: Miami defeats St. Cloud State
WCHA
Championship game: Minnesota State defeats Michigan Tech


Read more: http://www.uscho.com/rankings/pairwise-predictor/#ixzz3UzAGYFy8

 

[Priceless]

Re: The 2015 Pairwise, Bracketology and History Thread

If all higher seeds win tomorrow:

Fargo
#2 North Dakota vs RIT
Omaha vs Minnesota

South Bend
#3 Miami vs Harvard
Michigan Tech vs St Cloud

Providence
#1 Mankato vs Providence
Denver vs Colgate

Manchester
#4 Boston U vs Quinnipiac
Duluth vs Boston C