What's new
By:
USCHO Fan Forum

This is a sample guest message. Register a free account today to become a member! Once signed in, you'll be able to participate on this site by adding your own topics and posts, as well as connect with other members through your own private inbox!

  • The USCHO Fan Forum has migrated to a new plaform, xenForo. Most of the function of the forum should work in familiar ways. Please note that you can switch between light and dark modes by clicking on the gear icon in the upper right of the main menu bar. We are hoping that this new platform will prove to be faster and more reliable. Please feel free to explore its features.

1/28/13 Poll Speculation

sshablak said:
I have to say the voters(coaches and writers) really "probably" don't put much time into their voting. It's way more important to the fans here(some too much)...myself included I guess. I gave up two or three weeks ago wondering about the "validity" of these polls. Who the hell are the voters anyway...is there a list of them ??? Can we see who they are ?
Click to expand...
Remy is the lone voter, so the secrets out. He also controls the NCAAs effective April 1st.
 
Re: 1/28/13 Poll Speculation

I should preface this by saying that I profess that rankings mean little, but what in the wide wide world of sports is UC doing in the 8 slot. I shake my head.....That is all I got....:eek:
 
Re: 1/28/13 Poll Speculation

Who knows if they deserve it, but Bowdoin has never been ranked as high as 2nd:D On Feb 11 2002 they were ranked 3rd with a 16-0-3 record (and last week, of course). If history has taught us one thing it's that they'll suffer some sort of epic collapse now:eek: We can all enjoy until Friday, at any rate.
 
Re: 1/28/13 Poll Speculation

bakdraft21 said:
like I said b4 I like the D3 rankings better, but who is doing the voting??
Click to expand...

Two writers and 18 coaches vote in the poll. The coaches are distributed among each of the D-III leagues. USCHO.com does not release the names of the voters in any of its polls.
 
Re: 1/28/13 Poll Speculation

You'll go crazy if you try to make sense of the USCHO poll/D-3 hockey.com embarrassments... Thankfully, neither of them matters at all, anyway.

I'd advise you to hang your hat on the KRACH until the NCAA rankings are unveiled to us unworthy stinkards; that metric- and the PWR, which is very close to it - are apparently what the committee has used as its guide lately, and rationally so.
 
Re: 1/28/13 Poll Speculation

Fishman'81 said:
You'll go crazy if you try to make sense of the USCHO poll/D-3 hockey.com embarrassments... Thankfully, neither of them matters at all, anyway.

I'd advise you to hang your hat on the KRACH until the NCAA rankings are unveiled to us unworthy stinkards; that metric- and the PWR, which is very close to it - are apparently what the committee has used as its guide lately, and rationally so.
Click to expand...

Do you know anything about the metrics that the so-called "KRACH" (actually the Bradly-Terry system) uses? There is a lot of smoke and mirrors that goes on there - it uses logistic regression and was actually originally designed to rank chess players. The d3 hockey schedule doesn't meet all of the assumptions that are required for the use of logistic regression, so there are some "cluges" built into it that make it not as reliable as the KRACH fanboys would have you think. It is an interesting system and worthy of conversation, but it is not close to the be-all end all of ranking systems. (In no way am I implying that other well-known computer ranking systems are better, because due to low connectivity, no computer ranking system will ever be as reliable as we would like).
 
Re: 1/28/13 Poll Speculation

NUProf said:
Do you know anything about the metrics that the so-called "KRACH" (actually the Bradly-Terry system) uses? There is a lot of smoke and mirrors that goes on there - it uses logistic regression and was actually originally designed to rank chess players. The d3 hockey schedule doesn't meet all of the assumptions that are required for the use of logistic regression, so there are some "cluges" built into it that make it not as reliable as the KRACH fanboys would have you think. It is an interesting system and worthy of conversation, but it is not close to the be-all end all of ranking systems. (In no way am I implying that other well-known computer ranking systems are better, because due to low connectivity, no computer ranking system will ever be as reliable as we would like).
Click to expand...

What is the major assumption KRACH violates, for example is it something as simple as the sample to small or should we not be using a linear model as the model for rankings?
 
Re: 1/28/13 Poll Speculation

NUProf said:
Do you know anything about the metrics that the so-called "KRACH" (actually the Bradly-Terry system) uses? There is a lot of smoke and mirrors that goes on there - it uses logistic regression and was actually originally designed to rank chess players. The d3 hockey schedule doesn't meet all of the assumptions that are required for the use of logistic regression, so there are some "cluges" built into it that make it not as reliable as the KRACH fanboys would have you think. It is an interesting system and worthy of conversation, but it is not close to the be-all end all of ranking systems. (In no way am I implying that other well-known computer ranking systems are better, because due to low connectivity, no computer ranking system will ever be as reliable as we would like).
Click to expand...

+1
 
NUProf said:
Do you know anything about the metrics that the so-called "KRACH" (actually the Bradly-Terry system) uses? There is a lot of smoke and mirrors that goes on there - it uses logistic regression and was actually originally designed to rank chess players. The d3 hockey schedule doesn't meet all of the assumptions that are required for the use of logistic regression, so there are some "cluges" built into it that make it not as reliable as the KRACH fanboys would have you think. It is an interesting system and worthy of conversation, but it is not close to the be-all end all of ranking systems. (In no way am I implying that other well-known computer ranking systems are better, because due to low connectivity, no computer ranking system will ever be as reliable as we would like).
Click to expand...
+2
 
Re: 1/28/13 Poll Speculation

UCAlum32 said:
What is the major assumption KRACH violates, for example is it something as simple as the sample to small or should we not be using a linear model as the model for rankings?
Click to expand...

Bradley - Terry is derived from a choice model in which individuals express preferences in pairwise comparisons and logistic regression is applied. It assumes all comparisons are independent (in particular this means if two teams play a two game series back to back that there is no influence of game 1 on the result of game 2), and the network of comparisons must be connected. In the pure Bradley Terry method, undefeated teams (a preference which is always chosen) ends up with an infinite rating - a winless team ends up with a 0 rating. In order to adjust for that (and produce connectivity) a fictitious team is introduced - one that has tied every other team.

To deal with this problem, and the corresponding problems of teams with nearly-perfect records having ratings that are unreasonably high (or low), I pretend that each team has played, and tied, a "fictitious" game against a fictitious opponent with rating 100 (ie. average). Since this is the same for every team, the overall ranking is not unduly affected ("not biasedly affected" is the most accurate way to say it), and it ensures that each team has both gained and dropped a point -- in the fictitious game, if nowhere else -- so that the problem of infinitely large and zero ratings is sidestepped. I always include the fictitious games in the database, even when no teams have perfect (perfectly futile) records, so that comparisons from week to week can be made; as the season progresses, and more "real" games are played, the effect of the fictitious games diminishes.
Click to expand...

There is no doubt it is interesting, and the questions that I have about revolve around the independence assumption and the effect of the fictional team. The big question is whether the resulting hybrid model measures what Bradley-Terry originally intended to model. The other thing that I dislike about it is that there is no mention of a confidence level for the reported statistics. Any statistical computation has a margin of error - how accurate are the estimates.
 
Re: 1/28/13 Poll Speculation

This is why USCHO doesn't endorse KRACH. We provide it because people want to see it and because it keeps people from having to go elsewhere to find it. (Another hockey site that is D-I only endorses it and thinks it should be used by the NCAA.)
 
Re: 1/28/13 Poll Speculation

Ed Trefzger said:
This is why USCHO doesn't endorse KRACH. We provide it because people want to see it and because it keeps people from having to go elsewhere to find it. (Another hockey site that is D-I only endorses it and thinks it should be used by the NCAA.)
Click to expand...

I just use it to get a gauge on every team's winning percentage in games that count and to see relatively where each team's SOS stacks up. Other than that, I wait for the NCAA rankings.

I think there are useful tools that the KRACH shows you, but to take it as the gospel is a bit of a stretch as well.

Would fishman be touting the KRACH if Utica wasn't #1 in it? That's the $64,000 question.

And just to clarify, I have nothing agaisnt Utica. I think it's a travesty that they are as low as they are in the USCHO Poll.

I also stand by the fact that if they keep winning, they'll have a great shot at being the No.1 team in the East in the first NCAA rankings due out two weeks from now. I'm pretty sure next week the committee convenes and produces the first rankings that are not made available to the public and then the following week, Feb. 12 we'll see the first public rankings.
 
Re: 1/28/13 Poll Speculation

I didn't know if this has been answered but , does anyone know why Nick Lozorko left Castleton? He is now playing for the Huntsville Havoc of the SPHL. I figured I would put it here because a lot of people seem to check out this thread.
 
Re: 1/28/13 Poll Speculation

I got a "C" in sadistics. I followed most of it. When John Whelan and Prof get talking about the higher math stuff, my brain goes into lock (said by the guy who never worked so hard in his life for a "D" in Advanced Calculus).
 
Re: 1/28/13 Poll Speculation

joecct said:
I got a "C" in sadistics. I followed most of it. When John Whelan and Prof get talking about the higher math stuff, my brain goes into lock (said by the guy who never worked so hard in his life for a "D" in Advanced Calculus).
Click to expand...
How did you ever get out of Clarkson with a 'D' in Advanced Calculus?
 
You must log in or register to reply here.
Back
Top