February 23rd NCAA rankings

[norm1909]

Re: February 23rd NCAA rankings

... I just feel like your current "guilty until proven innocent" mentality isn't helping people understand the process, why some things are happening the way they are, and only furthers what I believe to be a mistaken belief the process is corrupt. I think we should all be focused on getting the best information to people we can and helping them understand what actually happens.

IMO, any process that CAN, yet choose to ignore definitive, quantifiable selection criteria and instead opt for "Additionally, input is provided by regional advisory committees for consideration by the NCAA Division III Men’s Ice Hockey Committee." - Why bother? There is ample "black & white" "Primary Criteria" & "Secondary Criteria" to fill the bracket. Go with what the numbers say, and there can be no "conspiracy theory" refuse and allow bias and prejudice*, and you have a conspiracy.

* Is there REALLY a need for this: (from here - page 6 of 20)

Additionally, input is provided by regional advisory committees for consideration by the NCAA Division III Men’s Ice Hockey Committee.

 

[NUProf]

Re: February 23rd NCAA rankings

Josh, the horse I have in this race is that the process should be like Caesar's wife - totally above suspicion. I don't doubt and was not "wearing a tin hat." I don't doubt that the process is honest. However, it is too open to suspicion - just like questions about congressmen who surface as lobbyists after they leave congress - and working for industries that they had under the purview of one of their committees. In many cases, I'm sure the reason they go to work in that industry is because they know it well - is there a possible problem, to use the phrase of a politician that I detest "you betcha."

I would like to see the seedings, and pool bids decided using a statistical technique in which the only human intervention is determining the criteria and weights of the criteria before the season starts and involves no human intervention after the season starts. Does the committee have a role, yes - it should be their job to determine the appropriate way to weight the criteria. No committee meetings about selection = no conspiracy theory = total transparency.

And once again, even while I was posting, Norm put what I was thinking in a succinct statement.

 

[OZFEST]

Re: February 23rd NCAA rankings

Well my sincere apologies if I miss the word eastern committee for I thought NUPROF was listing the NCAA committee. Also get off your high horse for I can't wait to degrade you for your misunderstanding of one's post. You have posted some good ones here. So before you degrade someone else take a long looook in the mirror and see just how imperfect you are.

Come on he used a smiley

 

[Josh Carey]

Re: February 23rd NCAA rankings

* Is there REALLY a need for this: (from here - page 6 of 20)

Working backwards, your quoted text only refers to the ability of the committee members from each region to take imput from their other committee members when making the rankings and apply that into their overall discussion.

There is ample "black & white" "Primary Criteria" & "Secondary Criteria" to fill the bracket. Go with what the numbers say, and there can be no "conspiracy theory" refuse and allow bias and prejudice*, and you have a conspiracy.

Kind of replying to NUProf's similar sentiment here as well. I would like to know how each of you would propose to construct such a system. Let's say we agreed on a hypothetical system that says we only consider winning percentage, opponent's strength, and head to head comparisons. Then take the following for example:

Team A
.750 winning percentage
.550 opponent's winning percentage
1-1-0 vs Team B

Team B
.700 winning percentage
.570 opponent's winning percentage
1-1-0 vs Team A

A purely even weighting would result in a tie. Weighting WIN and SOS on a single scale would tip the comparison in the favor of Team A. But are you comfortable with that? Because WIN values range from about .0000 (Thanks LVC!) to ~ .9300 every season, but OWP values only range from ~.4000 to ~ .6000. But if you then weight on those scales, you might find this comparison ends up in a statistically insigificant difference. Who makes the choice then? Do you go by the pure statistical model no matter how small the difference may be, or do you add in more factors? And no matter how many factors you add in, you don't eliminate the possibility of statistically insigificant distinctions (between conference members especially).

You generally see two schools of valuation in sports: Strict W-L (schedule strength be whatever it is) or a committee valuation based on several factors. The BCS had so many computer issues it eventually devalued them to the point they're meaningless and that's in the most popular collegiate sport and with professional mathematicians trying to solve the issues. I love statistical analysis of teams in sports (and use sabermetrics on my fantasy baseball teams), but I imagine if you surveyed a representative sample of hockey fans and said "Team of hockey professionals" or "Predetermined Computer Model" to determine the playoffs, the vast majority of fans would say they prefer the committee.

 

[NUProf]

Re: February 23rd NCAA rankings

Kind of replying to NUProf's similar sentiment here as well. I would like to know how each of you would propose to construct such a system. Let's say we agreed on a hypothetical system that says we only consider winning percentage, opponent's strength, and head to head comparisons. Then take the following for example:

Team A
.750 winning percentage
.550 opponent's winning percentage
1-1-0 vs Team B

Team B
.700 winning percentage
.570 opponent's winning percentage
1-1-0 vs Team A

Josh, here's my issue If team A is, say Norwich, and team B is UMB, we know exactly what would happen. We now have introduced into the system an irrelevant factor, namely "reputation." As soon as you introduce the human element into the system, decisions are made on something other than data. My point is determine the system ahead - you can look at your data and know exactly and precisely what you need to do and what your opponents need to do to give you a spot. As long as the system is known throughout the season. There is a reason that the Olympics Hockey tournament didn't have committee decide the seedings for the elimination rounds, and the NHL does doesn't have a committee to pick the wild cards and seedings of the Stanley cup. Whether the systems are perfect or not doesn't matter, what matters is that they come close and are objective.

You generally see two schools of valuation in sports: Strict W-L (schedule strength be whatever it is) or a committee valuation based on several factors. The BCS had so many computer issues it eventually devalued them to the point they're meaningless and that's in the most popular collegiate sport and with professional mathematicians trying to solve the issues. I love statistical analysis of teams in sports (and use sabermetrics on my fantasy baseball teams), but I imagine if you surveyed a representative sample of hockey fans and said "Team of hockey professionals" or "Predetermined Computer Model" to determine the playoffs, the vast majority of fans would say they prefer the committee.

I can't believe you cite the BCS as an argument for subjective choices. The BCS system is exactly an example of why an objective system needs to be used. The objective measures in the early days produced matchups in the final game that didn't agree with the Polls - which are what led to the devaluing of computer and objective measures. Reputation should have nothing to do with. The BCS is all about bringing in revenue for the bowls, not about determining a national champion.

 

[Russell Jaslow]

Re: February 23rd NCAA rankings

The BCS had so many computer issues it eventually devalued them to the point they're meaningless and that's in the most popular collegiate sport and with professional mathematicians trying to solve the issues.

Isn't one of the biggest issues with the BCS computer analysis is that there are not enough games played to have any statistical significance?

 

[NUProf]

Re: February 23rd NCAA rankings

Isn't one of the biggest issues with the BCS computer analysis is that there are not enough games played to have any statistical significance?

The sample size is small, which means that the margin of error is large, but the big problem is that the results don't match prejudice.

 

[NorthernLite]

Re: February 23rd NCAA rankings

.

Kind of replying to NUProf's similar sentiment here as well. I would like to know how each of you would propose to construct such a system. Let's say we agreed on a hypothetical system that says we only consider winning percentage, opponent's strength, and head to head comparisons. Then take the following for example:

Team A
.750 winning percentage
.550 opponent's winning percentage
1-1-0 vs Team B

Team B
.700 winning percentage
.570 opponent's winning percentage
1-1-0 vs Team A

Add in more factors that can be used to break the tie. In this case two games H2H offers sufficient statistical data so the decision can be decided by on ice results.

I love statistical analysis of teams in sports (and use sabermetrics on my fantasy baseball teams), but I imagine if you surveyed a representative sample of hockey fans and said "Team of hockey professionals" or "Predetermined Computer Model" to determine the playoffs, the vast majority of fans would say they prefer the committee

.

I don't think so Josh. The committee process certainly generates interest, discussion, and speculation.....but in the end someone will have an argument that they committee got it wrong. Predetermined criteria as NUProf suggests eliminates the arguments.

I enjoy the discussion and arguments......and will be glued to the computer on selection Sunday........to find out who got screwed this year.

 

[nysportsfan29]

Re: February 23rd NCAA rankings

I can't believe you cite the BCS as an argument for subjective choices. The BCS system is exactly an example of why an objective system needs to be used. The objective measures in the early days produced matchups in the final game that didn't agree with the Polls - which are what led to the devaluing of computer and objective measures. Reputation should have nothing to do with. The BCS is all about bringing in revenue for the bowls, not about determining a national champion.

You know you have a good system in sports when the Federal Government is trying to get involved and fix it.

On another note, in terms of SOS and OWP, if you play a team twice (SUNYAC) or even three times a season (NCHA) does it count multiple times or just the once? Also I'm assuming that these numbers are constantly updated as opposed to taking the team's record at the time of the game and only using that.

 

[NUProf]

Re: February 23rd NCAA rankings

You know you have a good system in sports when the Federal Government is trying to get involved and fix it.

On another note, in terms of SOS and OWP, if you play a team twice (SUNYAC) or even three times a season (NCHA) does it count multiple times or just the once? Also I'm assuming that these numbers are constantly updated as opposed to taking the team's record at the time of the game and only using that.

It is weighted by the number of times you play, and yes it is constantly updated. The end of the season data is what matters. Whether you played LVC at the first game of the year or the last, they go into your OWP as 0-24 (not 0-25, because your win against them isn't held against you) If you played them twice they count as 0-46 (0-25, subtract your two wins making 0-23, multiply by 2 because you played them twice). Add the wins and losses together and that's your OWP. That means that every ECAC W team has a 0-66 added to their OWP W-L record - not to worry though, this is balanced by the fact that those losses will be balanced by the fact that every one of the teams that you played will have gone 3-0 against them and those league losses get wiped out by league wins of your other league opponents. The only thing that doesn't wipe out are the 7 NC losses (21 losses for each ECAC W team because they played 3 times) I hope that's clear, but I bet somebody is gonna say ???????

 

[nysportsfan29]

Re: February 23rd NCAA rankings

It is weighted by the number of times you play, and yes it is constantly updated. The end of the season data is what matters. Whether you played LVC at the first game of the year or the last, they go into your OWP as 0-24 (not 0-25, because your win against them isn't held against you) If you played them twice they count as 0-46 (0-25, subtract your two wins making 0-23, multiply by 2 because you played them twice). Add the wins and losses together and that's your OWP. That means that every ECAC W team has a 0-66 added to their OWP W-L record - not to worry though, this is balanced by the fact that those losses will be balanced by the fact that every one of the teams that you played will have gone 3-0 against them and those league losses get wiped out by league wins of your other league opponents. The only thing that doesn't wipe out are the 7 NC losses (21 losses for each ECAC W team because they played 3 times) I hope that's clear, but I bet somebody is gonna say ???????

I got the gist of it. Math wasn't really my strong suit as it is yours. Thanks NUProf

 

[MW4Bucky&SNC]

Re: February 23rd NCAA rankings

You know you have a good system in sports when the Federal Government is trying to get involved and fix it.

Might be the first time in the history of this country someone might say "This would be so much better if the Federal Government would just run it."

 

[Russell Jaslow]

Re: February 23rd NCAA rankings

Might be the first time in the history of this country someone might say "This would be so much better if the Federal Government would just run it."

Not me. I'd rather have the BCS...

 

[OZFEST]

Re: February 23rd NCAA rankings

Not me. I'd rather have the BCS...

Yes, the lesser of two weevils.

 

[Josh Carey]

Re: February 23rd NCAA rankings

Josh, here's my issue If team A is, say Norwich, and team B is UMB, we know exactly what would happen. We now have introduced into the system an irrelevant factor, namely "reputation." As soon as you introduce the human element into the system, decisions are made on something other than data. My point is determine the system ahead - you can look at your data and know exactly and precisely what you need to do and what your opponents need to do to give you a spot. As long as the system is known throughout the season. There is a reason that the Olympics Hockey tournament didn't have committee decide the seedings for the elimination rounds, and the NHL does doesn't have a committee to pick the wild cards and seedings of the Stanley cup. Whether the systems are perfect or not doesn't matter, what matters is that they come close and are objective.

If UMB ever put up numbers like that, we'd have our answer. And when a team like Neumann (which spent years in the dumpster) can get an at-large bid into the tournament while Middlebury sits at home, I'm not sure I'm buying your argument big name programs get an advantage in the selection process. I do think big name programs get rated better - but that's usually because they are better. And if anything, this process is closer to what you want than the things you described. The Canadian Olympic hockey team just destroyed Russia, but Russia had a bye because Canada was in a group with the US. The NHL doesn't have a committee to pick the wild card of the Stanley Cup, so a 9th place team in the eastern conference may sit home while a worse 8th place western conference team moves on. Or a team may finish 10th in the conference, but due to an unbalanced schedule, has a lower winning percentage against tougher opposition. The Tampa Bay Rays may be better than 11 American League teams but sit home because they play with the Yankees and Red Sox. You're saying it's better to not consider these things at all than to have a human look at them?

I can't believe you cite the BCS as an argument for subjective choices. The BCS system is exactly an example of why an objective system needs to be used. The objective measures in the early days produced matchups in the final game that didn't agree with the Polls - which are what led to the devaluing of computer and objective measures. Reputation should have nothing to do with. The BCS is all about bringing in revenue for the bowls, not about determining a national champion.

The only point I was making in reference to the BCS was that when computers were given a substantial role in selecting participants in a sport's playoff system, it was found to be unacceptable. There are literally hundreds of things wrong with the BCS, but its issues are based on the fact the humans are given no criteria to work with and your final position is based on your starting position. Not things we're dealing with here.

I don't think so Josh. The committee process certainly generates interest, discussion, and speculation.....but in the end someone will have an argument that they committee got it wrong. Predetermined criteria as NUProf suggests eliminates the arguments.

I enjoy the discussion and arguments......and will be glued to the computer on selection Sunday........to find out who got screwed this year.

I want to make it clear here: I didn't say "USCHO message board users", I said "representative sample of hockey fans". That is Joe Goaliemask you find at the rink on Saturday. Message board users are inclined to the computer solution because they've already proven they care enough to come on the message board in the first place. But I think your average fan, and hell even the vast majority of coaches and players, prefer a committee system.

And none of those replies sketched out what such a system might look like. How do you prevent Team X from getting screwed in the first year when you realize you missed something and the system needs tweaking? How do you know what isn't worthy of tweaking the system? To determine whether or not the system is working correctly, you need to have a value judgment of who "should be" in the tournament - who makes those decisions? I want to like a computer solution - somebody give me a reason to.

 

[NUProf]

Re: February 23rd NCAA rankings

Josh,

Waaaay to much to respond to all at once, but my point is that I want human prejudice out of the system. My example of NU and UMB probably wasn't phrased the way I wanted, so let me try again.

Take two programs - one with a name and one with out. My claim is, right or wrong, it wouldn't matter which team was which the name team could have been A or B, and be the one selected because the committee would bend it.

Again, the reason they changed the BCS wasn't because the computer system wasn't fair - it was because it didn't align perfectly with the polls. Until something totally objective is used, Boise State will never play in the "big game"

The main point is, win your league - if that doesn't happen - good luck in an unfair process.

 

[norm1909]

Re: February 23rd NCAA rankings

Josh,

Waaaay to much to respond to all at once, but my point is that I want human prejudice out of the system...
The main point is, win your league - if that doesn't happen - good luck in an unfair process.

I'll second that, ALL humans are prejudice, this is not to say they are malicious or otherwise ill intended, but merely that all humans are subject to human frailties. Often we talk about “top of mind awareness”, “what have you done for me lately”, “what’s in for me”, and “history/legacy/etc”, these ARE part of the human psyche, for better or for worse, it is who we are – humans – biased individuals. Numbers “are what they are”, yes it is easy to use numbers to “justify an end”, but if the process is open, objective and applied exactly to all, we will have the fairest system. “Earn it and take it”.

 

[NUProf]

Re: February 23rd NCAA rankings

I'll second that, ALL humans are prejudice, this is not to say they are malicious or otherwise ill intended, but merely that all humans are subject to human frailties. Often we talk about “top of mind awareness”, “what have you done for me lately”, “what’s in for me”, and “history/legacy/etc”, these ARE part of the human psyche, for better or for worse, it is who we are – humans – biased individuals. Numbers “are what they are”, yes it is easy to use numbers to “justify and end”, but if the process is open, subjective and applied exactly to all, we will have the fairest system. “Earn it and take it”.

Dang it norm - one of us got a whole lot smarter since I retired (either that or I'm a better reader now that I have more free time) Well put! (except I think you said subjective when you mean objective in your last sentence )

 

[norm1909]

Re: February 23rd NCAA rankings

Dang it norm - one of us got a whole lot smarter since I retired (either that or I'm a better reader now that I have more free time) Well put! (except I think you said subjective when you mean objective in your last sentence )

Looks like "objective" to me Alas, I am not a morning person, but my current work situation is making me one (should let up by mid-next week).

 

[Ozz]

Re: February 23rd NCAA rankings

Why don't we just get rid of Pool B and Pool C and then that would take this discussion right out of the picture!