Big Wins And Bad Losses: TUC

[ScoobyDoo]

Re: Big Wins And Bad Losses: TUC

The bottom line is there is no mathematical system on earth that you could create to pick the field exactly and not having people whining about who #14 is (assuming 15, and 16 are your autobids from Atlantic and CHA, next year I suppose will be arguing about who #15 is, oh joy).

 

[TimU]

Re: Big Wins And Bad Losses: TUC

Actually that's what we have now, the selection committee selects the field. The PWR was made up by USCHO to predict how the committee would pick, not the other way around. From the USCHO website

"The PairWise Rankings (PWR) are a statistical tool designed to approximate the process by which the NCAA selection committee decides which 16 teams to invite to the Division I championship tournament. Although the PWR does not precisely duplicate the method used by the committee,"

Point taken, but it's also clear that the committee uses something other than just making a list out of thin air. It's also clear that if it's not exactly the same as the PWR, it's darn close. For all intents and purposes, PWR or something very much like it is determining the field - although the committee may or may not use it, PWR is very, very good at predicting the ultimate field year after year.

 

[Patman]

Re: Big Wins And Bad Losses: TUC

You may want to read this article. The NCAA changed the PWR threshhold?? I am not sure you understand how the 16 teams actually come to be.

Here is an excerpt that in particular is important

"The PWR's detractors cite the unusually large focus on statistics such as record against "teams under consideration" (TUC). Again, though, the PWR is not the system used by the selection committee, it is merely a mathematical process designed to mimic the NCAA's process as accurately as possible - and it always works."

http://blog.masslive.com/dailycollegiansports/2009/03/mens_ice_hockey_bracketology_v.html

Your first mistake is assuming a reporter understands anything involving math. Your second mistake is that we're the ones who have been looking at this process for years. Your third mistake is in not realizing that this procedure has matched exactly for the last decade save one year where the comittee decided to flip seeds on an admitted whim.

 

[jcarter7669]

Re: Big Wins And Bad Losses: TUC

Your first mistake is assuming a reporter understands anything involving math. Your second mistake is that we're the ones who have been looking at this process for years. Your third mistake is in not realizing that this procedure has matched exactly for the last decade save one year where the comittee decided to flip seeds on an admitted whim.

I'm not sure what exactly you are saying?

A reporter doesn't need to understand math to know that the selection committee picks the teams not the PWR.
Moreover, it would appear if you check the post directly before you, you would see that the reporter simply lifted that phrase from USHCO who presumably understand math as they came up with the PWR in the first place.

The number of years one looks at the PWR doesnt' change the fact that the selection committee picks the teams not the PWR.

Finally the PWR should match the committees selection as that is what it was designed to do. I never said differently, so I am confused as to how this is a mistake.

I am a little confused as to your posts. It appears that you say a UVM fan post and needed to respond in a "gotch" type manner. Listen, take the chip off your shoulder, I'm not disagreeing with anything you have said, and I can't seem to see where you are disagreeing with what I posted. I maybe misunderstanding something for which I apologize if that's the case.

 

[LynahFan]

Re: Big Wins And Bad Losses: TUC

It's not hard at all. It just doesn't solve the expressed problem with PWR, which is that it doesn't produce an accurate enough ranking to select the best tournament field.

The response to the perceived problem is that the PWR system does identify and include the elite teams in the field. If we're trying to refine the system to better define the last couple of teams in, then NO system can do that with a 36-game system and limited conference play. We'll always be arguing about the 13-16 seeds.

If you're going to use the more accurate ranking system, and then throw out certain results in order to get a more diverse tournament field, then why bother? That puts us right where we are now (which is fine with me, by the way), except that instead of simply using an agreed-upon system to select the field, we'd be adopting a "better" system and then throwing out certain of its results to get a tournament field that could have just been picked by hand to begin with - which would be even less scientific and would cause even more arguments.

The original complaint was purely about the TUC cliff, and how it doesn't make sense that wins against the 25th best team count for so much more than wins against the 26th - not that the PWR isn't "accurate" enough.

KRACH + conference limit solves the TUC cliff complaint. Period.

"Not accurate enough" is a completely specious concept, though. To measure accuracy, you have to have a target and then see how far you are from the target. When it comes to tournament selection, there's no way to define what the target is (if we knew the target, we'd just take THOSE teams), much less to measure how far from it the selected field really is. So forget accuracy. The only thing we can fall back on is trying to find a ranking method that has features that we like - that seem fair. Like, for example, there aren't any arbitrary cliffs, that winning always helps and losing always hurts, that beating a better team helps more than beating a worse team, and so on. PWR doesn't have ANY of these characteristics, and KRACH has ALL of these characteristics.

You do the math.

 

[TimU]

Re: Big Wins And Bad Losses: TUC

The original complaint was purely about the TUC cliff, and how it doesn't make sense that wins against the 25th best team count for so much more than wins against the 26th - not that the PWR isn't "accurate" enough.

KRACH + conference limit solves the TUC cliff complaint. Period.

"Not accurate enough" is a completely specious concept, though. To measure accuracy, you have to have a target and then see how far you are from the target. When it comes to tournament selection, there's no way to define what the target is (if we knew the target, we'd just take THOSE teams), much less to measure how far from it the selected field really is. So forget accuracy. The only thing we can fall back on is trying to find a ranking method that has features that we like - that seem fair. Like, for example, there aren't any arbitrary cliffs, that winning always helps and losing always hurts, that beating a better team helps more than beating a worse team, and so on. PWR doesn't have ANY of these characteristics, and KRACH has ALL of these characteristics.

All true. But it's also true that PWR only serves one function: to determine (or mimic the tool used to determine) the NCAA tournament field. The only reason to switch to a different system is to come up with a different tournament field. You say that KRACH's predictors of team quality are more fair (which I won't dispute). But if you adopt KRACH, and then change the results if it gives you a field that looks wrong, then I still don't see how you're any better off. And if you adopted KRACH and DIDN'T have a failsafe preventing more than five teams from a conference getting in, then you'd be worse off still.

It's also worth repeating that KRACH and PWR don't really disagree on the best of the best - it's only at the edges where there are conflicts and uncertainties, and they would not be solved by the use of KRACH because of the nature of the college hockey schedule. We would still have arguments over whether it was fair to include one team or exclude another as the 16 seed. If anything KRACH + conference limits would make things worse, since the chosen system could identify two teams as being the 8th and 9th best in the country, and then expressly exclude them in favor of the 18th- and 20th-best teams. Talk about arguments over arbitrary cliffs.

 

[Slap Shot]

Re: Big Wins And Bad Losses: TUC

I'm seeing a really poor defense of the PWR. It in essence states that, "We're sort of looking for the best at-large tournament field, but we'll artificially tweak it so to not make some people feel bad or because we don't want to look silly to the 3 non-college hockey fans that pay attention to the selection process". Nobody whines in the NFL if one division doesn't have a wild card team, nobody whines in the NHL if a division is highly represented in the top 8. Why should it be any different in college hockey?

Scrap the PWR and use the KRACH since it arguably rates teams more accurately top to bottom. If we really don't want over-respresentation by conference and you cap conference entries, that '*' is no more trite than assigning TUC status to only X number of teams, which is a '*' of its own.

 

[Innova]

Re: Big Wins And Bad Losses: TUC

And if you adopted KRACH and DIDN'T have a failsafe preventing more than five teams from a conference getting in, then you'd be worse off still.

How does PWR limit the number of teams from a conference? Is it in the formula, or a manual process?

 

[ScoobyDoo]

Re: Big Wins And Bad Losses: TUC

I'm seeing a really poor defense of the PWR. It in essence states that, "We're sort of looking for the best at-large tournament field, but we'll artificially tweak it so to not make some people feel bad or because we don't want to look silly to the 3 non-college hockey fans that pay attention to the selection process". Nobody whines in the NFL if one division doesn't have a wild card team, nobody whines in the NHL if a division is highly represented in the top 8. Why should it be any different in college hockey?

Scrap the PWR and use the KRACH since it arguably rates teams more accurately top to bottom. If we really don't want over-respresentation by conference and you cap conference entries, that '*' is no more trite than assigning TUC status to only X number of teams, which is a '*' of its own.

That's funny cause I'm seeing the opposite for KRACH. You seem to think that 8 teams from the WCHA a few years ago would have been acceptable? How so? The only reason those 8 teams from the WCHA were in using KRACH is because of insular scheduling and the skewing that can happen when one conference just happens to dominate a very small sample size of non-conference play.

Meanwhile the PWR (yes it's tweaked and tweaked again, but it's been transparent) has had a good run from 2003 on since they went to 16 teams of picking a great field and having great tournaments. And even though the rankings are mysterious to the layman, there are no funky rules about this many from that conference and this many from this conference which I positively hate as a concept. Where's the line? 6, 5? Seems that conference line is as perilous as the TUC cliff. In fact you KRACH fanatics have not addressed it. What happens when 8 of the top 10 teams in KRACH are from the WCHA again (oh, it'll happen) and you tell 3 of them they have to stay home? Won't you get the same amount of whining that you do now? Sure, KRACH is a much more elegant mathematical formula, blah, blah, blah, and frankly I don't care either way, but the mere fancy that any mathematical formula is going to work or not have incessant whining due to the small data size is folly.

 

[MaizeRage]

Re: Big Wins And Bad Losses: TUC

You may want to read this article. The NCAA changed the PWR threshhold?? I am not sure you understand how the 16 teams actually come to be.

Here is an excerpt that in particular is important

"The PWR's detractors cite the unusually large focus on statistics such as record against "teams under consideration" (TUC). Again, though, the PWR is not the system used by the selection committee, it is merely a mathematical process designed to mimic the NCAA's process as accurately as possible - and it always works."

http://blog.masslive.com/dailycollegiansports/2009/03/mens_ice_hockey_bracketology_v.html

That was once the case, but not any more. If you look, the entire PWR system is in the NCAA rule book and that's all the committee is allowed to look at.

 

[Slap Shot]

Re: Big Wins And Bad Losses: TUC

Scoobs - I'm no more a KRACH fanatic than you are a PWR fanatic. If anything I find the conversation interesting and don't mind analyzing the pros and cons of either system.

That's funny cause I'm seeing the opposite for KRACH. You seem to think that 8 teams from the WCHA a few years ago would have been acceptable? How so?

I repeat:

If we really don't want over-respresentation by conference and you cap conference entries, that '*' is no more trite than assigning TUC status to only X number of teams, which is a '*' of its own.

You sort of respond to that here:

What happens when 8 of the top 10 teams in KRACH are from the WCHA again (oh, it'll happen) and you tell 3 of them they have to stay home? Won't you get the same amount of whining that you do now?

There's always going to be complaints, I'm simply saying if we're going to "*" or tweak the at-large entrants, doing so by limiting the # of TUCs is less valid than limiting the number of confererence entrants given the KRACH more truly ranks teams top to bottom. And whether or not the PWR has provided competitive fields is irrelevant since there's no reason to believe a KRACH assigned field would not be equally competitive.

btw - for the record if the KRACH had in fact assigned 8 WCHA teams I don't know that I would give a rats behind. Something tells me an 8-team conference berth by KRACH would be a few and far between event.

 

[blockski]

Re: Big Wins And Bad Losses: TUC

How does PWR limit the number of teams from a conference? Is it in the formula, or a manual process?

Technically, it doesn't. Functionally, it does, as the only way you'll get one conference dominating the at-large bids is if that conference is actually dominant. The nature of the PWR's comparisons doesn't give as much weight to strength of schedule, particularly for the lower teams in a strong conference, as they end up with mediocre records against strong competition.

Again, the way they've tweaked the PWR and RPI from year to year is an adjustment to the process by changing the ranking system, rather than just leaving the ranking as-is and adjusting the selection process to meet goals for conference and regional equity, etc.

 

[MN Amh Fan]

Re: Big Wins And Bad Losses: TUC

I posted this a year ago... One way to objectively measure KRACH vs PWR is to see how they do in predicting the NCAA results. So, for 2004-2008, here are the results (no real change if you include 2009 and I'm going to wait to update until after the 2010 tourney).

Average National Champion Rank
KRACH – 6.4
PWR – 5.2
* not statistically different

Average National Runner-Up
KRACH – 8
PWR – 7.4
* not statistically different

Average Frozen Four participants
KRACH – 6.15
PWR – 6.15
* not statistically different

Out of the 75 games played (15 games per year for 5 years), did they pick the winner:
KRACH – 46/75 (61%)
PWR – 43/75 (57%)
* not statistically different

While there are differences in how the KRACH and PWR are calculated, the results fail to pick one as being any better than the other.

 

[jcarter7669]

Re: Big Wins And Bad Losses: TUC

I posted this a year ago... One way to objectively measure KRACH vs PWR is to see how they do in predicting the NCAA results. So, for 2004-2008, here are the results (no real change if you include 2009 and I'm going to wait to update until after the 2010 tourney).

Average National Champion Rank
KRACH – 6.4
PWR – 5.2
* not statistically different

Average National Runner-Up
KRACH – 8
PWR – 7.4
* not statistically different

Average Frozen Four participants
KRACH – 6.15
PWR – 6.15
* not statistically different

Out of the 75 games played (15 games per year for 5 years), did they pick the winner:
KRACH – 46/75 (61%)
PWR – 43/75 (57%)
* not statistically different

While there are differences in how the KRACH and PWR are calculated, the results fail to pick one as being any better than the other.

Hey !! Don't be bring pesky statistics into a perfectly good rant on why there is a bias against crappy teams in getting into the tourney.

 

[ScoobyDoo]

Re: Big Wins And Bad Losses: TUC

Scoobs - I'm no more a KRACH fanatic than you are a PWR fanatic. If anything I find the conversation interesting and don't mind analyzing the pros and cons of either system.

You said that you saw poor defense of PWR and seemed to be in love with KRACH. Sorry if I extrapolated too far.

And you poo pooed the 8 team thing but look at the numbers today for KRACH and tell me it's not an issue.

KRACH has 6 WCHA teams and 4 of them are in the top 5. That' not likely to change. Do you think one conference should have 6 teams in the tourney? Maybe it should, I don't know what the number would be and it seems to me 5 would be enough. Should 6 teams from a 10 team conference make the NCAA tournament? That's even more than makes the conference tournament.

 

[The Sicatoka]

Re: Big Wins And Bad Losses: TUC

I can solve all of everyone's qualms about who's and who's out and by which system.

58 NCAA DI mens ice hockey teams.

64 team bracket with six teams (how about the regular season conference winners) getting an opening bye.

 

[ScoobyDoo]

Re: Big Wins And Bad Losses: TUC

I can solve all of everyone's qualms about who's and who's out and by which system.

58 NCAA DI mens ice hockey teams.

64 team bracket with six teams (how about the regular season conference winners) getting an opening bye.

That'll work.

 

[Slap Shot]

Re: Big Wins And Bad Losses: TUC

KRACH has 6 WCHA teams and 4 of them are in the top 5.

How many made it the year of the WCHA Frozen Four via the PWR? And if the 6 from above doesn't include UM, it's probably correct.

 

[ScoobyDoo]

Re: Big Wins And Bad Losses: TUC

How many made it the year of the WCHA Frozen Four via the PWR? And if the 6 from above doesn't include UM, it's probably correct.

Ain't that the TRUTH.

 

[Happy]

Re: Big Wins And Bad Losses: TUC

You said that you saw poor defense of PWR and seemed to be in love with KRACH. Sorry if I extrapolated too far.

And you poo pooed the 8 team thing but look at the numbers today for KRACH and tell me it's not an issue.

KRACH has 6 WCHA teams and 4 of them are in the top 5. That' not likely to change. Do you think one conference should have 6 teams in the tourney? Maybe it should, I don't know what the number would be and it seems to me 5 would be enough. Should 6 teams from a 10 team conference make the NCAA tournament? That's even more than makes the conference tournament.

if the six teams from that conference are better than the teams from other conferences, than **** right they should be in the tourney. why should teams from the WCHA be discriminated because they do not get to play in an easy conference? clearly, a team that plays 25 TUC and 14 games against top ten will probably have worse record than a team that doesn't even play 10 tuc, so right now a team in 6th place is probably better than any team in the lower 3 conferences.

and, all WCHA teams make the WCHA tourney.