Pool C

Started by usee, October 28, 2008, 12:25:35 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Frank Rossi

Quote from: Mr. Ypsi on November 12, 2008, 05:43:11 PM
Quote from: Frank Rossi on November 12, 2008, 04:52:26 PM
Regarding Wooster, I think this is a place where the numbers lie a lot.  I really think the Committee is going to try to look at the OWP of the teams YOU BEAT as much as your overall OWP.  If that's the case, RPI's numbers fall, but Wooster's numbers freefall.  Then you need to look at Win % -- Wooster lost twice, RPI once.  I think a partially subjective analysis gives you a better comparison when it comes to those two teams since you really need some good reasons to place a two-loss team ahead of a one-loss team.  In this situation, those conditions do not really exist.

Very true, but the committee MIGHT decide to flip the logic - Wooster is surely the only two-loss team in the country where BOTH losses came to teams who are still undefeated!

But do we reward winning or losing in football?

Mr. Ypsi

Quote from: Frank Rossi on November 12, 2008, 05:44:54 PM
Quote from: Mr. Ypsi on November 12, 2008, 05:43:11 PM
Quote from: Frank Rossi on November 12, 2008, 04:52:26 PM
Regarding Wooster, I think this is a place where the numbers lie a lot.  I really think the Committee is going to try to look at the OWP of the teams YOU BEAT as much as your overall OWP.  If that's the case, RPI's numbers fall, but Wooster's numbers freefall.  Then you need to look at Win % -- Wooster lost twice, RPI once.  I think a partially subjective analysis gives you a better comparison when it comes to those two teams since you really need some good reasons to place a two-loss team ahead of a one-loss team.  In this situation, those conditions do not really exist.

Very true, but the committee MIGHT decide to flip the logic - Wooster is surely the only two-loss team in the country where BOTH losses came to teams who are still undefeated!

But do we reward winning or losing in football?

d'oh! ;)  But if we took that argument to its logical extreme, only undefeated teams would be allowed into the tourney.  I don't expect Woo to make it in either, but I dare say there will be no shortage of teams who DO get in who would have lost to both Wabash and CWRU.

Frank Rossi

Quote from: Mr. Ypsi on November 12, 2008, 05:50:06 PM
Quote from: Frank Rossi on November 12, 2008, 05:44:54 PM
Quote from: Mr. Ypsi on November 12, 2008, 05:43:11 PM
Quote from: Frank Rossi on November 12, 2008, 04:52:26 PM
Regarding Wooster, I think this is a place where the numbers lie a lot.  I really think the Committee is going to try to look at the OWP of the teams YOU BEAT as much as your overall OWP.  If that's the case, RPI's numbers fall, but Wooster's numbers freefall.  Then you need to look at Win % -- Wooster lost twice, RPI once.  I think a partially subjective analysis gives you a better comparison when it comes to those two teams since you really need some good reasons to place a two-loss team ahead of a one-loss team.  In this situation, those conditions do not really exist.

Very true, but the committee MIGHT decide to flip the logic - Wooster is surely the only two-loss team in the country where BOTH losses came to teams who are still undefeated!

But do we reward winning or losing in football?

d'oh! ;)  But if we took that argument to its logical extreme, only undefeated teams would be allowed into the tourney.  I don't expect Woo to make it in either, but I dare say there will be no shortage of teams who DO get in who would have lost to both Wabash and CWRU.

My interest is based on the possibility that Wooster and RPI could be jousting for a final slot this weekend.  If Wooster had beaten a quality team, I'd see the logic.  But both teams have mediocre wins -- and Wooster has one more loss.  The OWP deflates, and the OOWP isn't great.  I can't see a great reason to jump them over RPI if it comes down to it.

HScoach

Frank:  Great analysis!
I find easily offended people rather offensive!

Statistics are like bikinis; what they reveal is interesting, what they hide is essential.

usee

Frank,

Thanks for your work. This is as good an explanation as it gets. I will refer all to this post. We should definitely save this for future reference. Thanks!

Ralph Turner

Quote from: Frank Rossi on November 12, 2008, 05:05:28 PM
...

Ralph - Are you thinking that my Pool B assessment regarding Wesley being a definitive selection is correct?  It seems like there's a decent amount of debating going on about this point.
Good evening, Frank.  Yes!

With the release of the Rankings today, and the jump that we saw in Wesley's ranking from 10th to 8th, I am more confident that Wesley is one of the Pool B's, assuming that the following teams win:

1) CWRU
2) Wesley
3) Huntingdon/LaGrange (of course) or
3) Northwestern MN (and beating St Thomas is a well-respected win).

We also know that Salisbury is not #10.  I still think that it will be easier for LaGrange to slip in as an undefeated team than for Huntingdon to get the bid, but I am not at the Committee table.

Thanks for your analysis!

Frank Rossi

Appreciate the kudos guys.  I can't guarantee the results because we've seen the final pool be somewhat trumped by the National Committee in the past -- however, the analysis is based on what's in front of us at this time. 

As I said to Ralph earlier, it's clear that there for five clear Pool C leaders that can be interchanged in the first five picks -- and maybe Montclair could even be confirmed as a sixth pretty firm pick.  The dropoff from there seems somewhat dramatic with what's left on the board. 

If there are 3 or 4 upsets (including the two scenarios that weren't covered in that post since it was LLPP geared -- namely RPI losing or Hobart losing this weekend), then Pool C gets VERY muddled.  Stay tuned.

redswarm81

Quote from: Mr. Ypsi on November 12, 2008, 05:43:11 PM
Quote from: Frank Rossi on November 12, 2008, 04:52:26 PM
Regarding Wooster, I think this is a place where the numbers lie a lot.  I really think the Committee is going to try to look at the OWP of the teams YOU BEAT as much as your overall OWP.  If that's the case, RPI's numbers fall, but Wooster's numbers freefall.  Then you need to look at Win % -- Wooster lost twice, RPI once.  I think a partially subjective analysis gives you a better comparison when it comes to those two teams since you really need some good reasons to place a two-loss team ahead of a one-loss team.  In this situation, those conditions do not really exist.

Very true, but the committee MIGHT decide to flip the logic - Wooster is surely the only two-loss team in the country where BOTH losses came to teams who are still undefeated!

That ought to be reflected in the OWP.

Ultimately, I think that the combination of winning percentage and OWP/OOWP can be a fairly comprehensive system of comparison.  As Frank asked, do we reward winning or losing?  The winning percentage comparison rewards winning, the OWP/OOWP stats minimize the damage inflicted by losses to good teams.
Irritating SAT-lagging Union undergrads and alums since 1977

redswarm81

Quote from: Frank Rossi on November 12, 2008, 04:14:15 PM
This is an analysis of the state of affairs if Pool C were being picked today. 

How Pool C Works
------------------

So, now, let's look at how Pool C works.  There are six slots this year in Pool C. . . .

Using this week's Regional Rankings, here is the likely seeding of each region's Pool C nominees:

East:

1) Ithaca, 2) Montclair, 3) RPI, 4) Hartwick, 5) Rowan and 6) Curry

North:

1) Otterbein, 2) Wooster and 3) Elmhurst

South:

1) Hardin-Simmons, 2) Hampden-Sydney, 3) Wash. & Jeff. and 4) The Winner of LaGrange/Huntingdon if not chosen for Pool C

West:

1) UW-Whitewater, 2) Redlands and 3) Northwestern (Minn.) if not chosen for Pool C


Man oh man Frank, I'm totally geeking out about your analysis!  Great stuff.  It almost makes me excited enough to dig through each Pool C team's record to see what kind of records v. RROs are in the mix.  From my armchair, that's the only other criterion I've heard mentioned fairly consistently when "insiders" comment on the selection process retrospectively.
Irritating SAT-lagging Union undergrads and alums since 1977

Mr. Ypsi

Quote from: redswarm81 on November 12, 2008, 10:39:38 PM
Quote from: Mr. Ypsi on November 12, 2008, 05:43:11 PM
Quote from: Frank Rossi on November 12, 2008, 04:52:26 PM
Regarding Wooster, I think this is a place where the numbers lie a lot.  I really think the Committee is going to try to look at the OWP of the teams YOU BEAT as much as your overall OWP.  If that's the case, RPI's numbers fall, but Wooster's numbers freefall.  Then you need to look at Win % -- Wooster lost twice, RPI once.  I think a partially subjective analysis gives you a better comparison when it comes to those two teams since you really need some good reasons to place a two-loss team ahead of a one-loss team.  In this situation, those conditions do not really exist.

Very true, but the committee MIGHT decide to flip the logic - Wooster is surely the only two-loss team in the country where BOTH losses came to teams who are still undefeated!

That ought to be reflected in the OWP.

Ultimately, I think that the combination of winning percentage and OWP/OOWP can be a fairly comprehensive system of comparison.  As Frank asked, do we reward winning or losing?  The winning percentage comparison rewards winning, the OWP/OOWP stats minimize the damage inflicted by losses to good teams.

True, but I'm a retired stats prof, and a 10 (or less) game schedule (which rarely crosses regional lines) is just too small a sample to rely on.  In the football picks, I think the criteria MUST be augmented with 'common sense'.  Of course, 'common sense' is a very loaded 'criterion'.

Personally, I'm hoping for a whole bunch of 'upsets' Saturday to render most of this moot - then we can (for a few hours ;)) have this to do all over again! ;D

Ralph Turner

Quote from: Mr. Ypsi on November 12, 2008, 11:07:25 PM
Quote from: redswarm81 on November 12, 2008, 10:39:38 PM
Quote from: Mr. Ypsi on November 12, 2008, 05:43:11 PM
Quote from: Frank Rossi on November 12, 2008, 04:52:26 PM
Regarding Wooster, I think this is a place where the numbers lie a lot.  I really think the Committee is going to try to look at the OWP of the teams YOU BEAT as much as your overall OWP.  If that's the case, RPI's numbers fall, but Wooster's numbers freefall.  Then you need to look at Win % -- Wooster lost twice, RPI once.  I think a partially subjective analysis gives you a better comparison when it comes to those two teams since you really need some good reasons to place a two-loss team ahead of a one-loss team.  In this situation, those conditions do not really exist.

Very true, but the committee MIGHT decide to flip the logic - Wooster is surely the only two-loss team in the country where BOTH losses came to teams who are still undefeated!

That ought to be reflected in the OWP.

Ultimately, I think that the combination of winning percentage and OWP/OOWP can be a fairly comprehensive system of comparison.  As Frank asked, do we reward winning or losing?  The winning percentage comparison rewards winning, the OWP/OOWP stats minimize the damage inflicted by losses to good teams.

True, but I'm a retired stats prof, and a 10 (or less) game schedule (which rarely crosses regional lines) is just too small a sample to rely on.  In the football picks, I think the criteria MUST be augmented with 'common sense'.  Of course, 'common sense' is a very loaded 'criterion'.

Personally, I'm hoping for a whole bunch of 'upsets' Saturday to render most of this moot - then we can (for a few hours ;)) have this to do all over again! ;D
Yes!  I know what I would like to see.  :)

How many games in the D3 regional format are necessary to get a decent sample.

The 22-25 in-region games in basketball?  The 28-40 "in-region" games in baseball?  The 15-18 "in-region" games in soccer? 

Thanks, oh exalted statistician!   :)

Ralph Turner

Quote from: Mr. Ypsi on November 12, 2008, 11:07:25 PM
...
Personally, I'm hoping for a whole bunch of 'upsets' Saturday to render most of this moot - then we can (for a few hours ;)) have this to do all over again! ;D

The task...

it is Sisyphian!   :D

redswarm81

Quote from: Mr. Ypsi on November 12, 2008, 11:07:25 PM
Quote from: redswarm81 on November 12, 2008, 10:39:38 PM
Quote from: Mr. Ypsi on November 12, 2008, 05:43:11 PM
Quote from: Frank Rossi on November 12, 2008, 04:52:26 PM
Regarding Wooster, I think this is a place where the numbers lie a lot.  I really think the Committee is going to try to look at the OWP of the teams YOU BEAT as much as your overall OWP.  If that's the case, RPI's numbers fall, but Wooster's numbers freefall.  Then you need to look at Win % -- Wooster lost twice, RPI once.  I think a partially subjective analysis gives you a better comparison when it comes to those two teams since you really need some good reasons to place a two-loss team ahead of a one-loss team.  In this situation, those conditions do not really exist.

Very true, but the committee MIGHT decide to flip the logic - Wooster is surely the only two-loss team in the country where BOTH losses came to teams who are still undefeated!

That ought to be reflected in the OWP.

Ultimately, I think that the combination of winning percentage and OWP/OOWP can be a fairly comprehensive system of comparison.  As Frank asked, do we reward winning or losing?  The winning percentage comparison rewards winning, the OWP/OOWP stats minimize the damage inflicted by losses to good teams.

True, but I'm a retired stats prof, and a 10 (or less) game schedule (which rarely crosses regional lines) is just too small a sample to rely on.  In the football picks, I think the criteria MUST be augmented with 'common sense'.  Of course, 'common sense' is a very loaded 'criterion'.

Personally, I'm hoping for a whole bunch of 'upsets' Saturday to render most of this moot - then we can (for a few hours ;)) have this to do all over again! ;D

I'm with you, Prof--but why didn't your academic brethren listen to me when I was a mere undergrad, and they were dishing out grades of C to all of us within one standard deviation of the mean, in a class of 18?  I mean, I know the guys who took this class last year, Doc--they were a bunch of morons, we're all smarter than them!  :D

Do you have a favorite upset possibility?  I'm pulling for Alfred over St. John Fisher (to cap the 2008 E8 Clusterfu. . . nny story), and for some odd reason, I feel compelled to root for John Carroll v. Otterbein.

Note also that I said I think that the combination of winning percentage and OWP/OOWP can be a fairly comprehensive system of comparison.   There would have to be some weighting of those criteria, but I haven't come up with a clear idea of what the weighting scheme would be, other than winning percentage is somewhat more valuable than OWP/OOWP--but how much more valuable, I can't yet say.
Irritating SAT-lagging Union undergrads and alums since 1977

Frank Rossi

Quote from: Ralph Turner on November 12, 2008, 11:15:57 PM
Quote from: Mr. Ypsi on November 12, 2008, 11:07:25 PM
Quote from: redswarm81 on November 12, 2008, 10:39:38 PM
Quote from: Mr. Ypsi on November 12, 2008, 05:43:11 PM
Quote from: Frank Rossi on November 12, 2008, 04:52:26 PM
Regarding Wooster, I think this is a place where the numbers lie a lot.  I really think the Committee is going to try to look at the OWP of the teams YOU BEAT as much as your overall OWP.  If that's the case, RPI's numbers fall, but Wooster's numbers freefall.  Then you need to look at Win % -- Wooster lost twice, RPI once.  I think a partially subjective analysis gives you a better comparison when it comes to those two teams since you really need some good reasons to place a two-loss team ahead of a one-loss team.  In this situation, those conditions do not really exist.

Very true, but the committee MIGHT decide to flip the logic - Wooster is surely the only two-loss team in the country where BOTH losses came to teams who are still undefeated!

That ought to be reflected in the OWP.

Ultimately, I think that the combination of winning percentage and OWP/OOWP can be a fairly comprehensive system of comparison.  As Frank asked, do we reward winning or losing?  The winning percentage comparison rewards winning, the OWP/OOWP stats minimize the damage inflicted by losses to good teams.

True, but I'm a retired stats prof, and a 10 (or less) game schedule (which rarely crosses regional lines) is just too small a sample to rely on.  In the football picks, I think the criteria MUST be augmented with 'common sense'.  Of course, 'common sense' is a very loaded 'criterion'.

Personally, I'm hoping for a whole bunch of 'upsets' Saturday to render most of this moot - then we can (for a few hours ;)) have this to do all over again! ;D
Yes!  I know what I would like to see.  :)

How many games in the D3 regional format are necessary to get a decent sample.

The 22-25 in-region games in basketball?  The 28-40 "in-region" games in baseball?  The 15-18 "in-region" games in soccer? 

Thanks, oh exalted statistician!   :)

Number of games isn't so much the problem.  It's number of OUT-OF-CONFERENCE games that matter.  As Mr. Kaiser said to me on Monday, every team's OWP vs. conference opponents will be about .500 -- the only thing that makes a real difference is the out-of-conference opponents you play.  Well, depending on your conference, the statistically significant number of games played by every team is one, two, three or (in the case of the E8 and a couple others) four.  That's even a worse sample than the number of regional games played.  That's the reason he explained for opening up to secondary criteria and other more subjective processes to differentiate teams or validate the limited statistics.

Ralph Turner

Is that(Mr Kaiser's) interview available on podcast?