D3 Top 25 Fan Poll

Started by usee, October 20, 2010, 04:26:33 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

ExTartanPlayer

Quote from: 02 Warhawk on September 30, 2014, 01:35:35 PM
Quote from: ExTartanPlayer on September 30, 2014, 12:53:58 PM
Quote from: jknezek on September 30, 2014, 11:34:07 AM
I think you see that in some of the responses that say "well I can't move NCC that far down" or "I can't move UWSP that far up." That is the definition of poll bias and is the weakest argument you can make. It is not dependent on the data, it is simply dependent on what the voter thought originally and not wanting to be that "wrong" based on the latest data point. Yes, wrong in quotes since it's a poll and you can't be wrong on an opinion.

I will say that if you were starting the poll from scratch, today, based on what has happened in the last three weeks and nothing to do with history beyond that, there is no way you put NCC at 2-1 above 3-0 UWSP.

Yup.

"I don't think NCC is that bad to be in the 20's, nor is UWSP that good to be in the teens just so they can be ranked ahead of NCC."

I just cannot wrap my head around this kind of logic.  UWSP isn't good enough to be ranked ahead of a team that they just beat?

Since I was the one who said that, I guess I'll try and defend myself here. I suppose I've fallen victim to reviewing college football polls since I was a kid (back when I was a Notre Dame freak), and knowing how the Top 25 usually rounds out from week to week.

I've learned that if a #5 team loses to an un-ranked team, they will almost never drop far enough to be placed behind the team that just beat them. I'm not saying that's right or wrong (however, you can certainly make a strong case why that's wrong), but that's just how these polls usually end up. Judging how the voting went this week in our poll, I'm far from alone in voting that way:

13)   North Central        124      (10, 12, 12, 13, 13, 13, 15, 15, 16, 17)
18)   UW-Stevens Point      70      (13, 14, 15, 16, 18, 19, 20, 24, 25, --)

Also, the D3football.com poll seemed to go in the same direction:

13    North Central (Ill.)    2-1    289    5
17      UW-Stevens Point    3-0    188    --

Oh, you're far from alone in thinking that way, I agree. 

You're just all wrong :)

Being serious, I know where you got the idea, and you're absolutely right that's the way that major college polls have operated.  It's a phenomenon known in the social sciences as "anchoring-and-adjustment" - people use the prior beliefs as an "anchor" and then "adjust" rankings based on new results - but they tend to adjust insufficiently, resulting in the final answer being closer to the anchor than it would have been if they were starting from scratch.
I was small but made up for it by being slow...

http://athletics.cmu.edu/sports/fball/2011-12/releases/20120629a4jaxa

ExTartanPlayer

Quote from: wally_wabash on September 30, 2014, 01:42:02 PM
The requirement for Stevens Point to be ranked ahead of North Central today should not include them having to beat Whitewater later on or North Central losing again later on.  Might future results lead us to rank North Central higher down the road?  Of course they might.  But as of this minute there isn't a single logical reason for North Central to be ranked ahead of Stevens Point other than "meh, I'm not buying it". 

Yup
I was small but made up for it by being slow...

http://athletics.cmu.edu/sports/fball/2011-12/releases/20120629a4jaxa

02 Warhawk

#1292
Quote from: BashDad on September 30, 2014, 01:43:23 PM
Well, that's dumb. You're ranking teams according to how you're used to seeing polls adjust whatever team in whatever year after whatever game? That's dumb. Just rank the teams.

I did. With the loss I still think NCC is a top 15 team...while I don't think UWSP is, yet.

At least I wasn't the one who left UWSP off the ballot completely. I'm not that out of touch. ;D

02 Warhawk

#1293
Quote from: wally_wabash on September 30, 2014, 01:42:02 PM
The requirement for Stevens Point to be ranked ahead of North Central today should not include them having to beat Whitewater later on or North Central losing again later on.  Might future results lead us to rank North Central higher down the road?  Of course they might.  But as of this minute there isn't a single logical reason for North Central to be ranked ahead of Stevens Point other than "meh, I'm not buying it". 

But yet they are.

BashDad


ExTartanPlayer

Quote from: USee on September 30, 2014, 01:27:39 PM
Quote from: wally_wabash on September 30, 2014, 11:12:36 AM
Quote from: smedindy on September 29, 2014, 10:18:33 PM
The serious question is if you, as a voter, still think NC can beat UWSP three out of the next five?

This is kind of a cop out, isn't it?  Don't we use this who would win the most out of seven or ten games to more or less dismiss an actual game result that doesn't jive with our preconceptions?

While this is true for the most part I think we also need to remember teams evolve and change, sometimes dramatically, through the course of a year so results in weeks 1-4 may be entirely different  later in the season for a number of reasons. We have some data on this as we occasionally see teams play each other again in the playoffs. In 2002 Wheaton lost @Alma by 7 to start the season, ran the table and ended up @Alma in the playoffs, beating them by 28. Mt Union beat Capital by 20+ in the regular season in 2005 and had to come from behind to win by 3 in the playoffs. NCC lost to Concordia, WI by 6 to start the 2006 season and beat them by 29 in the playoffs, etc. etc. And of course we find teams like the 2008 Wheaton and 2011 St John Fisher who lose to weak teams during the season and become world beaters in the playoffs, making it to the semi's despite 2 regular season losses. Mt Union won the title in 2005 after losing at home mid-season to ONU.

So it's not necessarily misplaced conjecture to try and anticipate how teams change over the course of a season and how that would affect repeat results.  We have to be careful of absolutes. This rankings game is fun but it is part art part science.

I want to give a nod to this post while also disagreeing with the conclusion.

I'll concede the point that, indeed, teams change and evolve over the course of a season. 

I disagree that we should make speculation about that part of the rankings.

I'd rather rank the teams based on what's actually happened, then correct the rankings later when we have results to demonstrate that a team has evolved for the better or changed for the worse.

If time proves that NCC is better than UWSP and that result was a fluke, or a team not fully developed, fine.  That will be reflected down the line.  But I don't get why today's rankings should include that speculation.
I was small but made up for it by being slow...

http://athletics.cmu.edu/sports/fball/2011-12/releases/20120629a4jaxa

USee

Quote from: ExTartanPlayer on September 30, 2014, 02:02:29 PM
Quote from: USee on September 30, 2014, 01:27:39 PM
Quote from: wally_wabash on September 30, 2014, 11:12:36 AM
Quote from: smedindy on September 29, 2014, 10:18:33 PM
The serious question is if you, as a voter, still think NC can beat UWSP three out of the next five?

This is kind of a cop out, isn't it?  Don't we use this who would win the most out of seven or ten games to more or less dismiss an actual game result that doesn't jive with our preconceptions?

While this is true for the most part I think we also need to remember teams evolve and change, sometimes dramatically, through the course of a year so results in weeks 1-4 may be entirely different  later in the season for a number of reasons. We have some data on this as we occasionally see teams play each other again in the playoffs. In 2002 Wheaton lost @Alma by 7 to start the season, ran the table and ended up @Alma in the playoffs, beating them by 28. Mt Union beat Capital by 20+ in the regular season in 2005 and had to come from behind to win by 3 in the playoffs. NCC lost to Concordia, WI by 6 to start the 2006 season and beat them by 29 in the playoffs, etc. etc. And of course we find teams like the 2008 Wheaton and 2011 St John Fisher who lose to weak teams during the season and become world beaters in the playoffs, making it to the semi's despite 2 regular season losses. Mt Union won the title in 2005 after losing at home mid-season to ONU.

So it's not necessarily misplaced conjecture to try and anticipate how teams change over the course of a season and how that would affect repeat results.  We have to be careful of absolutes. This rankings game is fun but it is part art part science.

I want to give a nod to this post while also disagreeing with the conclusion.

I'll concede the point that, indeed, teams change and evolve over the course of a season. 

I disagree that we should make speculation about that part of the rankings.

I'd rather rank the teams based on what's actually happened, then correct the rankings later when we have results to demonstrate that a team has evolved for the better or changed for the worse.

If time proves that NCC is better than UWSP and that result was a fluke, or a team not fully developed, fine.  That will be reflected down the line.  But I don't get why today's rankings should include that speculation.

Today's might not but you will eventually be faced with the fact that NCC pounded UWP and lost to UWSP. What if UWP beats UWW? Then what? right now we have recent data and I agree the polls should reflect that. As weeks transpire the data becomes complex and at that point teams may indeed be different. Ranked teams lose to unranked teams and then beat more highly ranked teams. That happens. Right now we have clear data. In a couple of weeks it will almost decidedly be less clear.

USee

I actually digressed from my original point which was I don't think its wrong to project a team winning a rematch in the rankings. It actually happens.

ExTartanPlayer

Quote from: USee on September 30, 2014, 02:07:06 PM
Quote from: ExTartanPlayer on September 30, 2014, 02:02:29 PM
Quote from: USee on September 30, 2014, 01:27:39 PM
Quote from: wally_wabash on September 30, 2014, 11:12:36 AM
Quote from: smedindy on September 29, 2014, 10:18:33 PM
The serious question is if you, as a voter, still think NC can beat UWSP three out of the next five?

This is kind of a cop out, isn't it?  Don't we use this who would win the most out of seven or ten games to more or less dismiss an actual game result that doesn't jive with our preconceptions?

While this is true for the most part I think we also need to remember teams evolve and change, sometimes dramatically, through the course of a year so results in weeks 1-4 may be entirely different  later in the season for a number of reasons. We have some data on this as we occasionally see teams play each other again in the playoffs. In 2002 Wheaton lost @Alma by 7 to start the season, ran the table and ended up @Alma in the playoffs, beating them by 28. Mt Union beat Capital by 20+ in the regular season in 2005 and had to come from behind to win by 3 in the playoffs. NCC lost to Concordia, WI by 6 to start the 2006 season and beat them by 29 in the playoffs, etc. etc. And of course we find teams like the 2008 Wheaton and 2011 St John Fisher who lose to weak teams during the season and become world beaters in the playoffs, making it to the semi's despite 2 regular season losses. Mt Union won the title in 2005 after losing at home mid-season to ONU.

So it's not necessarily misplaced conjecture to try and anticipate how teams change over the course of a season and how that would affect repeat results.  We have to be careful of absolutes. This rankings game is fun but it is part art part science.

I want to give a nod to this post while also disagreeing with the conclusion.

I'll concede the point that, indeed, teams change and evolve over the course of a season. 

I disagree that we should make speculation about that part of the rankings.

I'd rather rank the teams based on what's actually happened, then correct the rankings later when we have results to demonstrate that a team has evolved for the better or changed for the worse.

If time proves that NCC is better than UWSP and that result was a fluke, or a team not fully developed, fine.  That will be reflected down the line.  But I don't get why today's rankings should include that speculation.

Today's might not but you will eventually be faced with the fact that NCC pounded UWP and lost to UWSP. What if UWP beats UWW? Then what? right now we have recent data and I agree the polls should reflect that. As weeks transpire the data becomes complex and at that point teams may indeed be different. Ranked teams lose to unranked teams and then beat more highly ranked teams. That happens. Right now we have clear data. In a couple of weeks it will almost decidedly be less clear.

Exactly!  We're saying the same thing!  In a couple of weeks, if UWP beats UWSP, then we can try to untangle which of the teams is best.  But right now we do have clear data.  UWSP beat NCC.  NCC beat UWP.  For now, UWSP>NCC>UWP, until there is an actual data point contradicting that.  That may well occur in a few weeks but it has not happened yet, and setting up the rankings by guessing such a result is a fool's errand.  Actually, all rankings are probably fool's errands.
I was small but made up for it by being slow...

http://athletics.cmu.edu/sports/fball/2011-12/releases/20120629a4jaxa

ExTartanPlayer

Ironically enough, today's SI.com Power Rankings for D1 address this very topic we are bandying about:

http://www.si.com/college-football/2014/09/30/college-football-power-rankings

Teams that aren't playing like top 25 teams don't belong there based on their preseason rankings alone. That's why there's a new No. 1 once again and why so many teams have shuffled. As SI.com's Andy Staples wrote in his Punt, Pass & Pork this week, "We may still have the wrong idea, but five weeks of data have provided a few hints."
I was small but made up for it by being slow...

http://athletics.cmu.edu/sports/fball/2011-12/releases/20120629a4jaxa

02 Warhawk

Quote from: BashDad on September 30, 2014, 01:59:14 PM
Because of dumb people.

Feel free to submit a poll then o' wise one.

02 Warhawk

Quote from: ExTartanPlayer on September 30, 2014, 01:47:08 PM
Quote from: 02 Warhawk on September 30, 2014, 01:35:35 PM
Quote from: ExTartanPlayer on September 30, 2014, 12:53:58 PM
Quote from: jknezek on September 30, 2014, 11:34:07 AM
I think you see that in some of the responses that say "well I can't move NCC that far down" or "I can't move UWSP that far up." That is the definition of poll bias and is the weakest argument you can make. It is not dependent on the data, it is simply dependent on what the voter thought originally and not wanting to be that "wrong" based on the latest data point. Yes, wrong in quotes since it's a poll and you can't be wrong on an opinion.

I will say that if you were starting the poll from scratch, today, based on what has happened in the last three weeks and nothing to do with history beyond that, there is no way you put NCC at 2-1 above 3-0 UWSP.

Yup.

"I don't think NCC is that bad to be in the 20's, nor is UWSP that good to be in the teens just so they can be ranked ahead of NCC."

I just cannot wrap my head around this kind of logic.  UWSP isn't good enough to be ranked ahead of a team that they just beat?

Since I was the one who said that, I guess I'll try and defend myself here. I suppose I've fallen victim to reviewing college football polls since I was a kid (back when I was a Notre Dame freak), and knowing how the Top 25 usually rounds out from week to week.

I've learned that if a #5 team loses to an un-ranked team, they will almost never drop far enough to be placed behind the team that just beat them. I'm not saying that's right or wrong (however, you can certainly make a strong case why that's wrong), but that's just how these polls usually end up. Judging how the voting went this week in our poll, I'm far from alone in voting that way:

13)   North Central        124      (10, 12, 12, 13, 13, 13, 15, 15, 16, 17)
18)   UW-Stevens Point      70      (13, 14, 15, 16, 18, 19, 20, 24, 25, --)

Also, the D3football.com poll seemed to go in the same direction:

13    North Central (Ill.)    2-1    289    5
17      UW-Stevens Point    3-0    188    --

Oh, you're far from alone in thinking that way, I agree. 

You're just all wrong :)

Being serious, I know where you got the idea, and you're absolutely right that's the way that major college polls have operated.  It's a phenomenon known in the social sciences as "anchoring-and-adjustment" - people use the prior beliefs as an "anchor" and then "adjust" rankings based on new results - but they tend to adjust insufficiently, resulting in the final answer being closer to the anchor than it would have been if they were starting from scratch.

Actually that's exactly what I do. I would consider re-evaluating my philosophy if my picks were totally titling our poll one way or the other. However, my picks are pretty consistent to what our composite results are.

I don't think one way is better than another, it's just a matter of opinion and interpretation.

jknezek

Quote from: 02 Warhawk on September 30, 2014, 02:35:55 PM

I don't think one way is better than another, it's just a matter of opinion and interpretation.

You can have an opinion and interpretation about anything. It's just difficult to defend one when the data says you are wrong. UWSP was better than NCC last week. That's the data point. UWSP is 3-0, no blemishes yet to diminish the win. In order to put NCC ahead of UWSP you have to ignore the result on the field, the most important piece of data, and the only relevant ancillary data we have this year (overall record).

There are a lot of people here in Alabama that would like to crown Auburn the National Champion last year. It's easy to do if you ignore the most important piece of data...

smedindy

Quote from: wally_wabash on September 30, 2014, 11:12:36 AM
Quote from: smedindy on September 29, 2014, 10:18:33 PM
The serious question is if you, as a voter, still think NC can beat UWSP three out of the next five?

This is kind of a cop out, isn't it?  Don't we use this who would win the most out of seven or ten games to more or less dismiss an actual game result that doesn't jive with our preconceptions?

NO! It's not a cop out. Weird results and outliers would pollute our rankings. Say Kenyon upsets Wittenberg. We'd never rank Kenyon ahead of Wittenberg.
Wabash Always Fights!

jknezek

Quote from: smedindy on September 30, 2014, 02:42:46 PM
Quote from: wally_wabash on September 30, 2014, 11:12:36 AM
Quote from: smedindy on September 29, 2014, 10:18:33 PM
The serious question is if you, as a voter, still think NC can beat UWSP three out of the next five?

This is kind of a cop out, isn't it?  Don't we use this who would win the most out of seven or ten games to more or less dismiss an actual game result that doesn't jive with our preconceptions?

NO! It's not a cop out. Weird results and outliers would pollute our rankings. Say Kenyon upsets Wittenberg. We'd never rank Kenyon ahead of Wittenberg.

Ridiculous. Kenyon has mitigating data in terms of other losses. UWSP is undefeated. There isn't any mitigating data. You just want to ignore the result.