11/23/2009 Final Regular Season Gridiron Power Index (GPI), Villanova No. 1

[CSN-info]

http://www.championshipsubdivisionnews.com/skins/andreas_01/img/GPI.JPG

11/23/2009 Final Regular Season Gridiron Power Index (GPI), Villanova No. 1
College Sporting News

The Gridiron Power Index (GPI), the index ranking for the NCAA Division I FCS and a top indicator of at-large playoff selection finishes the regular season with Villanova in the top spot.

The Colonial Athletic Association, the largest league in the FCS has seven teams in the top 25; the Big Sky Conference has five; the Missouri Valley Football and Southern Conferences have three each; the Ohio Valley and Southland Conferences have two each; and the Big South and Mid-Eastern Athletic Conferences plus the Ivy League have one each.

(Games through 11/22/09)

The Final 2009 GPI will be released once all the college football games have been played, in January 2010.

11/23/2009 GPI Top 25

1. Villanova (1.38)
2. Richmond (2.75)
3T. Montana (3.38)
3T. S Illinois (3.38)
5. William & Mary (4.50)
6. Appalachian St (6.00)
7. Elon (7.63)
8. New Hampshire (9.63)
9. S Dakota St (10.13)
10. E Washington (10.38)
11. S Carolina St (13.13)
12. Weber St (13.25)
13. Jacksonville St (14.13)
14T. Northern Iowa (14.25)
14T. McNeese St (14.25)
16. SF Austin (15.88)
17. James Madison (17.63)
18. Delaware (18.13)
19. Montana St (20.63)
20. Penn (21.38)
21. Liberty (22.25)
22. N Arizona (23.25)
23. Furman (24.00)
24. Massachusetts (24.75)
25. E Illinois (24.88)

Full GPI Detail:
http://www.collegesportingnews.com/stats/writer/GPI/20091123gpi.html

Conference Ranking:
Rank, League, Total Average

1. Colonial Athletic Association (25.40)
2. Big Sky Conference (30.29)
3. Southern Conference (31.15)
4. Great West Conference (31.72)
5. Missouri Valley Football Conference (33.40)
6. Southland Conference (40.50)
7. Ohio Valley Conference (49.95)
8. Patriot League (50.70)
9. Big South Conference (51.40)
10. Ivy League (51.75)
11. Mid-Eastern Athletic Conference (60.14)
12. Southwestern Athletic Conference (64.23)
13. Northeast Conference (66.22)
14. Pioneer Football League (73.40)
15. Independents (75.21)

2009 D-I PLAYOFF AT-LARGE GPI INDICATOR

The NCAA committee selected the GPI indicated teams for the D-I football playoffs. UNI tied for the last spot but the committee chose McNeese State instead.

2. Richmond (2.75)
5. William & Mary (4.50)
7. Elon (7.63)
8. New Hampshire (9.63)
9. S Dakota St (10.13)
10. E Washington (10.38)
12. Weber St (13.25)
14T. McNeese St (14.25)
14T. Northern Iowa (14.25)

Automatic Qualifiers:
1. Villanova (1.38)
3T. Montana (3.38)
3T. S Illinois (3.38)
6. Appalachian St (6.00)
11. S Carolina St (13.13)
16. SF Austin (15.88)
25. E Illinois (24.88)
30. Holy Cross (27.75)

Read more ... (http://www.championshipsubdivisionnews.com/index.php/2009/11/23/11-23-2009-final-regular-season-gridiron-1?blog=5)

[CSN-info]

At-large indicators:

GPI
2. Richmond
5. William & Mary
7. Elon
8. New Hampshire
9. S Dakota St
10. E Washington
12. Weber St
14T. McNeese St

AGS Poll
4. Richmond
6. William & Mary
8. McNeese St.
9. New Hampshire
10. Elon
11. South Dakota St.
14. Eastern Washington
15. Weber St.

TSN Poll
4. Richmond
6. William & Mary
8. McNeese St.
9. Elon
10. New Hampshire
11. South Dakota St.
13. Eastern Washington
15. Weber St.

FCP
4. Richmond
6. William & Mary
8. McNeese St.
9. Elon
10. New Hampshire
11. South Dakota St.
14. Eastern Washington
15. Northern Iowa

[CSN-info]

Flawless? The GPI?

[89Hen]

Conference Ranking:[/B]
Rank, League, Total Average

1. Colonial Athletic Association (25.40)
2. Big Sky Conference (30.29)
3. Southern Conference (31.15)
4. Great West Conference (31.72)
Great West finally dropped closer to where it should be.

[89Hen]

At-large indicators:

GPI
2. Richmond
5. William & Mary
7. Elon
8. New Hampshire
9. S Dakota St
10. E Washington
12. Weber St
14T. McNeese St

AGS Poll
4. Richmond
6. William & Mary
8. McNeese St.
9. New Hampshire
10. Elon
11. South Dakota St.
14. Eastern Washington
15. Weber St.
So once again the GPI doesn't out perform the AGS. :)

And actually since the GPI had a tie for last spot.... xwhistlex

[SumItUp]

Great West finally dropped closer to where it should be.

If each conference was graded based on their top 5 teams , the Great West would be slotted after the Ohio Valley Conference and before the Patriot League. I think that is a more accurate assessment.

[UMass922]

So once again the GPI doesn't out perform the AGS. :)

And actually since the GPI had a tie for last spot.... xwhistlex

Yeah, the GPI should only get partial credit--if even that. Having McNeese as the last team in is absurd. That's a McNeese team that was 8-1 against FCS with a win at Appalachian St. And the GPI thinks a 7-3 (against FCS) UNI team with no quality wins would have been just as good a pick for the last spot? I'd have to score this one as a victory for AGS over the GPI.

[CSN-info]

Human polls have been dragging down the GPI all year, the greatest disparity in recent memory. No poll outperformed the GPI this year as usual (it has only happened once since 1998).

[89Hen]

Human polls have been dragging down the GPI all year, the greatest disparity in recent memory. No poll outperformed the GPI this year as usual (it has only happened once since 1998).
xthumbsdownx The AGS did outperform it since UNI was tied with McNeese. The GPI has NEVER outperformed the AGS. The reverse can't be said.

[CSN-info]

The AGS did outperform it since UNI was tied with McNeese. The GPI has NEVER outperformed the AGS. The reverse can't be said.

We enjoy being the only national media to include the six year old AGS Poll in a significant way in our GPI. We do so because we believe it is the best poll in the nation. However, your statement intending to diminish the GPI is misleading. This year the GPI did indicate the 8 at-large teams that the selection committee picked. Only once in the 12 year history of the GPI was it outperformed. The goal of the GPI is not to outperform all others, it is to be an accurate indicator of at-large selection. It is not a competition.

[Skjellyfetti]

Socon will overtake Big Sky in final GPI.

[UMass922]

This year the GPI did indicate the 8 at-large teams that the selection committee picked.

No it didn't--it failed to distinguish the last team in from the first team out.

[Native]

People argue about the merits and flaws of this poll or that one, this rating or that rating. Many make arguments and judgments based only on their own small pool of personal experiences and abundance of personal pride, and cannot embrace the information or use the imagination required to take an honest and objective look at the entire big picture.

The truth is that there is no such thing as fair, and there are no perfect ranking criteria or selection criteria.

FCS win-loss records alone would include weak teams that would be hammered in the first round of the playoffs by "mediocre" teams with ambitious schedules from strong conferences but a poorer winning record. Head-to-head comparisons are helpful but there are not enough of them and they cannot predict how teams grow and progress over the course of the season - or fail to do so. Computers cannot (yet) factor how injuries will affect play, and they cannot account for the emotions or the intricate match-up asymmetries unique to each and every game.

Overall, the Gridiron Power Index has done a remarkable job for completeness and fairness. We are all better off for it, as one of the tools we rely on to try to make a difficult selection process as fair, open and objectives as it can be.

[GannonFan]

We enjoy being the only national media to include the six year old AGS Poll in a significant way in our GPI. We do so because we believe it is the best poll in the nation. However, your statement intending to diminish the GPI is misleading. This year the GPI did indicate the 8 at-large teams that the selection committee picked. Only once in the 12 year history of the GPI was it outperformed. The goal of the GPI is not to outperform all others, it is to be an accurate indicator of at-large selection. It is not a competition.

Come on, if it's not a competition then why the comment that the GPI is "...a top indicator..." - that clearly implies a ranking versus other indicators.

Like Hen said, the GPI has never outperformed the AGS poll and even this year the AGS poll correctly differentiated the final spot in the playoffs while the GPI left that question unanswered. And heck, this was probably one of the easiest years to pick the 8 at larges and the GPI still waffled in that regard. When it comes down to it, most of the members here, and the AGS poll, clearly have outperformed the GPI year after year. The AGS poll - still undefeated!!! xhurrayx

[UNH Fanboi]

Come on, if it's not a competition then why the comment that the GPI is "...a top indicator..." - that clearly implies a ranking versus other indicators.

Like Hen said, the GPI has never outperformed the AGS poll and even this year the AGS poll correctly differentiated the final spot in the playoffs while the GPI left that question unanswered. And heck, this was probably one of the easiest years to pick the 8 at larges and the GPI still waffled in that regard. When it comes down to it, most of the members here, and the AGS poll, clearly have outperformed the GPI year after year. The AGS poll - still undefeated!!! xhurrayx

Correctly picking the at larges does not necessarily prove that a poll is superior because sometimes the selection committee gets things wrong. Were polls that had W&M ahead of Maine last year flawed?

[89Hen]

your statement intending to diminish the GPI is misleading.
How so? The AGS was more accurate as it did not have a tie for the last spot and has a better record over it's 6 years of existance at predicting the at large field. My problem isn't really with the GPI itself, it's with the crappy computer models it uses. xpeacex

[APPS]

UD fans have beatdown denial. Win and you are in, one D-I championship 6 years ago is all that UD fans bring.

No DSU this year UD. No playoffs. Come on back now, y'hear? :p

[wideright82]

UD fans have beatdown denial. Win and you are in, one D-I championship 6 years ago is all that UD fans bring.

No DSU this year UD. No playoffs. Come on back now, y'hear? :p

App State fans have trouble connecting arguments to conclusions.




What exactly has gone on in this thread, that makes you think ANY UD fan wants UD in the playoffs?xoopsx

[89Hen]

UD fans have beatdown denial. Win and you are in, one D-I championship 6 years ago is all that UD fans bring.

No DSU this year UD. No playoffs. Come on back now, y'hear? :p
http://www.youtube.com/watch?v=0MRmxfLuNto

[UMass922]

Correctly picking the at larges does not necessarily prove that a poll is superior because sometimes the selection committee gets things wrong. Were polls that had W&M ahead of Maine last year flawed?

Flawed as an indicator of which teams got selected as at-larges? Yes, they were. What's being debated here is not whether or not the committee selects the right teams; it's about which ranking has been a better indicator of which teams actually get selected. The point is simply that the AGS poll has a better record of indicating those teams than the GPI does.

[mcveyrl]

So is it safe to say that we were 5 yards (a TD against UR) from getting in the field??

[19Duke97]

The GPI is just another tool in evaluating teams, especially since it removes emotion from the picture. Is it not perfect, but neither are the polls (insert Coaches poll). Taking both into account provides a decent picture of reality.

[GannonFan]

The GPI is just another tool in evaluating teams, especially since it removes emotion from the picture. Is it not perfect, but neither are the polls (insert Coaches poll). Taking both into account provides a decent picture of reality.

Or, as we are arguing here, the GPI takes the best indicator, that is the AGS Poll, and makes it less accurate by dragging it down with questionable computer models that are woefully short on a proper sample size to be statistically relevant and by saddling it with inferior polls from other outfits. That's why the AGS Poll has been supreme - the GPI hobbles itself by diluting the AGS Poll. xnodx

[MacThor]

Well, the inferior GPI had UR ranked higher than AGS going into the playoffs last year.

[JMUNJ08]

So is it safe to say that we were 5 yards (a TD against UR) from getting in the field??

Can only dream now. Who would they have picked? 7-4 Weber or us? I think we become this years W&M. Our losses hurt too much.

Edit: And then ASU would be the 4 seed and JMU would be heading there if we made it...

[CSN-info]

the GPI hobbles itself by diluting the AGS Poll.

Actually the polls have been off by quite a bit this year compared to computers so it may be more correct to say the AGS poll is hobbling the GPI. xnodx


School AGS Comp. Difference
Colgate 21 41.2 -20.2
Prairie View 17 35.8 -18.8
Holy Cross 18 33.6 -15.6
McNeese St 8 18 -10
S Carolina St 7 16.8 -9.8
E Illinois 19 28.8 -9.8
James Madison 25 12.8 12.2

Positive Number = under rank by AGS
Negative Number = over rank by AGS

[Native]

Actually the polls have been off by quite a bit this year compared to computers so it may be more correct to say the AGS poll is hobbling the GPI. xnodx


School AGS Comp. Difference
Colgate 21 41.2 -20.2
Prairie View 17 35.8 -18.8
Holy Cross 18 33.6 -15.6
McNeese St 8 18 -10
S Carolina St 7 16.8 -9.8
E Illinois 19 28.8 -9.8
James Madison 25 12.8 12.2

Positive Number = under rank by AGS
Negative Number = over rank by AGS


And the playoffs will prove the accuracy of the GPI.

[Native]

...woefully short on a proper sample size to be statistically relevant ...

Woefully short compared to what?!?? The morons that vote in most polls?

[AshevilleApp]

3. Southern Conference (31.15)


Whoa, I thought we played in a crappy conference since we didn't play anyone, are untested and do not have a significant win

[MacThor]

3. Southern Conference (31.15)


Whoa, I thought we played in a crappy conference since we didn't play anyone, are untested and do not have a significant win

xoopsxxdeadhorsex

[89Hen]

Actually the polls have been off by quite a bit this year compared to computers so it may be more correct to say the AGS poll is hobbling the GPI. xnodx


School AGS Comp. Difference
Colgate 21 41.2 -20.2
Prairie View 17 35.8 -18.8
Holy Cross 18 33.6 -15.6
McNeese St 8 18 -10
S Carolina St 7 16.8 -9.8
E Illinois 19 28.8 -9.8
James Madison 25 12.8 12.2

Positive Number = under rank by AGS
Negative Number = over rank by AGS

NOW who's COMPLETELY misleading?....

I could just as easliy say "the computers have been off by quite a bit this year compared to polls"

xthumbsdownxxnonoxxnonoxxnonox

[wideright82]

NOW who's COMPLETELY misleading?....

I could just as easliy say "the computers have been off by quite a bit this year compared to polls"

xthumbsdownxxnonoxxnonoxxnonox

Let me ask you something, Hen.


Are you arguing with CSN-Info about whether the AGS Poll or the GPI is better?


You argue with Colonel Sanders about whether Grilled KFC or Fried KFC is better, too?

[GannonFan]

Woefully short compared to what?!?? The morons that vote in most polls?

Woefully short to what would be considered enough data in the real world to use the model to make an accurate prediction. Only in college football would anyone even suggest these computer models are anywhere near statistically viable with the data they have.

[89Hen]

Let me ask you something, Hen.


Are you arguing with CSN-Info about whether the AGS Poll or the GPI is better?


You argue with Colonel Sanders about whether Grilled KFC or Fried KFC is better, too?
xeyebrowx The AGS is better. There's really no debating it.

[SumItUp]

xeyebrowx The AGS is better. There's really no debating it.

89Hen taking his ball and going home. xlolx

[Native]

Woefully short to what would be considered enough data in the real world to use the model to make an accurate prediction. Only in college football would anyone even suggest these computer models are anywhere near statistically viable with the data they have.

All it takes is a hundred or fewer data points for a valid statistical analysis in the real world. To the extent that scores, margins and schedules are valid data points, business and political sample sizes are a MUCH MUCH smaller portion of the available universe of data than are the sports computer models. xeyebrowx

Anyway, why argue? Some of the playoff match ups may give us a chance to get real data on the accuracy of AGS vs the GPI.

But even if, for example, SIU (GPI #3, AGS #2) beats Richmond (GPI#2, AGS #4), does it really prove that the AGS is superior to the GPI, or vica versa, since the teams are only two spots apart, at most? There is not all that much difference between the AGS and the GPI. The ONLY two teams in the playoffs for which the ranking is more than a couple of places different are Elon (#7 GPI, #10 AGS) and Eastern Washington (#10 GPI, #14 AGS).

Can't we all just get alooooong? xnodx

[AshevilleApp]

xoopsxxdeadhorsex

Whoa, I thought this was the dead horse beating forum.....my bad

[89Hen]

All it takes is a hundred or fewer data points for a valid statistical analysis in the real world.
Correct. Only problem is, there aren't a hundred data points comparing teams across conferences. With most teams only having 3 OOC games and most teams using one to play a I-A and another to either play a DII or non-schollie... there are very few data points. How do you compare a Weber State to a W&M? It's like playing the Kevin Bacon game. Thank God for the playoffs. xsmiley_wix

[JohnStOnge]

Yeah, the GPI should only get partial credit--if even that. Having McNeese as the last team in is absurd. That's a McNeese team that was 8-1 against FCS with a win at Appalachian St. And the GPI thinks a 7-3 (against FCS) UNI team with no quality wins would have been just as good a pick for the last spot? I'd have to score this one as a victory for AGS over the GPI.

The GPI is reported to the nearest hundreth. I'm guessing that if they go ahead and do whatever they do and consider all the digits there's a good chance that there wouldn't be an exact tie between McNeese and UNI. Wonder how that would turn out. I'd calculate it but I don't know the methodology.