\
  The most prestigious law school admissions discussion board in the world.
BackRefresh Options Favorite

ESPN CFB Power Index ranks Alabama #3 above LSU, Georgia, Auburn

http://www.espn.com/college-football/statistics/teamratings ...
DrakeMallard
  12/02/19
Massive drop off after teh T-14. Notre Dame is the GULC of c...
Judas Jones
  12/02/19
this is like the cooley ls rankings for football
waiting for the great leap forward
  12/02/19
...
zoomer
  12/02/19
...
DrakeMallard
  12/02/19
Bama fan here. even I'm like "WTF" at this one
michael doodikoff
  12/02/19
Yeah, they still rank Washington at 18 or something. I don'...
IronMonkey
  12/02/19
...
DrakeMallard
  12/02/19
its a computer ranking right? it takes margin of victory,...
Barack Carcetti
  12/02/19
None of their wins were against good teams, either, though. ...
IronMonkey
  12/02/19
its a computer ranking it has Bama and LSU virtually equa...
Barack Carcetti
  12/02/19
well, but Bama has now lost two games that were "close&...
Rafael Ironside, FIVE TIME ATP Year-End #1
  12/02/19
Sure no one's going to use some random ESPN computer rank...
Barack Carcetti
  12/02/19
How shitty the FPI ranking is
IronMonkey
  12/02/19
...
Rafael Ironside, FIVE TIME ATP Year-End #1
  12/02/19
...
DrakeMallard
  12/02/19
It's a shitty ranking system if if values blowing out shitty...
IronMonkey
  12/02/19
it's intended to be predictive. if, objectively, blowing out...
A Jurisprudence is Performed
  12/02/19
I understand, but why are you assuming that the system is ac...
IronMonkey
  12/02/19
I'm not, it's a weird math nerd black box that I could never...
A Jurisprudence is Performed
  12/02/19
'shitty ranking system' it depends on what you're using i...
Barack Carcetti
  12/02/19
Tua played against LSU. LSU beat Bama. LSU beat Auburn, whic...
Guy Debord
  12/02/19
(ESPN FPI Analyst)
Rafael Ironside, FIVE TIME ATP Year-End #1
  12/02/19
...
DrakeMallard
  12/02/19
...
Hysterical bitchmade deepstate faggot
  12/02/19
Cr a meaningless stat like "H2H" doesn't matter at...
Heteronormative, Cisgendered Bro
  12/02/19
its a computer rank ---- if Tua was healthy, Vegas wou...
Barack Carcetti
  12/02/19
the "favored in a rematch" is such a fucking awful...
Rafael Ironside, FIVE TIME ATP Year-End #1
  12/02/19
Cr let's ignore WHAT ACTUALLY HAPPENED and guess what may ha...
Heteronormative, Cisgendered Bro
  12/02/19
It is also mentally challenging to have to beat a good oppon...
Rafael Ironside, FIVE TIME ATP Year-End #1
  12/02/19
Bama's not getting 2 bites at the apple based on this rankin...
Barack Carcetti
  12/02/19
yes it's still that shocking even as a computer model. (I'm ...
borders
  12/02/19
These are power ratings. Logically, losing a game on a last ...
A Jurisprudence is Performed
  12/02/19
...
spritezero
  12/02/19
That doesn't explain Alabama being ahead of LSU.
Rafael Ironside, FIVE TIME ATP Year-End #1
  12/02/19
...
Heteronormative, Cisgendered Bro
  12/02/19
sure it does. LSU beat Bama by five points and outgained the...
A Jurisprudence is Performed
  12/02/19
let's say they don't score on their last, meaningless posses...
Rafael Ironside, FIVE TIME ATP Year-End #1
  12/02/19
FPI just sees numbers, it doesn't make judgments like that. ...
A Jurisprudence is Performed
  12/02/19
it would be better if ESPN made more of their stuff open, ...
Barack Carcetti
  12/02/19
Lol cmon brother. Now look at game v common opponent Aub...
Heteronormative, Cisgendered Bro
  12/02/19
...
Guy Debord
  12/02/19
I have to agree here. This is pretty hard to argue.
Rafael Ironside, FIVE TIME ATP Year-End #1
  12/02/19
...
DrakeMallard
  12/02/19
imagine the SECCG was LSU-UA instead of LSU-Georgia. Do you...
A Jurisprudence is Performed
  12/02/19
I get what you're saying. It just doesn't make sense in this...
Heteronormative, Cisgendered Bro
  12/02/19
Even if you take the Auburn and LSU games as tossups for Ala...
IronMonkey
  12/02/19
I didn't program the model.
A Jurisprudence is Performed
  12/02/19
But you seemed to be defending it.
IronMonkey
  12/02/19
"this math nerd Nate Silver, "well actually,"...
A Jurisprudence is Performed
  12/02/19
its hard to create a model that handles every situation well...
Barack Carcetti
  12/02/19
jesus fucking christ isn't he? god damnit already
No Paye No Gain
  12/02/19
In a vacuum in a head to head matchup your point makes sense...
borders
  12/02/19
well said
Barack Carcetti
  12/02/19
...
LathamTouchedMe
  12/02/19
sports fans have a tendency to be too results (and thus, luc...
SomeOtherGhost
  12/02/19
Looks like Saban spent the western carolina week learning py...
Woah Cr friend!
  12/02/19
lol
Barack Carcetti
  12/02/19
...
DrakeMallard
  12/02/19
...
spritezero
  12/02/19
...
Harmon Killebrew
  12/02/19
...
zoomer
  12/02/19
Heather Dinich basically does nothing but regurgitate the FP...
Hysterical bitchmade deepstate faggot
  12/02/19
AHH MAH GHEEEERRRRD THA PLAYOFFS ARE SO DARN BAD YOU GUISE! ...
.....;;,,.........;.;.;.;.,;,;,;.;.;,;
  12/02/19
Thought this would be a throwaway ~10 poast thread. Very pl...
DrakeMallard
  12/02/19
We're all chill, fratty 6'4" SEC white bros here
Judas Jones
  12/02/19
If you get into horse racing, you learn that you can't boil ...
Windsor Canadian
  12/02/19
http://autoadmit.com/thread.php?thread_id=4392319&mc=11&...
chandler (retired)
  12/02/19


Poast new message in this thread



Reply Favorite

Date: December 2nd, 2019 11:14 AM
Author: DrakeMallard (Let's get dangerous)

http://www.espn.com/college-football/statistics/teamratings

1 Ohio State, Big Ten 12-0 -- -- -- -- 33.4

2 Clemson, ACC 12-0 -- -- -- -- 31.4

3 Alabama, SEC 10-2 -- -- -- -- 28.6

4 LSU, SEC 12-0 -- -- -- -- 27.6

5 Georgia, SEC 11-1 -- -- -- -- 25.3

6 Auburn, SEC 9-3 -- -- -- -- 21.9

7 Florida, SEC 10-2 -- -- -- -- 21.8

8 Oklahoma, Big 12 11-1 -- -- -- -- 21.4

9 Wisconsin, Big Ten 10-2 -- -- -- -- 21.3

10 Penn State, Big Ten 10-2 -- -- -- -- 21.1

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39201790)



Reply Favorite

Date: December 2nd, 2019 11:17 AM
Author: Judas Jones

Massive drop off after teh T-14. Notre Dame is the GULC of college football.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39201804)



Reply Favorite

Date: December 2nd, 2019 11:18 AM
Author: waiting for the great leap forward (ggtp)

this is like the cooley ls rankings for football

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39201809)



Reply Favorite

Date: December 2nd, 2019 3:28 PM
Author: zoomer (uspo)



(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39203258)



Reply Favorite

Date: December 2nd, 2019 11:26 AM
Author: DrakeMallard (Let's get dangerous)



(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39201844)



Reply Favorite

Date: December 2nd, 2019 11:58 AM
Author: michael doodikoff

Bama fan here. even I'm like "WTF" at this one

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202020)



Reply Favorite

Date: December 2nd, 2019 12:00 PM
Author: IronMonkey

Yeah, they still rank Washington at 18 or something. I don't understand their metrics

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202035)



Reply Favorite

Date: December 2nd, 2019 12:22 PM
Author: DrakeMallard (Let's get dangerous)



(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202143)



Reply Favorite

Date: December 2nd, 2019 12:27 PM
Author: Barack Carcetti

its a computer ranking right?

it takes margin of victory, and probably stuff like drive efficiency into account,

both of their losses were close games, none of their wins were that close

this shouldn't be that shocking

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202165)



Reply Favorite

Date: December 2nd, 2019 12:29 PM
Author: IronMonkey

None of their wins were against good teams, either, though. Compare that to LSU, sitting right behind Alabama in FPI.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202170)



Reply Favorite

Date: December 2nd, 2019 12:38 PM
Author: Barack Carcetti

its a computer ranking

it has Bama and LSU virtually equal

when they played they were pretty close to equal

the computer thinks Bama was somewhat better against the crappy teams they played than LSU was against the (maybe somewhat less) crappy teams LSU played

if there's a critism to be made here, the computer almost certainly isn't factoring in that Tua wouldn't be playing if the teams had a rematch now

whereas Bama with a healthy Tua makes up most of the inputs generating this ranking

(I suspect if Tua was healthy, Vegas would favor Bama in a rematch)

-------

--------

https://www.espn.com/blog/statsinfo/post/_/id/122612/an-inside-look-at-college-fpi



(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202228)



Reply Favorite

Date: December 2nd, 2019 12:40 PM
Author: Rafael Ironside, FIVE TIME ATP Year-End #1

well, but Bama has now lost two games that were "close" (though I'd dispute that the LSU game was that close)

perhaps they can't get it done in crunch time?

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202241)



Reply Favorite

Date: December 2nd, 2019 12:44 PM
Author: Barack Carcetti

Sure

no one's going to use some random ESPN computer ranking to put them in the playoff

nor should they

----

what exactly are we getting MAF about here?

or do we just do that anytime 'ESPN' and 'Alabama' are in a sentence in proximity to each other?

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202266)



Reply Favorite

Date: December 2nd, 2019 12:46 PM
Author: IronMonkey

How shitty the FPI ranking is

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202280)



Reply Favorite

Date: December 2nd, 2019 12:46 PM
Author: Rafael Ironside, FIVE TIME ATP Year-End #1



(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202283)



Reply Favorite

Date: December 2nd, 2019 12:46 PM
Author: DrakeMallard (Let's get dangerous)



(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202287)



Reply Favorite

Date: December 2nd, 2019 12:43 PM
Author: IronMonkey

It's a shitty ranking system if if values blowing out shitty teams and losing to good teams (Alabama) more heavily than blowing out shitty teams and beating multiple good teams without ever losing (LSU).

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202259)



Reply Favorite

Date: December 2nd, 2019 12:45 PM
Author: A Jurisprudence is Performed (Dictated But Not Read)

it's intended to be predictive. if, objectively, blowing out bad teams is a better predictor of future performance than losing closely to good teams, so be it. That's irrelevant in terms of rankings, seedings, etc., which are based on actual results.

remember this is all ultimately based on vegas shit.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202278)



Reply Favorite

Date: December 2nd, 2019 12:46 PM
Author: IronMonkey

I understand, but why are you assuming that the system is actually good at its purpose?

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202288)



Reply Favorite

Date: December 2nd, 2019 12:50 PM
Author: A Jurisprudence is Performed (Dictated But Not Read)

I'm not, it's a weird math nerd black box that I could never understand. I'm just explaining what its purpose is.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202323)



Reply Favorite

Date: December 2nd, 2019 12:48 PM
Author: Barack Carcetti

'shitty ranking system'

it depends on what you're using it for,

if you're using it to seed the playoff, than yeah, it is

-----

if you're using it as a tool as to expect what will happen in future matchups

idk, its almost certainly a better tool than say, the AP rankings

its a rough approximation of the process that goes into setting Vegas lines

if you think its way off

there's money to be made figuring out in exactly what way its off

(or you can bluster loudly on XO, either way)

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202295)



Reply Favorite

Date: December 2nd, 2019 12:45 PM
Author: Guy Debord

Tua played against LSU. LSU beat Bama. LSU beat Auburn, which just beat Bama. LSU just crushed TAMU by a much larger margin than Bama beat TSU. Ranking makes no sense, even if done by a computer.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202272)



Reply Favorite

Date: December 2nd, 2019 12:29 PM
Author: Rafael Ironside, FIVE TIME ATP Year-End #1

(ESPN FPI Analyst)

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202172)



Reply Favorite

Date: December 2nd, 2019 12:36 PM
Author: DrakeMallard (Let's get dangerous)



(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202215)



Reply Favorite

Date: December 2nd, 2019 1:57 PM
Author: Hysterical bitchmade deepstate faggot



(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202806)



Reply Favorite

Date: December 2nd, 2019 12:33 PM
Author: Heteronormative, Cisgendered Bro

Cr a meaningless stat like "H2H" doesn't matter at all when comparing relative rank

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202193)



Reply Favorite

Date: December 2nd, 2019 12:40 PM
Author: Barack Carcetti

its a computer rank

----

if Tua was healthy, Vegas would favor Bama in a rematch against LSU

----

It is important to note what FPI is not -- FPI is not a playoff predictor, and it is not designed to identify the four teams most deserving of making the College Football Playoff. ESPN has other metrics, including Strength of Record, that can be used to identify the most deserving teams.

https://www.espn.com/blog/statsinfo/post/_/id/122612/an-inside-look-at-college-fpi

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202242)



Reply Favorite

Date: December 2nd, 2019 12:47 PM
Author: Rafael Ironside, FIVE TIME ATP Year-End #1

the "favored in a rematch" is such a fucking awful metric

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202289)



Reply Favorite

Date: December 2nd, 2019 12:48 PM
Author: Heteronormative, Cisgendered Bro

Cr let's ignore WHAT ACTUALLY HAPPENED and guess what may happen if they played again!!!



(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202302)



Reply Favorite

Date: December 2nd, 2019 1:02 PM
Author: Rafael Ironside, FIVE TIME ATP Year-End #1

It is also mentally challenging to have to beat a good opponent twice, and so in my mind not representative.

That's why i vehemently disagree with giving teams two bites at the apple in a conference championship game or the playoff. If LSU beats Bama, one of the benefits of that win should be not having to beat Bama again in the conference championship game or the playoff

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202414)



Reply Favorite

Date: December 2nd, 2019 1:13 PM
Author: Barack Carcetti

Bama's not getting 2 bites at the apple based on this ranking

if it were, I'd agree with you

--------

ESPN analytic stuff is so black box, I actually don't have very much confidence that its all that good

but,

"Team A beat Team B"

and

"Team B has more absolute losses than Team A"

therefore

"Team B should be ranked ahead of Team A"

aren't good criticisms of a black box computer ranking

(and honestly, they're not great criticisms of the human based rankings either)

---

---

fwiw

imo better criticisms start here "What goes into the preseason ratings?"

and how they would handle a starting QB getting injured

but whatever

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202513)



Reply Favorite

Date: December 2nd, 2019 1:55 PM
Author: borders

yes it's still that shocking even as a computer model. (I'm a statistician and build shit like this all the time)

The model is clearly not well calibrated / validated, or too algorithmically simplistic, if it has such a disproportionate amount of weight on margin of victory relative to head-to-head and quality of wins / strength of schedule.

If I built a model like this - saw you could have a team like LSU win a head-to-head, remain undefeated, win against the team that gave Alabama its other loss, and have a win on the road against a far more difficult out of conference opponent than Alabama - I would immediately tear it down / re-train it as it's not ready for production / primetime to be put into use.

Maybe I should be more clear - it's shocking that ESPN is publishing this thing, not shocking that their model sucks. It's clearly too reliant on margin of victory and offensive / defensive efficiency ratings, but you have to wonder about your model when you have this kind of fatal flaw of not taking actual wins and head to head matchups into account.

edit: re: discussion above about whether it's a predictive model or a pure "rankings". The fact that it's not a predictive index makes the model look even worse. You could argue that based on the model inputs, it imperfectly makes predictions and okay it actually thinks 'bama would beat LSU in a rematch then so be it.

But the fact it's a pure rankings makes it even more pure shit - it's outputting a ranking that's favoring a team with two losses within the same conference to a team with a very similar but more difficult schedule, that

a) defeated that team in a head-to-head matchup on the road at that team's home stadium

b) defeated the team that gave alabama its second loss

c) won on the road against against a more difficult opponent in out of conference play

d) overall has more impressive wins - as LSU won on the road in Austin, and at Alabama, beat Auburn, beat Florida

e) LSU has defeated 3 teams that *finished* in the top 10/11,

f) Alabama's only win against a ranked opponent is against #24 Texas A&M, their next best win is what?

g) I just manually calculated the margin of victory for the 5 games that LSU and Alabama had in common - LSU +126 vs. Alabama +116,

h) If FPI is a pure ranking, There should be an algorithmic override for the production of final results that checks undefeated / head-to-head / opponents shared in common etc. - maybe at some absurd extreme you could imagine Alabama eviscerating very good opponents they shared in common, then having extremely close loss to auburn / lsu, whereas LSU barely squeaks by everybody

But as is - that ranking is fundamentally retarded and as a pure ranking is obviously flawed - I personally wouldn't publish such an obviously flawed model under my name.

The regular season is almost over after conference championship -

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202794)



Reply Favorite

Date: December 2nd, 2019 12:31 PM
Author: A Jurisprudence is Performed (Dictated But Not Read)

These are power ratings. Logically, losing a game on a last second field goal that hits the upright is meaningless in terms of power ratings. Like, if the kick goes two inches to the right or the left, that doesn't affect how "good" a team is. FPI is specifically designed to measure this.

But the point of playing football is to win games, not to be the "better" team in some Platonic sense. And this math nerd Nate Silver, "well actually," shit takes a lot of the fun and enjoyment out of the game.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202183)



Reply Favorite

Date: December 2nd, 2019 12:32 PM
Author: spritezero



(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202187)



Reply Favorite

Date: December 2nd, 2019 12:32 PM
Author: Rafael Ironside, FIVE TIME ATP Year-End #1

That doesn't explain Alabama being ahead of LSU.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202190)



Reply Favorite

Date: December 2nd, 2019 12:33 PM
Author: Heteronormative, Cisgendered Bro



(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202200)



Reply Favorite

Date: December 2nd, 2019 12:36 PM
Author: A Jurisprudence is Performed (Dictated But Not Read)

sure it does. LSU beat Bama by five points and outgained them by 18 total yards. In FPI terms, that's a tossup, and disappears with the adjustment for home field advantage. In math nerd land, LSU didn't really "beat" Bama in a statistically significant sense.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202213)



Reply Favorite

Date: December 2nd, 2019 12:39 PM
Author: Rafael Ironside, FIVE TIME ATP Year-End #1

let's say they don't score on their last, meaningless possession. still statistically insignificant?

LSU was never in danger of losing that game.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202233)



Reply Favorite

Date: December 2nd, 2019 12:42 PM
Author: A Jurisprudence is Performed (Dictated But Not Read)

FPI just sees numbers, it doesn't make judgments like that. It's only insignificant because Bama didn't recover the onside.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202255)



Reply Favorite

Date: December 2nd, 2019 1:21 PM
Author: Barack Carcetti

it would be better if ESPN made more of their stuff open,

so you could kick the tires on their models in this fashion more,

but that's not a thing that they do



(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202575)



Reply Favorite

Date: December 2nd, 2019 12:44 PM
Author: Heteronormative, Cisgendered Bro

Lol cmon brother.

Now look at game v common opponent Auburn. LSU slaughtered them in yards and other significant categories (time of poss, 1st downs). AND they won that game, which Alabama did not do.

Whatever modeling they are doing is broken when it causes a zero loss team to be ranked lower than a 2-loss team that got beat in H2H and v common opponent. Just no fucking way.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202271)



Reply Favorite

Date: December 2nd, 2019 12:46 PM
Author: Guy Debord



(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202279)



Reply Favorite

Date: December 2nd, 2019 12:46 PM
Author: Rafael Ironside, FIVE TIME ATP Year-End #1

I have to agree here. This is pretty hard to argue.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202282)



Reply Favorite

Date: December 2nd, 2019 12:46 PM
Author: DrakeMallard (Let's get dangerous)



(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202284)



Reply Favorite

Date: December 2nd, 2019 12:49 PM
Author: A Jurisprudence is Performed (Dictated But Not Read)

imagine the SECCG was LSU-UA instead of LSU-Georgia. Do you think the line should automatically be LSU -5 because that's what the score was last time? that's not how these things work.

Put it this way - if App State and Michigan had a rematch the week after their famous 2007 game, Michigan would've been favored by two TDs. App State still won the ballgame.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202309)



Reply Favorite

Date: December 2nd, 2019 12:52 PM
Author: Heteronormative, Cisgendered Bro

I get what you're saying. It just doesn't make sense in this case. They are clearly valuing the wrong things that 99.9% of fans can see with a naked eyeball test. EDIT - especially with tua out

These "systems" are capable of writing bad lines. They do it all the time. If they were to spit out a Bama -2 in your hypothetical SECCG (which lol @ 2-loss Bama being in), I and probably 90% of the betting public would be all over the dog.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202347)



Reply Favorite

Date: December 2nd, 2019 12:45 PM
Author: IronMonkey

Even if you take the Auburn and LSU games as tossups for Alabama, LSU has multiple quality victories over good teams, while Alabama has none (or, at best, two "ties" against good teams). Why would Alabama be ahead of LSU unless the model weighs stupid things too heavily.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202276)



Reply Favorite

Date: December 2nd, 2019 12:46 PM
Author: A Jurisprudence is Performed (Dictated But Not Read)

I didn't program the model.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202286)



Reply Favorite

Date: December 2nd, 2019 12:48 PM
Author: IronMonkey

But you seemed to be defending it.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202298)



Reply Favorite

Date: December 2nd, 2019 12:51 PM
Author: A Jurisprudence is Performed (Dictated But Not Read)

"this math nerd Nate Silver, "well actually," shit takes a lot of the fun and enjoyment out of the game"

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202328)



Reply Favorite

Date: December 2nd, 2019 1:15 PM
Author: Barack Carcetti

its hard to create a model that handles every situation well,

this probably isn't a situation that this model handles well

"all models are wrong, some are useful"

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202534)



Reply Favorite

Date: December 2nd, 2019 3:28 PM
Author: No Paye No Gain

jesus fucking christ isn't he? god damnit already

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39203260)



Reply Favorite

Date: December 2nd, 2019 2:32 PM
Author: borders

In a vacuum in a head to head matchup your point makes sense, especially about the field goal breaking one way or the other.

But a model that doesn't take into account shared opponents, strength of schedule of quality wins etc. is questionable. I can certainly understand offensive / defensive efficiency being inputs into the model - but as you said, the goal of the game is to win, and even if total yards etc. is close, you have LSU defeating Alabama on the road, LSU defeating the opponent that gave Alabama its other loss, far more impressive wins overall, and a higher margin of victory against the teams they both played against I just calculated lsu +126 vs bama + 116 (not including their game against each other).

So maybe we're debating some minor points about weighting factors into the model - but e.g. imagine an extreme case where a team has a monster red zone defense but defense gives up tons of yards, and the converse on offense. this model will rank this team poorly even if they are able to beat everyone.

Your point about statistical significance only makes sense if they're calling it a wash - like saying okay this is inside the confidence interval we're calling it even. It makes no sense to say okay mu_a > mu_b, but was within margin of error so we're ranking mu_b ahead of mu_a,..... they're putting bama ahead, which shows an obvious flaw of e.g. weighing offensive production against crappy teams far out of proportion relative to actual head to head matchups, strength of schedule, and outcomes against shared opponents within the season.

The difference all comes down to ESPN's FPI model having too much importance placed on crushing shitty teams, rather than actually winning games against difficult opponents / shared opponents / head to head.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202977)



Reply Favorite

Date: December 2nd, 2019 1:00 PM
Author: Barack Carcetti

well said

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202407)



Reply Favorite

Date: December 2nd, 2019 5:07 PM
Author: LathamTouchedMe



(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39203769)



Reply Favorite

Date: December 2nd, 2019 9:38 PM
Author: SomeOtherGhost

sports fans have a tendency to be too results (and thus, luck) driven. the "nerds" are a necessary corrective, if you have a brain.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39205304)



Reply Favorite

Date: December 2nd, 2019 12:54 PM
Author: Woah Cr friend!

Looks like Saban spent the western carolina week learning python and r

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202362)



Reply Favorite

Date: December 2nd, 2019 1:13 PM
Author: Barack Carcetti

lol

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202516)



Reply Favorite

Date: December 2nd, 2019 1:32 PM
Author: DrakeMallard (Let's get dangerous)



(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202646)



Reply Favorite

Date: December 2nd, 2019 2:05 PM
Author: spritezero



(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202854)



Reply Favorite

Date: December 2nd, 2019 2:13 PM
Author: Harmon Killebrew



(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202905)



Reply Favorite

Date: December 2nd, 2019 3:28 PM
Author: zoomer (uspo)



(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39203255)



Reply Favorite

Date: December 2nd, 2019 1:57 PM
Author: Hysterical bitchmade deepstate faggot

Heather Dinich basically does nothing but regurgitate the FPI in her commentary. It now makes sense.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39202810)



Reply Favorite

Date: December 2nd, 2019 3:20 PM
Author: .....;;,,.........;.;.;.;.,;,;,;.;.;,;


AHH MAH GHEEEERRRRD THA PLAYOFFS ARE SO DARN BAD YOU GUISE! I WISH THEM THAR COMPEWTURS COULD HELP OUT OLE NICK AND ROLL TIDE LIKE THEM THAR GOOD OLE DAYS.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39203219)



Reply Favorite

Date: December 2nd, 2019 5:14 PM
Author: DrakeMallard (Let's get dangerous)

Thought this would be a throwaway ~10 poast thread. Very pleased with the outperformance

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39203833)



Reply Favorite

Date: December 2nd, 2019 5:24 PM
Author: Judas Jones

We're all chill, fratty 6'4" SEC white bros here

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39203906)



Reply Favorite

Date: December 2nd, 2019 6:27 PM
Author: Windsor Canadian

If you get into horse racing, you learn that you can't boil predicting complicated events like sports into a foolproof ranking system that is always or mostly right.

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39204267)



Reply Favorite

Date: December 2nd, 2019 6:29 PM
Author: chandler (retired)

http://autoadmit.com/thread.php?thread_id=4392319&mc=11&forum_id=2

(http://www.autoadmit.com/thread.php?thread_id=4393275&forum_id=2#39204275)