If you think a single game or a single team completely invalidates a ratings system

Sign up, and you can make all message times appear in your timezone. Sign up

Report problem with this ad

Jan 3, 2025
10:12:01am

Indy Coug Loquacious Lummox

If you think a single game or a single team completely invalidates a ratings system

I think you're going to need a lot more than that to make your case. I've already provided how you validate/debunk the FPI — and it's not through screaming about Bama losing on the road to a 7-6 Vanderbilt team.

I'm not saying the FPI is best system out there — but you need to find a better approach to demonstrate how good/bad it is.

This message has been modified

Originally posted on Jan 3, 2025 at 10:12:01am

Message modified by Indy Coug on Jan 3, 2025 at 10:12:48am

Report problem with this ad

Indy Coug

Bio page: Indy Coug
Joined: Dec 20, 2001
Last login: Jan 4, 2025
Total posts: 190,207 (10,160 FO)

Report problem with this ad

Messages

Author

Time

Great work by Bamacoug in exposing ESPN's FPI rankings:

MHallDidNothingWr0ng

Jan 3, 9:55am

BYU and ASU are both suppressed by bad play in 2023 and low recruiting rankings.

Jan 3, 9:58am

Are you sure that's updated?

RobAdams

Jan 3, 10:00am

Yep... FPI updates nightly. BYU moved up 7 spots because of the Colorado win &

BamaCoug

Jan 3, 10:30am

FPI never valued BYU.

Adrielle

Jan 3, 10:32am

Thank you.

ParentingNotForSissy

Jan 3, 11:23am

If you want to truly expose the FPI, then what you do is take it and the other analytic rankings

Indy Coug

Jan 3, 10:00am

RE: If you want to truly expose the FPI, then what you do is take it and the

Jan 3, 10:01am

I don’t have the interest/energy in doing that.

Indy Coug

Jan 3, 10:02am

RE: I don’t have the interest/energy in doing that.

Accipiter

Jan 3, 1:00pm

And it’s still January and 8 months before the next game…

bjc

Jan 3, 4:51pm

A 9-4 Alabama who lost to Vanderbilt is at #4 in these rankings. What more do

MHallDidNothingWr0ng

Jan 3, 10:06am

If you think a single game or a single team completely invalidates a ratings system

Indy Coug

Jan 3, 10:12am

Colorado is still ranked above BYU despite getting absolutely manhandled in the

MHallDidNothingWr0ng

Jan 3, 10:17am

Ratings after 13-14 games this late into the season are based off a lot more than the outcome of a single game

Indy Coug

Jan 3, 10:20am

Like the stats and math you used last season to "prove" that Jay Hill was not an

MHallDidNothingWr0ng

Jan 3, 10:29am

I never made that argument. I did argue that BYU's defensive results in 2023 were poor

Indy Coug

Jan 3, 10:35am

I just think it's funny that you pick such weird hills to die on. Tuiaki, and

MHallDidNothingWr0ng

Jan 3, 10:39am

So why is Colorado ahead of BYU?

Beggar

Jan 3, 10:35am

You don’t need a replacement model to prove that a bad one is bad.

kcca

Jan 3, 2:44pm

Exactly. There are many games and teams that discredit their stupid metric.

Jan 3, 10:21am

Cherry picking games out on the fringes isn't how you debunk something

Indy Coug

Jan 3, 10:23am

There are too many outliers to take it seriously. Four ACC teams being ahead of

Jan 3, 1:06pm

And what does the empirical data say?

Indy Coug

Jan 3, 1:35pm

The empirical data says that there's lies, (danged) lies, and statistics. And

MHallDidNothingWr0ng

Jan 3, 1:46pm

The FPI (and other analytics) is based on actual, on the field results. So what is the best source of truth for this?

Indy Coug

Jan 3, 1:51pm

BYU beat every WAC team in 1985 except for UTEP

Belboz

Jan 3, 10:54am

RE: BYU beat every WAC team in 1985 except for UTEP

Indy Coug

Jan 3, 10:55am

Ok, so how does that explain Colorado being ahead of us this year?

Jan 3, 12:14pm

My point is that there should be more to a ranking than the outcome of 1 game

Belboz

Jan 3, 1:28pm

I agree with your overarching idea, but what about Colorado makes you think they

here's to hoping

Jan 3, 4:27pm

I don't think Colorado should be ranked higher than BYU. That wasn't my point

Belboz

Jan 4, 11:10am

I believe they have four losses, including one to Vanderbilt.

thirdgencoug

Jan 3, 1:40pm

What about losing to 7-5 Michigan and keeping the same position at 4?

Viking VI

Jan 3, 3:28pm

Thankfully it's not just a single game to invalidate the idea that Alabama is the 4th best team. We have 3 other losses

Spiff

Jan 3, 6:13pm

Plus, an accurate model would get more accurate as the season goes on. Would

Boring

Jan 3, 10:35am

Yeah, this was just more of a "Zoom out/look & FPI model doesn't value Big12...

BamaCoug

Jan 3, 10:51am

“Just” 48% against the spread? Cmon.

TCuz

Jan 3, 11:13am

KS needs to blow out teams if that is the case.

coolblue

Jan 3, 11:30am

Here’s a good website to compare ranking systems

canuckchuck

Jan 3, 12:09pm

This is cool, thanks. I would love to see it break down by conference. e.g. how

here's to hoping

Jan 3, 4:46pm

Woah. That seems like a lot of work

TNT

Jan 3, 2:23pm

See my link below for a website that does that

MyTwoCents

Jan 3, 6:12pm

Not sure at the end of the season we should be picking "best" team instead of

ABCoug

Jan 3, 8:40pm

I’m too dumb to understand everything, but is he suggesting that espn arbitrarily is keeping big XII teams lower?

shimmy

Jan 3, 10:02am

He needs a more persuasive argument than the one he used.

Indy Coug

Jan 3, 10:03am

Two things...

Adrielle

Jan 3, 10:13am

So it’s a bad system or their thumb is on the scale?

shimmy

Jan 3, 11:13am

The system is an algorithm. They created the algorithm.

here's to hoping

Jan 3, 4:50pm

My (very cynical) view of FPI is that it gives ESPN one more talking point...

dut99002

Jan 3, 10:10am

This makes the most sense to me. They don't try to be accurate. In fact,

RobAdams

Jan 3, 10:11am

And that's fine, but it's pretty clear that the committee uses the FPI as well, and if the computer system is flawed,

Adrielle

Jan 3, 10:18am

Yep... they have used it on air for years AS A RANKING (they post the national

BamaCoug

Jan 3, 10:54am

Exactly.

dut99002

Jan 3, 11:13am

Let us know when you link the great work he did.

TCuz

Jan 3, 10:13am

Let us know when you decide to become a decent poster instead of an unfunny

MHallDidNothingWr0ng

Jan 3, 10:19am

911, I’d like to report a dead body. Yes his name is TCuz.

shimmy

Jan 3, 11:13am

Talk about a poster that thinks he's funny but really isn't...

Jan 3, 12:01pm

We'll definitely get in the Big 12 now!

Jeffacka

Jan 3, 10:19am

One major issue is we are using a “predictive” model to assess actual results.

Boring

Jan 3, 10:27am

But the predictions are based on data from the past, so I'm not sure what your point is.

Indy Coug

Jan 3, 10:29am

As an actuary, I think you should be able to understand that not every source of

royal_blue

Jan 3, 12:26pm

I think if P&C actuaries used credit scores for setting car insurance rates....

Indy Coug

Jan 3, 12:32pm

It needs to be based on merit. Period. If teams are receiving a boost because of

royal_blue

Jan 3, 12:50pm

OK you go bet money against Vegas with your "fair" model and someone else can use the FPI (52.8% against the spread)

Indy Coug

Jan 3, 12:56pm

You can keep your FPI and use it to bet all you want, just keep it out of the

royal_blue

Jan 3, 2:28pm

So what should be used? Please provide the top 25 list and explain why

Indy Coug

Jan 3, 2:37pm

Hahahahaha. "Come up with 25 alternatives or I won't change my mind."

royal_blue

Jan 3, 3:24pm

How am I supposed to know if you have a valid point if you can't provide

Indy Coug

Jan 3, 3:31pm

Your standard of predictive accuracy and "beating Vegas" is every bit as

royal_blue

Jan 3, 3:56pm

Prior season data continues to be predictive through the end of the season

bythenumbers

Jan 3, 4:06pm

Ok so that being the case, I would assume that you would be in favor of the NFL

royal_blue

Jan 3, 4:18pm

The NFL doesn’t need this nonsense because it only has 32 teams, 17 games, equal home and home divisional play and

Indy Coug

Jan 3, 4:31pm

Well at least we agree that it's nonsense.

royal_blue

Jan 3, 4:40pm

Rankings and prior season results are derived from on-field performance

bythenumbers

Jan 3, 12:58pm

It's fair because every team starts at the same position ensuring an equal

royal_blue

Jan 3, 2:59pm

It's not racist to rank college football teams

bythenumbers

Jan 3, 3:19pm

It's not an analogy.

royal_blue

Jan 3, 3:28pm

You would prefer we draw names out of a hat?

bythenumbers

Jan 3, 3:35pm

Nice straw man.

royal_blue

Jan 3, 4:11pm

I get what you are saying, however completely uninformed doesn't make sense to

sah22

Jan 3, 6:44pm

Prior season results are extremely predictive

bythenumbers

Jan 3, 12:34pm

This is the problem with college football, it is turning into a pro league

Y'S CAT

Jan 3, 10:52am

It's a pro league already. Time to separate it from university control.

Les IsMore

Jan 3, 8:07pm

I think part of it is BYU purposely rarely blows teams out. We don’t try and run

rowlinthunder

Jan 3, 1:05pm

That’s most of it. People want to blame recruiting rankings, or past results, and all types of other stuff

TCuz

Jan 3, 1:12pm

If BYU wanted a better rating then don't wait until the last 10 seconds to beat 3-9 OSU and 5-7 Utah

Indy Coug

Jan 3, 1:32pm

Yup. All those weeks where people would post those charts where BYU won but it was expected that they lose

TCuz

Jan 3, 1:37pm

25?

ParentingNotForSissy

Jan 3, 1:39pm

Alabama got blown out by Oklahoma.

Jan 3, 1:56pm

Great job Jeff... keep up the good work!

Hill4Heisman

Jan 3, 2:16pm

A huge indictment. This needs to be shared …. as much as possible.

Jan 3, 2:57pm

ESPN is the new "fake news of sports"

VenomFarmer

Jan 3, 3:04pm

Problem with "predictive rankings" is that they largely ignore game results

Hill4Heisman

Jan 3, 3:21pm

If they ignore game results, then you should be able to show that the ratings at the end of the season

Indy Coug

Jan 3, 3:24pm

Love BamaCoug but he's completely wrong on this one

MyTwoCents

Jan 3, 5:59pm

I'll admit i don't really know how FPI works, but betting on a game and ranking

sah22

Jan 3, 6:47pm

If you can most accurately predict who wins games and by what score, then why would that not also be telling you who the

Indy Coug

Jan 3, 7:04pm

First, like I said, I don't know how FPI works. Does whoever have the lower FPI

sah22

Jan 3, 7:50pm

Just read post below after posting. Predicting a spread is better than just

sah22

Jan 3, 7:54pm

They were about 71% in predicting the winner of games

MyTwoCents

Jan 3, 8:23pm

From a ranking posted in a link above it looks like they were 9th, but top 10

sah22

Jan 3, 8:36pm

Not really a computer if subjective human inputs are large part of the equation

ABCoug

Jan 3, 8:52pm

Best potential vs best accomplishments?

ABCoug

Jan 3, 8:49pm

I think this is a good point. I think the weight of current performance should

sah22

Jan 3, 8:56pm

FPI gives the expected point spread against an average team on a neutral field

bythenumbers

Jan 3, 7:16pm

Interesting. Thanks for explaining.

sah22

Jan 3, 7:56pm

Thank you for the link.

ParentingNotForSissy

Jan 3, 7:25pm

It's possible to be fairly accurate overall and still have bias (intentional or

here's to hoping

Jan 4, 12:45am

Very interesting to see the top 5 Big12 teams bunched within 6 spots.

Brandoblueblood

Jan 3, 9:37pm

Recruiting rankings should not have any effect. That is a flawed and subjective exercise

Colorado

Jan 4, 4:25pm

I think this is likely the biggest problem. If the recruiting sites were more

here's to hoping

Jan 4, 5:36pm

Even if flawed, recruiting rankings could still be predictive

bythenumbers

Jan 4, 6:46pm

My guess is recruiting rankings help predict future performance

bythenumbers

Jan 4, 6:36pm

Predicting performance and rankings should be different IMO. I get the use of it

sah22

Jan 4, 8:00pm

The problem is the sample size is too small

bythenumbers

Jan 4, 8:41pm

"I'm not arguing the model should have the final word." This is exactly what I'm

sah22

Jan 4, 8:48pm

Then every ranking system is fundamentally flawed

bythenumbers

Jan 4, 9:12pm

Report problem with this ad

Posting on CougarBoard

In order to post, you will need to either sign up or log in.

Report problem with this ad