Sagarin Conference Rankings week 3

[nddulac]

I think I read he did use a least squares method for the older data but for current years (after he really started ranking current teams) he went to a Bayesian method. Not sure when he went to Bayesian models but I do think you're correct about his least squares for really older data.

It may have coincided with his inclusion of his ELO-CHESS method that theBCS wanted when they mandated that the ratings systems remove margin of victory from their formulae.

 

[KansasSooner]

The biggest problem is the size of the data set in that so few games are played it is really hard to get a good confidence level on a sample of just 12 games, or 13 in some cases. Yeah you have a large number of teams (128 and counting) but only a few data points per team. I have really thought about simulation methods but even then it still comes down to there are just not enough games to get a handle on all the variables that go into a game.

 

[KansasSooner]

Truth is if I could I would really like to do a simulation model, it's just there is so much team data to sort through and the spreadsheets (databases if you prefer) would take so much time to set up I just can't find the time to do it. But damn, would it be fun to do it. I could really get into analyzing the data and models if I could just find a way and the time to set up the data. The biggest impediment is really finding available data as most sites average data over the year and finding a break down of each team for game by game data is not an easy chore to import to a spreadsheet/database for analysis and modeling.

 

[4down20]

Adjust what though? Nothing from last year is relevant to this year as far as doing a true regression model. Weighted averages are better than using last years data in my opinion and even those are hard to use given the number of cupcakes scheduled the first few weeks. No, for true regression models I'll wait until the data is complete and normal before even attempting to make a rating model.

Adjusts to the change in data, same as every other week. You simply remove the previous years data from the equation once the current year is able to stand on it's own.

Each week is calculated fresh on that data, and previous weeks rankings have absolutely no bearing on those results(as ranks are actually just the order of the results anyway).

Personally, I used my power ratings to predict games and was able to keep basically the same % rates during the early weeks using the previous years data as I was during the later weeks. While some of that is because so many cupcake games at the year vs harder games towards the end of the year, overall the results were pretty good. This to me suggested it had a decent grasp on the overall quality of teams.

 

[KansasSooner]

Personally, I used my power ratings to predict games

I'm not trying to make a predictive model though, just a ranking model. And using last year's data does nothing for the current year rankings in my opinion. I want to know how teams line up this year based on what they have done JUST this year. Thus I wait until enough data is available for this year to do any modeling.

 

[4down20]

I'm not trying to make a predictive model though, just a ranking model. And using last year's data does nothing for the current year rankings in my opinion. I want to know how teams line up this year based on what they have done JUST this year. Thus I wait until enough data is available for this year to do any modeling.

 

[KansasSooner]

Like I said before, not enough games to make a good prediction model. The model maybe unbiased but the the confidence interval is not that great. In my models only 70% of the variation is ever accounted for usually and the prediction confidence interval is almost 50% at best, no matter which of the three models I use. Hell, might as well flip a coin. A good simulation model as I mentioned earlier would probably be better but it would still need just this year's data to be worth squat in my opinion. Really, if I could and if I had the time I would do one just for kicks. And I mean a simulation like I used to do for diabetes meters testing, 1 million random normally distributed simulations based on the average with a standard deviation that was not outrageous.

 

[Across The Field]

You're welcome to think so...but it's not.

The B1G conference has 14 teams. If you want to look at how strong the B1G is, you actually have to look at ALL of the teams. Not just the top ones. ALL 14 of them.

Ditto with the ACC having 14 teams.

Ditto on the SEC having 14 teams.

Ditto on the PAC having 12 teams.

Ditto on the Big 12 having 10.


You want to say that Ohio State is a fantastic football team? I'd agree with you. Want to talk about the B1G, you have to also include Illinois in the conversation...

Uh huh.

So, OSU looks sincerely like one of 3 teams that could lay claim to being best in the country right now. michigan looks solid, if not great. Michigan State just won at ND, Wisconsin claims a win over LSU, Nebraska just beat Oregon, and 3 other teams are unbeaten in Indiana, Maryland and Minnesota. No other P5 conference can claim anywhere near the resume the B1G has. But because our bottom two teams are 1-2, that means we're the worst P5 conference?

Kansas rushed the field after beating Rhode Island, they're that bad. Iowa State is 0-3. The best win so far for the entire conference is the 3rd or 4th best win in the B1G, but they're better than the B1G?

The Pac-12 has fewer unbeatens than the B1G East (and half as many as the B1G as a whole) and their best OOC win is Texas followed by Kansas State. Yet even with a 1-3 record vs. the B1G, they're better than the B1G?

The ACC and SEC are the only conferences that can contend with the B1G for the best conference claim right now. Anyone who says the Pac or Big 12 are better is an idiot. I truly have no idea how you can argue that.

 

[ericd7633]

I'm having a tough time deciding between the B1G and SEC for conference supremacy this year. I'd lean SEC based on depth, even though I think the B1G could potentially be better in the top third of each respective conference. It's tough though because the SEC has already started playing conference games against each other so records won't look good. I'd still say Ole Miss is the 2nd best team in the SECW they just happened to play Bama. Imagine if Michigan played tOSU in week 3 just for comparison purposes? I think the B1G and SEC are the top 2 conferences this year.

1. SEC
2. B1G
3. ACC
4. PAC 12
5. Big 12

 

[Across The Field]

I'm having a tough time deciding between the B1G and SEC for conference supremacy this year. I'd lean SEC based on depth, even though I think the B1G could potentially be better in the top third of each respective conference. It's tough though because the SEC has already started playing conference games against each other so records won't look good. I'd still say Ole Miss is the 2nd best team in the SECW they just happened to play Bama. Imagine if Michigan played tOSU in week 3 just for comparison purposes? I think the B1G and SEC are the top 2 conferences this year.

1. SEC
2. B1G
3. ACC
4. PAC 12
5. Big 12

I'm not sold on all of them, but you really can't argue against 8 ranked teams. I think the top 4 in each conference would make for amazing games, and you could throw the ACC in there, as well:

Alabama: Ohio State: Louisville
Texas A&M: michigan: Clemson
Georgia: Michigan State: Florida State
Tennessee: Wisconsin: Miami(FL)

 

[HuskerinBig10]

You're welcome to think so...but it's not.

The B1G conference has 14 teams. If you want to look at how strong the B1G is, you actually have to look at ALL of the teams. Not just the top ones. ALL 14 of them.

Ditto with the ACC having 14 teams.

Ditto on the SEC having 14 teams.

Ditto on the PAC having 12 teams.

Ditto on the Big 12 having 10.


You want to say that Ohio State is a fantastic football team? I'd agree with you. Want to talk about the B1G, you have to also include Illinois in the conversation...

That is a wash.

Big 12 has Kansas and Iowa State. Iowa State is 0-3. 20% of the conference is truly bad. 10% of the conference could finish 0-12. The other eight teams in the conference are almost guaranteed two more wins.

B1G has Illinois and Purdue. Jury is out on Northwestern. No 0-3 teams. 15% of the conference is truly bad. Might be able to up that soon.

ACC has Virginia, Boston College, Syracuse. Jury is out on Wake Forest. Virginia is 0-3. 20% of the conference is truly bad. 7% of the conference could go 0-12.

SEC has Vanderbilt and Kentucky. Jury is out on Missouri. No 0-3 teams. 15% of the conference is truly bad.

PAC 12 has Oregon State. Jury is out on Wash St. No 0-3 teams. 8% of the conference is truly bad.

 

[Across The Field]

That is a wash.

Big 12 has Kansas and Iowa State. Iowa State is 0-3. 20% of the conference is truly bad. 10% of the conference could finish 0-12. The other eight teams in the conference are almost guaranteed two more wins.

B1G has Illinois and Purdue. Jury is out on Northwestern. No 0-3 teams. 15% of the conference is truly bad. Might be able to up that soon.

ACC has Virginia, Boston College, Syracuse. Jury is out on Wake Forest. Virginia is 0-3. 20% of the conference is truly bad. 7% of the conference could go 0-12.

SEC has Vanderbilt and Kentucky. Jury is out on Missouri. No 0-3 teams. 15% of the conference is truly bad.

PAC 12 has Oregon State. Jury is out on Wash St. No 0-3 teams. 8% of the conference is truly bad.

Exactly. Putting equal weight on the bottom 2-3 teams in a conference as you do the top 5-6 is just stupid.

 

[Rolltide94]

So the B1G winning against basically all of the quality opponents on their OOC schedule means the probability is low of them beating quality opponents?

Dude, he also rated Ohio State #1 and Michigan #5 based on probability of winning...I agree with you, he is clearly retarded.

Just accept that you're a good team in a top heavy conference and move on.

 

[Rolltide94]

Exactly. Putting equal weight on the bottom 2-3 teams in a conference as you do the top 5-6 is just stupid.

what if you put equal weight on the top 7 and the bottom 7...what happens then?

 

[Across The Field]

what if you put equal weight on the top 7 and the bottom 7...what happens then?

Then I'd be interested to see how you could rationalize the B1G being the worst P5 conference.

 

[rmilia1]

What the heck...

B1G beats Oklahoma, Colorado, Oregon, Notre Dame, LSU. Oregon State(B sub FD) Iowa State(B sub FD), Duke.

B1G lost to Pittsburgh, North Carolina. Rutgers lost to Washington. Wow, Rutgers.

8 Power 5 wins
3 Power 5 losses

Big 12

Big 12 Beat Missouri, Pittsburgh, Notre Dame

Big 12 lost to Arkansas, Ohio State, Iowa, California, Stanford, Az State

3 Power 5 wins
6 Power 5 losses

Yup, makes sense to me.

Computer polls still use last year's data and preseason rankings for a baseline in their rankings. It's why computer rankings are useless til week 9 or 10. They're still useless after that roo but they at least make sense occasionally

 

[HuskerCradle2Grave]

And my great-great-great-great-uncle or something like that fought in the Revolutionary War. On the winning side even!

College football>>>>>>>>>>>revolutionary war

 

[Across The Field]

Dude, he also rated Ohio State #1 and Michigan #5 based on probability of winning...I agree with you, he is clearly retarded.

Just accept that you're a good team in a top heavy conference and move on.

How is having Ohio State #1 retarded? We're #1 in the nation in TO margin, #1 in TOs forced and #3 in PPG scored. That isn't retarded.

A top heavy conference? So you're seriously going to try and argue that the Big 12 or Pac 12 are somehow better quality top to bottom? Even the ACC is worse at the bottom than we are (our worst team did just beat their 4th worst team).

 

[Rolltide94]

Then I'd be interested to see how you could rationalize the B1G being the worst P5 conference.

The same way you can "rationalize" having Ohio State #1, you put the numbers in and look at the outcome.

 

[Rolltide94]

How is having Ohio State #1 retarded? We're #1 in the nation in TO margin, #1 in TOs forced and #3 in PPG scored. That isn't retarded.

A top heavy conference? So you're seriously going to try and argue that the Big 12 or Pac 12 are somehow better quality top to bottom? Even the ACC is worse at the bottom than we are (our worst team did just beat their 4th worst team).

Christ almighty you are fucking dense. If any of those stats where in the Sagarin rating you might have a point, but I somehow doubt he did that.

Here you are trying to justify Sagarin having you #1 based on qualitative measures while at the same time arguing that his quantitative rankings of the other teams in your conference is somehow wrong.

Bitch about his methodology if you want, but stating he is wrong because your opinion is X, is RETARDED.