1. SEC (8.25)
2. Pac-12 (8.24)
3. Big Ten (8.09)
4. Big 12 (8.03)
5. ACC (7.90)
--------------------------------
6. Sun Belt (7.31)
7. Mountain West (7.19)
8. AAC (7.12)
9. Conference USA (6.93)
10. MAC (6.89)

https://masseyratings.com/cf/fbs/ratings?c=1

For the first time this season, the Pac-12 falls from the #1 spot. The Sun Belt widens its lead over the rest of the G5.

I'm always excited about the SEC, but...um...how are we ranked ahead of the Pac right now?

Maybe they hired Quo?

(09-17-2023 09:10 AM)bryanw1995 Wrote:  I'm always excited about the SEC, but...um...how are we ranked ahead of the Pac right now?

Maybe they hired Quo?

Stanford. ASU.

As I've said before, I do not look at computer rankings, which by inference includes composites of computer rankings, until about six or so games have been played, because IIRC early in a season, some computers use "filler" from last year to make up for the lack of data points for this year.

It takes until about then for the old filler data to be washed out.

So I suspect the reason the SEC rates high despite its poor performance is because filler from last year is being used in some of these computers, and the SEC was mighty last year.

Although I have to admit, the G5 rankings mee the "eye test" as I would see it. But the P5 do not.

Hard to understand these computer models
ODU beats ULL last week and now rated 21 spots lower? Must be too much human influence in the formula.

About as useless as Coaches and AP rankings.

(09-17-2023 09:21 AM)jrj84105 Wrote:  

(09-17-2023 09:10 AM)bryanw1995 Wrote:  I'm always excited about the SEC, but...um...how are we ranked ahead of the Pac right now?

Maybe they hired Quo?

Stanford. ASU.

That's fitting, they conspired to kill off the Pac over the past year, the least they could do is also drag down their ratings now.

Personally, I like to check the Massey when I want to have a good scoff

(09-17-2023 09:26 AM)Gemofthehills Wrote:  Hard to understand these computer models
ODU beats ULL last week and now rated 21 spots lower? Must be too much human influence in the formula.

Its simply the strength of competition and its only 3 games. It would probably take human influence to rank ODU higher.

Sagarin has an ELO system which is based on a chess ranking system. The formula is more complicated now, but it still works basically the same way as when it was simpler.

If you were just starting in chess and had played 3 games against players rated 1700, 1200, and 1000 and went 1 win, 1 loss and 1 draw, your rating would be the average of those opponents' ratings, 1300. You got 500 points for a win and lost 500 for a loss. That meant if you played 3 1800 players and lost to all 3, you would also have a 1300 rating. These football ratings include all sorts of different things, but opponent strength is a major component of almost all of them.

As Quo says, they often use last year's data as filler to give you a starting point (you can't start everyone at 0 and get anything meaningful with just 3 games).

(09-17-2023 10:19 AM)bullet Wrote:  

(09-17-2023 09:26 AM)Gemofthehills Wrote:  Hard to understand these computer models
ODU beats ULL last week and now rated 21 spots lower? Must be too much human influence in the formula.

Its simply the strength of competition and its only 3 games. It would probably take human influence to rank ODU higher.

Sagarin has an ELO system which is based on a chess ranking system. The formula is more complicated now, but it still works basically the same way as when it was simpler.

If you were just starting in chess and had played 3 games against players rated 1700, 1200, and 1000 and went 1 win, 1 loss and 1 draw, your rating would be the average of those opponents' ratings, 1300. You got 500 points for a win and lost 500 for a loss. That meant if you played 3 1800 players and lost to all 3, you would also have a 1300 rating. These football ratings include all sorts of different things, but opponent strength is a major component of almost all of them.

As Quo says, they often use last year's data as filler to give you a starting point (you can't start everyone at 0 and get anything meaningful with just 3 games).

So a human rates the teams at the start of the season?

(09-17-2023 10:20 AM)Gemofthehills Wrote:  

(09-17-2023 10:19 AM)bullet Wrote:  

(09-17-2023 09:26 AM)Gemofthehills Wrote:  Hard to understand these computer models
ODU beats ULL last week and now rated 21 spots lower? Must be too much human influence in the formula.

Its simply the strength of competition and its only 3 games. It would probably take human influence to rank ODU higher.

Sagarin has an ELO system which is based on a chess ranking system. The formula is more complicated now, but it still works basically the same way as when it was simpler.

If you were just starting in chess and had played 3 games against players rated 1700, 1200, and 1000 and went 1 win, 1 loss and 1 draw, your rating would be the average of those opponents' ratings, 1300. You got 500 points for a win and lost 500 for a loss. That meant if you played 3 1800 players and lost to all 3, you would also have a 1300 rating. These football ratings include all sorts of different things, but opponent strength is a major component of almost all of them.

As Quo says, they often use last year's data as filler to give you a starting point (you can't start everyone at 0 and get anything meaningful with just 3 games).

So a human rates the teams at the start of the season?

Normally they simply use last year's data. I know Sagarin will start with last year's data and weight it less and less as the season goes on until the 4th or 5th game when he drops prior year data. So when that happens, you may see significant changes in Sagarin.

With the transfer portal last year team means less and less. Also creates bias in the numbers but I guess you take what you get.

(09-17-2023 11:23 AM)Gemofthehills Wrote:  With the transfer portal last year team means less and less. Also creates bias in the numbers but I guess you take what you get.

I dont think anyone is arguing that point, because its more valid now with the portal than ever. However, they do need a starting point. Without some sort of foundational starting point, all teams would bounce around wildly until the 5th game or so. So you might as well not even try to rank the current year until half the season is over.

But because people want some inference of what is going on, they give a starting bias, and it loses its weight week after week.

Interesting out of conference results so far. In overall record, the Big 10, Big 12 and ACC are all an identical 29-13, but they have different amounts of FCS and conference games. AAC and MWC trail the pack. Pac 12 leads the pack. Pac, Big 12 and Big 10 have the bulk of their ooc done for the season. The only one who could catch the Pac in the regular season is the SEC. And that would require that they go at least 18-1 and those 18 games are all against FBS competition. The Pac only has 2 ooc games left.

Pac 12 84.0%
Big 10 69.0%
SEC 68.0%
ACC 61.5%
Big 12 58.6%

Sun Belt 45.8%
CUSA 35.3%
MAC 26.9%
AAC 26.7%
MWC 24.0%

Ind 41.7%

Note: these are the records vs. FBS only

(09-17-2023 12:38 PM)bullet Wrote:  Interesting out of conference results so far. In overall record, the Big 10, Big 12 and ACC are all an identical 29-13, but they have different amounts of FCS and conference games. AAC and MWC trail the pack. Pac 12 leads the pack. Pac, Big 12 and Big 10 have the bulk of their ooc done for the season. The only one who could catch the Pac in the regular season is the SEC. And that would require that they go at least 18-1 and those 18 games are all against FBS competition. The Pac only has 2 ooc games left.

Pac 12 84.0%
Big 10 69.0%
SEC 68.0%
ACC 61.5%
Big 12 58.6%

Sun Belt 45.8%
CUSA 35.3%
MAC 26.9%
AAC 26.7%
MWC 24.0%

Ind 41.7%

4 of those top 8 from the Pac are joining the B1G next year, too. Would be pretty funny if the B1G broke their title drought next year with help from one of the newcomers.