Updated: BCS Computers

28 posts / 0 new
Login or register to post comments
Last post
November 27th, 2011 at 9:53 PM
#1
Sckon
Joined: 01/29/2011
MGoPoints: 296
Updated: BCS Computers

Quick note to those freaking out. The computers are basing a lot of the rankings right now on predictions and not the actual level of performance. Note the Sargarin has Texas A and M at #12. This will drop when the computers readjusted from predictions to actual results last week. This is why Michigan is all but assured in assuming LSU takes care of business.

 

http://www.usatoday.com/sports/sagarin/fbt11.htm

 

 

UPDATE:

Clarification on what is happening. The BIg 12 went 20 something and 3 for out of conference (which is a heavy basis for the intitial computations). Hence the inflated rating at the moment, the readjustment will take place at the end of the year, with the in conference wins/losses are taken into consideration.

This is why you do not have to worry about Baylor as much as you should. Although it would be comforting if they were to lose because I have only been speaking to the Saragin computer and who knows what the others are doing.

But, this is all avoided by moving up a tick or two in the human polls which carry significant more weight.

 

 

Top
Tags:

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.
November 27th, 2011 at 9:57 PM |  What do you mean that the (Score:1)
Gorgeous Borges
Joined: 10/07/2011
MGoPoints: 680

 What do you mean that the computers will readjust from predictions to actual results? Does this mean Baylor won't jump Michigan?

November 27th, 2011 at 9:57 PM | You keep saying this (Score:2)
AAB
Joined: 01/14/2009
MGoPoints: 7198

where are you getting it?

There is no time, sir, at which ties do not matter.

November 27th, 2011 at 10:03 PM | Saragin, bless his heart, (Score:1)
Sckon
Joined: 01/29/2011
MGoPoints: 296

Saragin, bless his heart, tries to explain it in the link. But basically they switch the way they calculate the teams, but that it not reflected until later. That is why a team like Texas A and M is at #12 because they are calculated under the old measurement, and not to new one which kicks in halfway? through the season.

 

**Note  I am trying to find the article where Saragin expressed concern via this manner, and basically said he wants it to update weekly. His poll has not changed much at all for the past couple of weeks.

November 27th, 2011 at 10:09 PM | The ELO-Chess number (Score:2)
AAB
Joined: 01/14/2009
MGoPoints: 7198

the one used in the BCS, is listed on the page.  You;re saying that's not the real ELO-Chess number?  

There is no time, sir, at which ties do not matter.

November 27th, 2011 at 10:09 PM | But it says in his write up (Score:1)
pfholland
Joined: 01/07/2011
MGoPoints: 280

But it says in his write up that the BCS uses ELO_CHESS, (which has Texas A&M 17th), not the RATING (which has them 12th).

November 27th, 2011 at 10:15 PM | again, I believe SOS is not (Score:1)
Sckon
Joined: 01/29/2011
MGoPoints: 296

again, I believe SOS is not recalculated within that computation on a week to week basis.

November 27th, 2011 at 10:16 PM | Sorry for being a dick but (Score:4 Normal)
AAB
Joined: 01/14/2009
MGoPoints: 7198

[Citation needed]

There is no time, sir, at which ties do not matter.

November 27th, 2011 at 10:55 PM | Looking at 2010 (Score:1)
goblue7612
Joined: 03/02/2010
MGoPoints: 1021

I hope that you are right, and that Sagarin's rankings will change a lot in this last week, but I went back and looked at 2010. Specifically looking at how teams in 2010 moved in Jeff Sagarin's rankings before and after the final week of the season (the time we're currently at). From my examination, only two teams moved more than 3 spots in either direction. Nebraska dropped three spots after losing to 7th/5th ranked Oklahoma, and Arizona lost to unranked Arizona State. It doesn't seem like there was much movement if there was some adjustment. Thoughts?

November 27th, 2011 at 9:59 PM | while I would say that is a (Score:2)
orobs
Joined: 10/03/2010
MGoPoints: 1734

while I would say that is a good thing in theory...it drops michigan from 9th to 23rd and baylor jumps from 24 to 7.  so the adjustment only fucks us more

November 27th, 2011 at 9:59 PM | Most likley not. (Score:1)
Sckon
Joined: 01/29/2011
MGoPoints: 296

Most likley not. Simplistically, most of the computers used a predetermied weight for the conferences (hence the Big 12 getting a huge boost). These are not recalculated until the end of the season though, and with non-conference teams that the Big 12 played not doing so hot, they are all expected to take a hit outside of Oklahoma and Okie St (which have the benefit of the human polls).

November 27th, 2011 at 10:06 PM | Can anyone explain the OP to me? (Score:2)
Yeoman
Joined: 06/08/2011
MGoPoints: 2279

The Sagarin ratings I'm looking at are for "College Football 2011 through games of November 26 Saturday"

That would seem to include A&M's result on Saturday.

To confuse me even more, the Sagarin rating that's used by the BCS is the ELO rating, because the Predictor and the combined Predictor+ELO ratings use point spread and thus don't conform to BCS regs. In the Sagarin/ELO A&M is 18th, not 12th as claimed.

http://www.usatoday.com/sports/sagarin/fbt11.htm

Can anyone tell me what the OP is on about? I'm inclined to think he's wrong but maybe I just don't understand what he's saying.

November 27th, 2011 at 10:09 PM | His rankings are up to date. (Score:1)
Sckon
Joined: 01/29/2011
MGoPoints: 296

His rankings are up to date. But they are not accurate because SOS gets readjusted.

November 27th, 2011 at 10:06 PM | It is all going to work out, (Score:2 Normal)
UMLaw73
UMLaw73's picture
Joined: 11/19/2011
MGoPoints: 92

It is all going to work out, the BCS always does. 

NO Music, Concerts, Ads, "Playing Dress-Up", Night Games, 3:30 Games, Neutral Site Games, And No MASCOTS

November 27th, 2011 at 10:07 PM | The BCS Computers, In My Mind... (Score:1)
LSAClassOf2000
LSAClassOf2000's picture
Joined: 01/07/2011
MGoPoints: 8923

"Funny isn't it, how naughty dentists always make that one fatal mistake."

Follow the random tweets of a Michigan alum - http://twitter.com/#!/LorneEC3

November 27th, 2011 at 10:07 PM | Is Sargin the only computer (Score:2)
Gobluegr
Joined: 07/07/2011
MGoPoints: 649

Is Sargin the only computer ranking that works this way?

November 27th, 2011 at 10:12 PM | I think for the most part, (Score:1)
Sckon
Joined: 01/29/2011
MGoPoints: 296

I think for the most part, the computers all differ in their calculation, but the Saragin model was created by the BCS but just run by Saragin.

If you look at his other body of work, you will see that is he usually right. Wait for the adjustment, all will be fine unless the SEC goes full-retard.

November 27th, 2011 at 10:12 PM | More from the Sagarin link... (Score:2)
Yeoman
Joined: 06/08/2011
MGoPoints: 2279

For the first few weeks of the season, the starting ratings have weight in the process(BAYESIAN), but once the teams are all WELL CONNECTED, then the starting ratings are no longer used and all teams are started equal and the RATING, ELO-CHESS, and PURE POINTS (PREDICTOR) are then done in an UNBIASED manner from that point on.

The teams are now WELL CONNECTED and so all three ratings are UNBIASED.

Any errors caused by the initial estimates of team and conference strength were adjusted for several weeks ago. I don't know if that's true for all the computers used but it's definitely true for the Sagarin. As soon as all teams in the system are connected by a common opponent, at whatever remove, the prior weightings are removed. I'm guessing that's typical of these systems but don't know for sure.

November 27th, 2011 at 10:17 PM | Texas A&M (Score:1)
goblue7612
Joined: 03/02/2010
MGoPoints: 1021

How is 6-6 Texas A&M 12th overall in Sagarin's rankings, 17th in elo_chess and 11th in the predictor? That's why I sort of believe the OP's stance.

November 27th, 2011 at 10:25 PM | I don't know if Sagarin's ever specific about how he does it (Score:2)
Yeoman
Joined: 06/08/2011
MGoPoints: 2279

but he states that the overall rating is a "synthesis" of the ELO and the predictor. However he's combining them, it doesn't seem surprising that the overall ranking would be somewhere between the two.

November 27th, 2011 at 10:28 PM | I meant more, there is no way (Score:1)
goblue7612
Joined: 03/02/2010
MGoPoints: 1021

I meant more, there is no way that Texas A&M should be ranked at this point. They're 6-6....with few quality wins.

November 27th, 2011 at 10:36 PM | Strength of schedule includes (Score:2)
Yeoman
Joined: 06/08/2011
MGoPoints: 2279

Strength of schedule includes losses as well as wins and their losses were to quality teams. I'm looking at the Massey, which hasn't been updated for Saturday yet, and their five losses were to teams ranked 3,6,7,10 and 20. Two of the losses were in OT, two others were 30-29 and 42-38. To a computer that considers spreads and not just w/l that would look pretty good.

November 27th, 2011 at 11:02 PM | Too bad we didn't get that (Score:1)
Gorgeous Borges
Joined: 10/07/2011
MGoPoints: 680

Too bad we didn't get that kind of love in 2005 when we had all of the close Lloyd llosses.

November 27th, 2011 at 11:37 PM | Scoring margin is not a factor. (Score:1)
Mr Miggle
Mr Miggle's picture
Joined: 10/07/2010
MGoPoints: 1637

Scoring margin is not a factor, at least not in Sagarin's poll that the BCS uses. A&M could have have lost six 58-0 games and still be ranked ahead of every BiG team.

They should throw Sagarin's crazy ass out of the BCS formula. He's been doing this for a long time and if he can't avoid ridiculous outcomes by now, he's just incompetent. They no longer consider margin of victory, in part because his formula used to overvalue it to a ridculous degree. Now he has the Big 12 teams ranked 1-10 in strength of schedule because they play each other. That's based on a handful of competitve non-conference games and leads to 6-6 A&M and 7-5 Missouri in his top 20.  

Computer polls are supposed to use criteria similar to the human voters, but without the bias. When 5-7 Texas Tech is three spots behind Wisconsin it should be obvious even to Sagarin that his algorithm is a worthless piece of crap. 

  

Eating is murder.

November 28th, 2011 at 1:03 AM | But margin is a factor in the (Score:2)
Yeoman
Joined: 06/08/2011
MGoPoints: 2279

But margin is a factor in the Sagarin/Predictor, which is where A&M was ranked #12 and was what was being discussed above.

And whether you like its results or not, the Predictor is a better predictor of results than the human polls. It's the use of computers to try to recreate human voters that's the absurdity here.

November 28th, 2011 at 12:51 AM | Margin of victory is (Score:1)
cbuswolverine
Joined: 08/14/2008
MGoPoints: 1778

Margin of victory is irrelevant.  The BCS computer polls are not allowed to factor it in any longer.  This changed several years ago.

November 28th, 2011 at 1:13 AM | Basically because Georgia's (Score:2)
Yeoman
Joined: 06/08/2011
MGoPoints: 2279

Basically because Georgia's higher in the computer rankings so it's harder for them to move up. Pretty much everybody above them also won so they stayed where they were; State was able to pass Clemson and Penn St.

 

Theme provided by Roopletheme; sidebars adapted from Chris Murphy.