menu
Logo
Ohio Football Topic
Topic: Computers
Page: 1 of 1
The Optimist
General User
Member Since: 3/16/2007
Location: CLE
Post Count: 5,611
mail
The Optimist
mail
Posted: 10/8/2012 8:30 PM
All that matters is that we keep going 1-0, but I am certain I am not the only one wondering where the BCS gets the computer component of their rankings...  So I looked around a bit.  Here is what I found:

Highest and lowest rating dropped from the following 6 computers, leaving each team with 4 rankings that are put together.  If you are very interested, there is a fair amount of data out there about how the computers calculate their rankings.

All except Massey updated through bye week

Sagarin: Specifically the ELO Chess Factor (in red)  

Ohio-18
http://usatoday30.usatoday.com/sports/sagarin/fbt12.htm


Anderson and Hester
Ohio-20
http://andersonsports.com/football/ACF_frnk.html


Billingsley
Ohio-32
http://www.cfrc.com/Ratings_2012/WK_8.htm


Colley Matrix
Ohio-27
http://www.colleyrankings.com/


Massey Not Updated
Ohio-39
http://www.masseyratings.com/rate.php?lg=cf&mid=6


Wolfe
Ohio-14
http://prwolfe.bol.ucla.edu/cfootball/home.htm
Last Edited: 10/22/2012 12:29:56 AM by The Optimist
anorris
General User
Member Since: 7/7/2010
Location: Bristol, CT
Post Count: 2,262
mail
anorris
mail
Posted: 10/8/2012 8:50 PM
Was actually going to post some of these earlier but ran out of time before work. I've been tracking them each week in a spreadsheet and it has been interesting to see how some weight SOS and such differently.
Victory
General User
V
Member Since: 3/11/2012
Post Count: 2,519
person
mail
Victory
mail
Posted: 10/8/2012 9:02 PM
That is the Wrong Massey.  That's like the composite Sagarin.  He also has a Predictor and no victory margin ranking.  The no victory margin is used by the BCS.

http://www.masseyratings.com/rate.php?lg=cf&mid=6
The Optimist
General User
Member Since: 3/16/2007
Location: CLE
Post Count: 5,611
mail
The Optimist
mail
Posted: 10/8/2012 9:07 PM
Victory wrote:expand_more
That is the Wrong Massey.  That's like the composite Sagarin.  He also has a Predictor and no victory margin ranking.  The no victory margin is used by the BCS.

http://www.masseyratings.com/rate.php?lg=cf&mid=6

Thank you.  I like this one better.
mf279801
General User
M279801
Member Since: 8/6/2010
Location: Newark, DE
Post Count: 2,486
person
mail
mf279801
mail
Posted: 10/9/2012 9:52 AM
Anyone know why the USAToday sagarin (EloChess 35, with a value of 78.87) is different than the sagarin.com sagarin (EloChess 22, with a value of 81.78)? Both have us at 6-0, so its not like one is an old chart.
oucs 1986
General User
Member Since: 12/20/2004
Location: Mason, OH
Post Count: 251
mail
oucs 1986
mail
Posted: 10/10/2012 1:44 AM
Ask Mr. Sagarin.

He does answer emails, even if he's kind of a jerk in them.

-john
Victory
General User
V
Member Since: 3/11/2012
Post Count: 2,519
person
mail
Victory
mail
Posted: 10/10/2012 1:47 PM
It seems to me that the one on Sagarin.com has a different ELO algorithm only.  The predictor is the same.  Some parameter has been changed.  It seems that teams with one good win and a bunch of bad ones improve the most on the Sagarin.com one and teams with one bad loss and a bunch of OK losses to good teams got hurt the most.  Of course everone's changed because everyone is rated off their opponents ratings.

I'd be curious to hear the answer.  And, yes, he does have the reputation of answering but being arrogant about it.  In fairness to him, he probably gets many emails like that every day and many, many more of the form:

Hey stupid,
Do you not even watch the games?  [My school who is 4-7] just beat the #@!$ out of the @#%$ [My schools rival who is 7-4].  How can a #%$ like you rank those $%!% higher?  You are an $%# and a !$@%!
Last Edited: 10/10/2012 1:48:16 PM by Victory
Victory
General User
V
Member Since: 3/11/2012
Post Count: 2,519
person
mail
Victory
mail
Posted: 10/10/2012 1:59 PM
I emailed him as to which is the BCS ranking.  I will let you guys know what he says.
Victory
General User
V
Member Since: 3/11/2012
Post Count: 2,519
person
mail
Victory
mail
Posted: 10/10/2012 9:14 PM
No response....but he did change his webpage to match USAToday.
mf279801
General User
M279801
Member Since: 8/6/2010
Location: Newark, DE
Post Count: 2,486
person
mail
mf279801
mail
Posted: 10/10/2012 11:53 PM
Ha!
JSF
General User
Member Since: 1/29/2005
Location: Houston, TX
Post Count: 6,580
mail
JSF
mail
Posted: 10/11/2012 12:40 AM
ELO_Chess is what is used in the BCS. Predictor uses margin of victory and is therefore not allowed, though Sagarin says it's a better system.
The Optimist
General User
Member Since: 3/16/2007
Location: CLE
Post Count: 5,611
mail
The Optimist
mail
Posted: 10/14/2012 2:20 PM
Most of these have not been updated.  The Sagarin's have, and we are up all the way to 25!  I truly believe we will be ranked in the only poll that matters... The BCS.

I will update the original post as I see them come out.
Last Edited: 10/14/2012 2:21:48 PM by The Optimist
anorris
General User
Member Since: 7/7/2010
Location: Bristol, CT
Post Count: 2,262
mail
anorris
mail
Posted: 10/14/2012 2:24 PM
The Optimist wrote:expand_more
Most of these have not been updated. The Sagarin's have, and we are up all the way to 25! I truly believe we will be ranked in the only poll that matters... The BCS.
I think many of them are embargoed until after the BCS Countdown show at 8 on ESPN, so tune in! [/ShamelessSelfPromotion]
Last Edited: 10/14/2012 2:25:27 PM by anorris
The Optimist
General User
Member Since: 3/16/2007
Location: CLE
Post Count: 5,611
mail
The Optimist
mail
Posted: 10/14/2012 2:35 PM
anorris wrote:expand_more
Most of these have not been updated. The Sagarin's have, and we are up all the way to 25! I truly believe we will be ranked in the only poll that matters... The BCS.
I think many of them are embargoed until after the BCS Countdown show at 8 on ESPN, so tune in! [/ShamelessSelfPromotion]


I just commented on Twitter that I will be watching this show for the 1st time ever.  I haven't cared this much about a non-game on ESPN since July 8th, 2010.  
The Optimist
General User
Member Since: 3/16/2007
Location: CLE
Post Count: 5,611
mail
The Optimist
mail
Posted: 10/14/2012 9:42 PM
Interesting changes.  Colley dropped from 17 to 25 while we went from 35 to 25 in the Saragin.  Wolfe that was previously unreleased really helped.

Edit: To be clear, the highest rating is dropped (along with the lowest), so the Wolfe actually didn't help at all.  
Last Edited: 10/14/2012 9:53:28 PM by The Optimist
anorris
General User
Member Since: 7/7/2010
Location: Bristol, CT
Post Count: 2,262
mail
anorris
mail
Posted: 10/14/2012 10:05 PM
The Optimist wrote:expand_more
Interesting changes. Colley dropped from 17 to 25 while we went from 35 to 25 in the Saragin. Wolfe that was previously unreleased really helped.

Edit: To be clear, the highest rating is dropped (along with the lowest), so the Wolfe actually didn't help at all.
Well, it helped in that it was dropped instead of the 24, which would've dropped if Wolfe had come in higher.
The Optimist
General User
Member Since: 3/16/2007
Location: CLE
Post Count: 5,611
mail
The Optimist
mail
Posted: 10/21/2012 12:46 PM
Sagarin is the only one updated.... Up from 25 to 18!!!  
Victory
General User
V
Member Since: 3/11/2012
Post Count: 2,519
person
mail
Victory
mail
Posted: 10/21/2012 1:43 PM
My guess is that Sagarin-Elo was not totally unbiased last week and MAY have a very slight tinge of preseason bias in it this week.  The website says that it is unbiased but I wonder.  He used to leave the Elo biased a week or two longer than the predictor.

Alabama was #1 in the ELO last week which didn't make much sense.  Florida had pretty obviously played a tougher schedule last week by almost anyone's standard.

I wonder if the ELO result we noticed on his personal website that was different a few weeks ago were, in fact, that weeks unbiased rankings.
Showing Messages: 1 - 18 of 18
MAC News Links



extra small (< 576px)
small (>= 576px)
medium (>= 768px)
large (>= 992px)
x-large (>= 1200px)
xx-large (>= 1400px)