menu
Logo
person
Bobcat Attack Menu
Logo
person
Ohio Football Topic
Topic: Clearly the Worst 7-0 Team in the Country
Page: 2 of 3
L.C.
General User
LC
Member Since: 9/1/2005
Post Count: 10,584
person
mail
L.C.
mail
Posted: 10/9/2012 4:39 PM
settings_applications edit_note
I look forward to the day when they are the worst 14-0 team in the country.
OhioCatFan
General User
Member Since: 12/20/2004
Location: Athens, OH
Post Count: 15,700
mail
OhioCatFan
mail
Posted: 10/9/2012 11:41 PM
settings_applications edit_note
L.C. wrote:expand_more
I look forward to the day when they are the worst 14-0 team in the country.


+1                                                                                                                                                                   

Last Edited: 10/9/2012 11:45:49 PM by OhioCatFan
JSF
General User
Member Since: 1/29/2005
Location: Houston, TX
Post Count: 6,580
mail
JSF
mail
Posted: 10/10/2012 12:02 AM
settings_applications edit_note
OUVan wrote:expand_more
I can't believe everything isn't perfect with our football team all the time.  And it really irks me that the other team tries. 


Ever think of this?
giacomo
General User
G
Member Since: 11/20/2007
Post Count: 2,764
person
mail
giacomo
mail
Posted: 10/10/2012 9:16 AM
settings_applications edit_note
Sagarin has us #66 and Penn State at #31. Clearly, our schedule hurts us.
Voice of Reason
General User
Member Since: 7/29/2010
Post Count: 249
mail
Voice of Reason
mail
Posted: 10/10/2012 10:03 AM
settings_applications edit_note
How does our schedule hurt us?  We would not be 6-0 with Miami's schedule or other mid-majors schedule!  This schedule has us in the conversation.  When have we ever been in the conversation?  The fact that we are in it speaks volumes of our schedule and the commitment that we have made as an athletic department to the football program.  We are not forcing our team to play murderers row for a pay day.  Instead we are putting money into the program to make up that gap from not playing murderers row.  This schedule is perfect and look down the road at our scheduling as it gets even better.  They are taking care of the fans with 7 home games the next year, unheard of for a MAC school and again showing our commitment.  For the next several years our schedule sets us up to be a part of the discussion, NATIONALLY.  You may say that the schedule hurts us, but I say it is perfect considering we are not at the level of a Top 15 program right now but are getting that type of publicity.  We are relevant in the college football world and have a chance to be relevant for several years as we build into the next Boise!  Think about what we have going on right now and if the schedule being harder would really benefit us.  I say it is perfect as is.
Deciduous Forest Cat
General User
DFC
Member Since: 12/20/2004
Location: OH
Post Count: 4,559
person
mail
Deciduous Forest Cat
mail
Posted: 10/10/2012 11:41 AM
settings_applications edit_note
I don't know about our current schedule versus another team's. I'd like to see it be a little stronger, but mostly I'd just like to see what we could do if we were at least moderately healthy for an entire season.
LoganElm_grad09
General User
LE09
Member Since: 9/9/2010
Location: South Bloomingville, OH
Post Count: 934
person
mail
LoganElm_grad09
mail
Posted: 10/10/2012 12:06 PM
settings_applications edit_note
Voice of Reason wrote:expand_more
How does our schedule hurt us?  We would not be 6-0 with Miami's schedule or other mid-majors schedule!  This schedule has us in the conversation.  When have we ever been in the conversation?  The fact that we are in it speaks volumes of our schedule and the commitment that we have made as an athletic department to the football program.  We are not forcing our team to play murderers row for a pay day.  Instead we are putting money into the program to make up that gap from not playing murderers row.  This schedule is perfect and look down the road at our scheduling as it gets even better.  They are taking care of the fans with 7 home games the next year, unheard of for a MAC school and again showing our commitment.  For the next several years our schedule sets us up to be a part of the discussion, NATIONALLY.  You may say that the schedule hurts us, but I say it is perfect considering we are not at the level of a Top 15 program right now but are getting that type of publicity.  We are relevant in the college football world and have a chance to be relevant for several years as we build into the next Boise!  Think about what we have going on right now and if the schedule being harder would really benefit us.  I say it is perfect as is.


Answered your own question.  The "hurt" was probably meant toward the ranking, not the record.
anorris
General User
Member Since: 7/7/2010
Location: Bristol, CT
Post Count: 2,262
mail
anorris
mail
Posted: 10/10/2012 12:46 PM
settings_applications edit_note
In the only part of the Sagarin that matters, we're 35 and they're 39.
giacomo
General User
G
Member Since: 11/20/2007
Post Count: 2,764
person
mail
giacomo
mail
Posted: 10/10/2012 5:49 PM
settings_applications edit_note
"hurt" is a relative term in this case. We are successful and that's good. But wouldn't you like to see what we could do against a few more BCS teams? We already have several weak sisters in the MAC, so playing Norfolk State and UMass doesn't help us get to a better bowl game, if that's the goal. Let's say we play Michigan State and Virginia instead and lose both games, but win the MAC. We're in the same bowl that we would be in if we go 12-0 with our current shedule.
OUbobcat9092
General User
OU9092
Member Since: 12/28/2004
Location: Fairfax, VA
Post Count: 1,279
person
mail
OUbobcat9092
mail
Posted: 10/10/2012 6:05 PM
settings_applications edit_note
giacomo wrote:expand_more
"hurt" is a relative term in this case. We are successful and that's good. But wouldn't you like to see what we could do against a few more BCS teams? We already have several weak sisters in the MAC, so playing Norfolk State and UMass doesn't help us get to a better bowl game, if that's the goal. Let's say we play Michigan State and Virginia instead and lose both games, but win the MAC. We're in the same bowl that we would be in if we go 12-0 with our current shedule.


seeing that we are required to play UMass since they are in the conference, there is nothing you can do to upgrade that game on the schedule.  We would have to lose Norfolk State, Marshall or NMSU to upgrade our opponents.  

I would like to see us drop the FCS annual game and at least play one of the other non-AQ conferences for our 4th non-con game.

oucs 1986
General User
Member Since: 12/20/2004
Location: Mason, OH
Post Count: 251
mail
oucs 1986
mail
Posted: 10/10/2012 7:52 PM
settings_applications edit_note
After all this yack, I've been curious about the Sagarin Rating system for years, so I've spent an hour or so playing with mathematica, trying to determine what's going on, actually.

This graph expresses it about as well as anything I've seen.

sagarin graph top 100 teams

This is only the first 100; it looks the same, basically, for all 247 teams...

The "overall rank" CLOSELY matches the PREDICTOR model. 

ELO rank seems to have a very loose corelation.  Mathematica choked on the data, but on a smaller data set, the final rank seems to be something like  

rank = predictor * 0.85 + elo * 0.10 + sched * 0.05

This isn't exactly right; you can see from the graph that the function is curvilinear, but it's a "good enough approximation" for discussion.

SCHEDule strength doesn't matter much, really.

SCHED and ELO do seem to have some co-relation.  

So, from this chart, I think two things jump off the page at me....

There's no need to schedule a particularly hard schedule if you want to rank well on Sagarin's rating system. However, margin of victory does play into the PREDICTOR model, so I think you might actually benefit a bit from having a weaker schedule; blow the other guys out, and you'll do well in PREDICTOR.

The BCS uses ELO, so right now schedule does matter... but only a little.

GO BOBCATS!
-john




Victory
General User
V
Member Since: 3/11/2012
Post Count: 2,519
person
mail
Victory
mail
Posted: 10/10/2012 8:22 PM
settings_applications edit_note
Rank isn't going to be .85*PRE +.1*ELO + .05*sched.  The rank is synthesized from PRE and ELO only.  His page says that straight out and of course this is the case.  Schedule strength won't figure in that synthesis at all.  That isn't to say it doesn't matter.  It matters a lot.  It is fundamental to the whole thing.  Your rating for predictive and ELO is figured off of the game result  and the rating of your opponent.

As L.C described earlier.  If you win a game by 10 points over a team with a 70 rating on a neutral field your predictive raking for that game is about 80.  If you lose a game by 2 at home to a team with a 60 rating your predictive ranking for that game is about 60  -  2 -  3 (for home field) or about 55.  Of course your opponents rating is solved the same way off of your rating and their opponents ratings and their game results.  It is 246 equations in 246 unknowns.  This is easily solved by a computer.  The Sagarin Predictive rating is in actuality probably just a little more complicated than that description but not much.  I have an Excel Program that can do just what I described and the result comes out very close to Sagarin Predictor.

You can Google ELO Ratings if you want.  I won't bother to describe it here.

Basically, if you had the same game score results as a team whose opponents in each game was exactly 15 points lower than you then their rating would be exactly 15 points lower than yours.  So SOS is fundamental to the whole thing.  It just won't correlate the way you are looking at it.
Last Edited: 10/10/2012 9:54:56 PM by Victory
L.C.
General User
LC
Member Since: 9/1/2005
Post Count: 10,584
person
mail
L.C.
mail
Posted: 10/10/2012 9:12 PM
settings_applications edit_note
It isn't surprising that strength of schedule isn't related to Predictor rankings, but is related to ELO. For the Predictor rating, if you play a team rated 30 and beat them by 50, that is the same as playing a team rated 79 and beating them by 1. Thus you can accomplish the same rating with an easy schedule, or with a hard one. For ELO it is different since margin of victory doesn't matter. It will help you a lot more to beat the team rated 79 than it will to beat a team rated 30.

I personally don't think the ELO is going to be very reliable with only 12-14 games. I think it would work a lot better for Basketball or Baseball where there are a lot more games.
Victory
General User
V
Member Since: 3/11/2012
Post Count: 2,519
person
mail
Victory
mail
Posted: 10/10/2012 9:28 PM
settings_applications edit_note
Over time the ELO and Predictor should move closer together as games increase.  Just as you can get a 90 predictive rating by playing even with teams of a rating of 90 or beating teams with a rating of 70 by an average of 20 points you can get a ELO rating of 90 by going 5-5 against teams with 90 ratings or by beating teams with a rating of 70 the vast majority of the time.  History would show that you should beat a team that you are favored by 20 over about 94% of the time so if you played a bunch of teams with a rating of 70 and had a winning percentage of 94% you should, I'd expect, have a ELO rating of about 90.

So, as you see the strength of your opponents is fundamental to your rating.  SOS won't correlate to rating like you showed because it isn't either easier or harder to get a higher rating against a tough schedule.  You have to beat weaker teams worse to get a higher rating but it is in fact easier to beat weaker teams worse.  You have to beat weaker teams more often to get a good ELO but it is in fact easier to beat weaker teams more often.

The fact that it doesn't correlate means that it is working like it should.

Now, though, having said that,  Schedule Strength will start to correlate to rating a little bit on your graph as the season wears on.  This isn't because it is easier to get a high rating against a tough schedule.  It is simply because good programs tend to end up in conferences with other good programs and conference play is beginning in earnest.
L.C.
General User
LC
Member Since: 9/1/2005
Post Count: 10,584
person
mail
L.C.
mail
Posted: 10/10/2012 9:42 PM
settings_applications edit_note
But how many games against teams with a rating of 50 would you have to beat to get a 90 rating? The problem is, there aren't that many games in a year. Thus, I don't think a team that plays mostly teams with ratings in the 50-60 range, like the MAC, could ever reach a rating of 90 in ELO with a 12 game schedule.

If Ohio wins out, they will raise the current 78 rating a few points, but they would need a harder schedule than they have to get to a 90, I suspect. They need to get to 85 or so to get into range for BCS busting. Maybe winning out will get them there, maybe not. I hope we find out.
RSBobcat
General User
Member Since: 8/23/2010
Location: Columbus, OH
Post Count: 4,504
mail
RSBobcat
mail
Posted: 10/10/2012 9:55 PM
settings_applications edit_note
 Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est.

Actually, I think the only thing that really matters is to keep winning..................  
Victory
General User
V
Member Since: 3/11/2012
Post Count: 2,519
person
mail
Victory
mail
Posted: 10/10/2012 10:06 PM
settings_applications edit_note
LC, This is exactly right.  I extrapolate from my curve fit that a 40 point favorite should win about 998 or 999 out of 1000.  I think this also pretty close to history.  Until Stanford beat USC about 4 or 5 years ago no 40+ point underdog had EVER won a game in college football history.  The previous high was 37 in 1985. 

Yeah, in reality, you need to play several games closer to your own level of competition to determine an accurate rating and this is much more true if you are trying to determine a teams actual level by looking a only wins and losses and not victory margin. It takes a lot more games to know anything if you are playing teams substantially worse and substantially better.  This is even true if you are watching on TV or however you are trying to judge.
Last Edited: 10/10/2012 10:12:31 PM by Victory
Victory
General User
V
Member Since: 3/11/2012
Post Count: 2,519
person
mail
Victory
mail
Posted: 10/10/2012 10:16 PM
settings_applications edit_note
RSBobcat wrote:expand_more
 Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est.

Actually, I think the only thing that really matters is to keep winning..................  


I wasn't intentionally trying to turn a football thread into a math lesson.  But there are people who want to have some idea of how these systems figure into our BCS ranking function.  I know there are some that just want to watch football and don't care about the rankings.
oucs 1986
General User
Member Since: 12/20/2004
Location: Mason, OH
Post Count: 251
mail
oucs 1986
mail
Posted: 10/10/2012 10:57 PM
settings_applications edit_note
L.C. wrote:expand_more
It isn't surprising that strength of schedule isn't related to Predictor rankings, but is related to ELO. For the Predictor rating, if you play a team rated 30 and beat them by 50, that is the same as playing a team rated 79 and beating them by 1. Thus you can accomplish the same rating with an easy schedule, or with a hard one For ELO it is different since margin of victory doesn't matter. It will help you a lot more to beat the team rated 79 than it will to beat a team rated 30.

I personally don't think the ELO is going to be very reliable with only 12-14 games. I think it would work a lot better for Basketball or Baseball where there are a lot more games.


Edited to allow for more careful thinking...

However.. it would seem that the diminishing returns you get from a comfortable win would limit the highest ranking we could achieve.

So...  to really game the Predictor system would take an unlikely turn of events... you'd need to beat, handily, a team that handily beats a bunch of other teams.

In other words, two undefeated MAC teams playing late in the season would each have, oh call it a 65 ranking... If one were to beat the other by 20, then they might "jump up" quite a bit in the overall rankings.  Which would explain why the SEC is ranked as highly as it is, and why the MAC is ranked as lowly.  There are an awful lot of losses on the MAC side of the ledger.

Thanks LC & Legend.  We can take this elsewhere if other folks is mathed out.

-john

Last Edited: 10/10/2012 11:07:19 PM by oucs 1986
RSBobcat
General User
Member Since: 8/23/2010
Location: Columbus, OH
Post Count: 4,504
mail
RSBobcat
mail
Posted: 10/10/2012 11:05 PM
settings_applications edit_note
Victory wrote:expand_more
 Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est.

Actually, I think the only thing that really matters is to keep winning..................  


I wasn't intentionally trying to turn a football thread into a math lesson.  But there are people who want to have some idea of how these systems figure into our BCS ranking function.  I know there are some that just want to watch football and don't care about the rankings.


No problem - geek away - I'm no math wiz, and I did get some entertainment and enlightenment from the thread read. End of day though, there is Only One Way to take destiny more into your own hands - Winning. And the Bigger, the Better. Embarassing your wife might help the cause more than the "schedule" too!  Will be listening for you Saturday!
Last Edited: 10/10/2012 11:08:10 PM by RSBobcat
oucs 1986
General User
Member Since: 12/20/2004
Location: Mason, OH
Post Count: 251
mail
oucs 1986
mail
Posted: 10/10/2012 11:14 PM
settings_applications edit_note
RSBobcat wrote:expand_more
 Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est.

Actually, I think the only thing that really matters is to keep winning..................  


I wasn't intentionally trying to turn a football thread into a math lesson.  But there are people who want to have some idea of how these systems figure into our BCS ranking function.  I know there are some that just want to watch football and don't care about the rankings.


No problem - geek away - I'm no math wiz, and I did get some entertainment and enlightenment from the thread read. End of day though, there is Only One Way to take destiny more into your own hands - Winning. And the Bigger, the Better. Embarassing your wife might help the cause more than the "schedule" too!  Will be listening for you Saturday!



Stop by 104 and help me embarrass her!  Row 18.. uh. seats ?20? ?21? heck, i'd have to look and the tickets are in the filing cabinet.  If that's not right, it's close enough for govt. work.

-john

oucs 1986
General User
Member Since: 12/20/2004
Location: Mason, OH
Post Count: 251
mail
oucs 1986
mail
Posted: 10/10/2012 11:21 PM
settings_applications edit_note
Victory wrote:expand_more
Rank isn't going to be .85*PRE +.1*ELO + .05*sched.  The rank is synthesized from PRE and ELO only.  His page says that straight out and of course this is the case.  Schedule strength won't figure in that synthesis at all.  That isn't to say it doesn't matter.  It matters a lot.  It is fundamental to the whole thing.  Your rating for predictive and ELO is figured off of the game result  and the rating of your opponent.

As L.C described earlier.  If you win a game by 10 points over a team with a 70 rating on a neutral field your predictive raking for that game is about 80.  If you lose a game by 2 at home to a team with a 60 rating your predictive ranking for that game is about 60  -  2 -  3 (for home field) or about 55.  Of course your opponents rating is solved the same way off of your rating and their opponents ratings and their game results.  It is 246 equations in 246 unknowns.  This is easily solved by a computer.  The Sagarin Predictive rating is in actuality probably just a little more complicated than that description but not much.  I have an Excel Program that can do just what I described and the result comes out very close to Sagarin Predictor.

You can Google ELO Ratings if you want.  I won't bother to describe it here.

Basically, if you had the same game score results as a team whose opponents in each game was exactly 15 points lower than you then their rating would be exactly 15 points lower than yours.  So SOS is fundamental to the whole thing.  It just won't correlate the way you are looking at it.


So I'm assuming (perhaps wrongly) that there would be a large matrix... 247 unknowns  247 coefficients... 247 constants...  

I'm guessing the 247 constants would be your current PREDICTOR rating, and the 247 coefficients would be "expected margin of victory"?  No?

-john

JSF
General User
Member Since: 1/29/2005
Location: Houston, TX
Post Count: 6,580
mail
JSF
mail
Posted: 10/11/2012 12:39 AM
settings_applications edit_note
Giacomo has taken his complaining about scheduling, ignorance of why some teams are on the schedule, and his lack of knowledge on how rankings work to a new board.
Victory
General User
V
Member Since: 3/11/2012
Post Count: 2,519
person
mail
Victory
mail
Posted: 10/11/2012 3:35 PM
settings_applications edit_note
oucs 1986 wrote:expand_more
Rank isn't going to be .85*PRE +.1*ELO + .05*sched.  The rank is synthesized from PRE and ELO only.  His page says that straight out and of course this is the case.  Schedule strength won't figure in that synthesis at all.  That isn't to say it doesn't matter.  It matters a lot.  It is fundamental to the whole thing.  Your rating for predictive and ELO is figured off of the game result  and the rating of your opponent.

As L.C described earlier.  If you win a game by 10 points over a team with a 70 rating on a neutral field your predictive raking for that game is about 80.  If you lose a game by 2 at home to a team with a 60 rating your predictive ranking for that game is about 60  -  2 -  3 (for home field) or about 55.  Of course your opponents rating is solved the same way off of your rating and their opponents ratings and their game results.  It is 246 equations in 246 unknowns.  This is easily solved by a computer.  The Sagarin Predictive rating is in actuality probably just a little more complicated than that description but not much.  I have an Excel Program that can do just what I described and the result comes out very close to Sagarin Predictor.

You can Google ELO Ratings if you want.  I won't bother to describe it here.

Basically, if you had the same game score results as a team whose opponents in each game was exactly 15 points lower than you then their rating would be exactly 15 points lower than yours.  So SOS is fundamental to the whole thing.  It just won't correlate the way you are looking at it.


So I'm assuming (perhaps wrongly) that there would be a large matrix... 247 unknowns  247 coefficients... 247 constants...  

I'm guessing the 247 constants would be your current PREDICTOR rating, and the 247 coefficients would be "expected margin of victory"?  No?

-john



I am sure that there is more than one way to get the answer.  I took each teams predictive rating to be an unknown.  I set up the equation calculating from game result and opponent rating as described in simple system above.  I used an Newton-Raphson Numerical Method to solve for the ratings.
oucs 1986
General User
Member Since: 12/20/2004
Location: Mason, OH
Post Count: 251
mail
oucs 1986
mail
Posted: 10/11/2012 4:52 PM
settings_applications edit_note
Victory wrote:expand_more
Rank isn't going to be .85*PRE +.1*ELO + .05*sched.  The rank is synthesized from PRE and ELO only.  His page says that straight out and of course this is the case.  Schedule strength won't figure in that synthesis at all.  That isn't to say it doesn't matter.  It matters a lot.  It is fundamental to the whole thing.  Your rating for predictive and ELO is figured off of the game result  and the rating of your opponent.

As L.C described earlier.  If you win a game by 10 points over a team with a 70 rating on a neutral field your predictive raking for that game is about 80.  If you lose a game by 2 at home to a team with a 60 rating your predictive ranking for that game is about 60  -  2 -  3 (for home field) or about 55.  Of course your opponents rating is solved the same way off of your rating and their opponents ratings and their game results.  It is 246 equations in 246 unknowns.  This is easily solved by a computer.  The Sagarin Predictive rating is in actuality probably just a little more complicated than that description but not much.  I have an Excel Program that can do just what I described and the result comes out very close to Sagarin Predictor.

You can Google ELO Ratings if you want.  I won't bother to describe it here.

Basically, if you had the same game score results as a team whose opponents in each game was exactly 15 points lower than you then their rating would be exactly 15 points lower than yours.  So SOS is fundamental to the whole thing.  It just won't correlate the way you are looking at it.


So I'm assuming (perhaps wrongly) that there would be a large matrix... 247 unknowns  247 coefficients... 247 constants...  

I'm guessing the 247 constants would be your current PREDICTOR rating, and the 247 coefficients would be "expected margin of victory"?  No?

-john



I am sure that there is more than one way to get the answer.  I took each teams predictive rating to be an unknown.  I set up the equation calculating from game result and opponent rating as described in simple system above.  I used an Newton-Raphson Numerical Method to solve for the ratings.


Right, but you need a "rating" for the opponent...  You're using the one prior to the current one i assume

Interested in working on it more.. thx for the yack.

-john
Showing Messages: 26 - 50 of 75
reply
vertical_align_top
vertical_align_center
vertical_align_bottom
add_comment
chevron_backward
chevron_forward
MAC News Links



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