|
Replies: 69
| visibility 1
|
CU Medallion [54087]
TigerPulse: 100%
Posts: 43644
Joined: 11/17/03
|
|
|
|
 |
110%er [6329]
TigerPulse: 100%
Posts: 5113
Joined: 7/19/13
|
I am not sure what I am looking at
Sep 5, 2013, 11:17 AM
|
|
Surely this is not his rankings. I mean, I always thought they were bogus, but he should get laughed out of the business if these are his actual rankings.
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
It's only after one game. Math will not work out well
Sep 5, 2013, 11:20 AM
|
|
For another few weeks. He obviously has something in there that is based on last season, else Stanford couldn't even exist in the rankings, for example.
|
|
|
|
|
|
110%er [6329]
TigerPulse: 100%
Posts: 5113
Joined: 7/19/13
|
Hence why his rankings should be thrown out of the
Sep 5, 2013, 11:23 AM
|
|
equation....
Any ranking that has us below uga right now is a messed up ranking system. It is just easier to see the flaws now because of the one game and the fact that uga is ahead of Clemson. When you are 8 weeks into the season and almost 500 games to account for, it is a lot harder to break it down. Here, the bias and the flawed system is glaring.
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
There is no equation right now. Only one that matters
Sep 5, 2013, 11:24 AM
|
|
Is the last one.
|
|
|
|
|
|
110%er [6329]
TigerPulse: 100%
Posts: 5113
Joined: 7/19/13
|
If that is true, I recommend him redoing his math***
Sep 5, 2013, 11:25 AM
|
|
|
|
|
|
|
|
All-In [28802]
TigerPulse: 100%
Posts: 58393
Joined: 11/14/03
|
why? because the rankings don't work after one game?***
Sep 5, 2013, 11:26 AM
|
|
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
No rankings work after one game, including human polls.***
Sep 5, 2013, 11:27 AM
|
|
|
|
|
|
|
|
110%er [6329]
TigerPulse: 100%
Posts: 5113
Joined: 7/19/13
|
They are certainly more logical than that***
Sep 5, 2013, 11:28 AM
|
|
|
|
|
|
|
|
All-In [28802]
TigerPulse: 100%
Posts: 58393
Joined: 11/14/03
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
Of course they would be.***
Sep 5, 2013, 11:29 AM
|
|
|
|
|
|
|
|
110%er [6329]
TigerPulse: 100%
Posts: 5113
Joined: 7/19/13
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
What makes you say that?
Sep 5, 2013, 11:28 AM
|
|
Why, specifically, do you think his system sucks?
|
|
|
|
|
|
110%er [6329]
TigerPulse: 100%
Posts: 5113
Joined: 7/19/13
|
Re: What makes you say that?
Sep 5, 2013, 11:32 AM
|
|
Because there is obviously an element that is skewing a rational outcome. Your rationale is that he is using last years results. If true, that system sucks. Teams change by an average of 20-25% every year.
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
Re: What makes you say that?
Sep 5, 2013, 11:34 AM
|
|
http://www.bcsknowhow.com/better-know-a-bcs-computer-jeff-sagarin-ratings
Sagarin includes a preseason ranking system and weighs it through the first few weeks.
However, when the first BCS rankings are released, Sagarin removes the preseason ranking, allowing for each team to be ranked as if they was not a preseason ranking — all starting at the same spot.
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
|
|
|
|
|
Oculus Spirit [75940]
TigerPulse: 100%
Posts: 108953
Joined: 10/26/03
|
Are you doing yours again this year?***
Sep 5, 2013, 1:07 PM
|
|
|
|
|
|
|
|
All-In [28802]
TigerPulse: 100%
Posts: 58393
Joined: 11/14/03
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
I wish he had more specific info on his formula there.
Sep 5, 2013, 11:26 AM
|
|
It seems like I remember reading somewhere one time that any pre-season numbers drop after after a certain week...like week 3 or 4, after all the teams are "well-connected" statistically.
|
|
|
|
|
|
CU Medallion [60282]
TigerPulse: 100%
Posts: 42584
Joined: 11/30/98
|
is home field worth a TD (6.49 points) now?
Sep 5, 2013, 11:40 AM
|
|
I thought +/-3 points was the general rule of thumb.
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
It's 6.49 after one week of games. That will change, too.
Sep 5, 2013, 11:42 AM
|
|
At the end of last year it was 2.84 points.
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
Which is another reason why UGA is higher than Clemson
Sep 5, 2013, 11:43 AM
[ in reply to is home field worth a TD (6.49 points) now? ] |
|
Clemson beat UGA by less than the home-field advantage, so in a purely mathematical sense after this small sample size, UGA was better than Clemson.
|
|
|
|
|
|
All-In [28802]
TigerPulse: 100%
Posts: 58393
Joined: 11/14/03
|
actually...
Sep 5, 2013, 11:59 AM
|
|
it says that home field is worth three points **(in the tutorial), so Clemson beat Georgia by exactly the home- field advantage. Woulda' been nice if we would've played any defense at all on Georgia's last drive to keep us ahead by 10, but we still won.
** Also, based on the prediction score for Clemson, even if you add the current 6- point home field advantage, Clemson won by more than what Sagarin would've predicted (~ a one point win).
Message was edited by: camcgee®
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
No, home field right now is 6.49. If you read closely...
Sep 5, 2013, 12:00 PM
|
|
He says that the "three points" used in the explanation is just for illustrative purposes. The actual number used for the rankings is show at the top of the rankings.
|
|
|
|
|
|
All-In [28802]
TigerPulse: 100%
Posts: 58393
Joined: 11/14/03
|
right, but you're still doing it wrong...
Sep 5, 2013, 12:05 PM
|
|
To predict a game, you're supposed to add the home field advantage to whatever the predictor score is. Clemson's predictor score is about 5 points lower than Georgia's (even after beating them), so he would've had us winning by 1 after adding the home- field advantage. Clemson actually outperformed his prediction (even, somehow, with us actually beating them already factored in).
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
It's not about out-performing a prediction.
Sep 5, 2013, 12:08 PM
|
|
We're talking about who is better. If Virginia Tech had lost by 7 to Bama, they would have beat the prediction. But that doesn't mean that VT was better than Bama.
But if VT went into Bama's house and lost by a point in front of 90,000 people, then you might say VT on a neutral field is better.
|
|
|
|
|
|
All-In [28802]
TigerPulse: 100%
Posts: 58393
Joined: 11/14/03
|
"out- performing the prediction" wasn't the important point***
Sep 5, 2013, 12:11 PM
|
|
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
OK, I think we got our signals crossed. All I was saying
Sep 5, 2013, 12:14 PM
|
|
Was that with a 6.49 home-field factor, Georgia losing by three at Clemson actually meant they were the better team, based on the rudimentary math of a week's worth of games, regardless of what the predictor says or said.
I thought you were disagreeing with that. Maybe you weren't.
|
|
|
|
|
|
Orange Blooded [4956]
TigerPulse: 100%
Posts: 6983
Joined: 10/12/06
|
Re: It's only after one game. Math will not work out well
Sep 5, 2013, 11:25 AM
[ in reply to It's only after one game. Math will not work out well ] |
|
That is the crazy part. They say the BCS ranking don't come out until week 6 in order to get a good basis for ranking yet the computer factor in a preseason ranking? Amazing how a person can design a logical computer system to come up with irrational results.
|
|
|
|
|
|
All-In [29049]
TigerPulse: 100%
Posts: 36110
Joined: 8/28/00
|
And in week 6
Sep 5, 2013, 1:01 PM
|
|
When the BCS comes out, the Sagarin rankings used will have no pre-season data involved.
|
|
|
|
|
|
Orange Blooded [2563]
TigerPulse: 100%
Posts: 4860
Joined: 10/2/08
|
IKR..... Must have missed the little $EC multiplier off
Sep 5, 2013, 11:17 AM
|
|
to the right....
|
|
|
|
|
|
CU Medallion [60282]
TigerPulse: 100%
Posts: 42584
Joined: 11/30/98
|
SC State is #198
Sep 5, 2013, 11:27 AM
|
|
Clemson will drop out of the top 50 in next week's rankings.
|
|
|
|
|
|
All-TigerNet [14921]
TigerPulse: 100%
Posts: 12314
Joined: 3/28/06
|
This is why computers and formulas should never
Sep 5, 2013, 11:27 AM
|
|
decide who plays for the MNC...
PLAYOFF!!!
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
Certainly, Week 1 computer rankings should never be used
Sep 5, 2013, 11:27 AM
|
|
To decide anything.
|
|
|
|
|
|
Heisman Winner [108390]
TigerPulse: 100%
Posts: 64974
Joined: 2/25/06
|
what's wrong spud? a bcs computer still had nd #1 after
Sep 5, 2013, 11:32 AM
|
|
getting blasted by bama.
it's human, they have to explain their assumptions...
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
Found it:
Sep 5, 2013, 11:33 AM
|
|
"Sagarin includes a preseason ranking system and weighs it through the first few weeks.
However, when the first BCS rankings are released, Sagarin removes the preseason ranking, allowing for each team to be ranked as if they was not a preseason ranking — all starting at the same spot."
http://www.bcsknowhow.com/better-know-a-bcs-computer-jeff-sagarin-ratings
|
|
|
|
|
|
110%er [6329]
TigerPulse: 100%
Posts: 5113
Joined: 7/19/13
|
Well, that is idiotic
Sep 5, 2013, 11:36 AM
|
|
In my opinion anyway.
Preseason polls are useless. It would make more sense to me to just start from scratch and not add that in at any time.
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
Then you wouldn't even have a ranking.
Sep 5, 2013, 11:37 AM
|
|
Or teams like Eastern Washington would be in the top 10.
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
The only one that matters is the last one.
Sep 5, 2013, 11:38 AM
[ in reply to Well, that is idiotic ] |
|
All the others are just made to generate page clicks and discussion.
|
|
|
|
|
|
110%er [6329]
TigerPulse: 100%
Posts: 5113
Joined: 7/19/13
|
True, it is all that matters
Sep 5, 2013, 11:48 AM
|
|
You would have a ranking, and you may have Eastern Michigan in the top 10 and Alabama not in the top25, but it would be real time accurate based on that systems guidelines.
I would rather see E. Michigan get credit for something they have done instead of seeing Alabama get credit for something they have not done.
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
I'd rather see something more realistic.
Sep 5, 2013, 11:50 AM
|
|
But then again, I don't remember ever looking at computer rankings after Week 1. It's not going to be a true ranking yet.
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
Nobody would take Sagarin's Week 1 rankings less seriously
Sep 5, 2013, 11:45 AM
|
|
Than Jeff Sagarin.
|
|
|
|
|
|
All-Conference [444]
TigerPulse: 91%
Posts: 535
Joined: 6/30/11
|
I like how Kentucky is 10 pt. favorite over Western Kentucky
Sep 5, 2013, 11:52 AM
|
|
Shows that it is very in tune with college football.
|
|
|
|
|
|
Legend [19654]
TigerPulse: 84%
Posts: 17312
Joined: 7/25/07
|
Re: Uh whuuuuut.....
Sep 5, 2013, 11:55 AM
|
|
This is exactly why the BCS is going away. Because it uses this crap.
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
No, it doesn't.***
Sep 5, 2013, 11:56 AM
|
|
|
|
|
|
|
|
All-In [25021]
TigerPulse: 100%
Posts: 42972
Joined: 7/31/10
|
According to the article, YES it does.***
Sep 5, 2013, 12:36 PM
|
|
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
Link to an article that says the BCS uses Sagarin's
Sep 5, 2013, 12:40 PM
|
|
Week 1 rankings for anything?
|
|
|
|
|
|
All-In [25021]
TigerPulse: 100%
Posts: 42972
Joined: 7/31/10
|
|
|
|
|
|
All-In [29049]
TigerPulse: 100%
Posts: 36110
Joined: 8/28/00
|
His formula, by the time the BCS considers it
Sep 5, 2013, 2:26 PM
|
|
Is all current-season data
|
|
|
|
|
|
Commissioner [967]
TigerPulse: 97%
Posts: 1181
Joined: 2/9/02
|
That's a complete joke
Sep 5, 2013, 12:02 PM
|
|
nm
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
Getting mad at Week 1 computer rankings would be akin
Sep 5, 2013, 12:04 PM
|
|
To getting mad because Clemson had to start their first drive at the 19 yard line because of a short kickoff return. There's some game left to be played...
|
|
|
|
|
|
All-In [29049]
TigerPulse: 100%
Posts: 36110
Joined: 8/28/00
|
Slight change
Sep 5, 2013, 12:08 PM
|
|
"Clemson had to start their first drive at the 19 yard line because of a short kickoff return..." in the Spring Game.
The formula isn't designed to produce accurate results at this point. Honestly, it simply couldn't be.
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
Correct. My computer rankings don't even compute yet
Sep 5, 2013, 12:10 PM
|
|
And neither would Sagarin's if he didn't include some pre-season factor.
|
|
|
|
|
|
All-In [29049]
TigerPulse: 100%
Posts: 36110
Joined: 8/28/00
|
It's simply an equation
Sep 5, 2013, 12:06 PM
|
|
An equation which has been honed over years to give a numeric ranking at the end of the year to college football.
At this point the sample size is very, very small, so in order to seed the equation (and since the rankings right now mean less than nothing)he includes some data stretching back to last year. It makes the numbers screwy right now, but everyone knows that.
Right now we have considerably less than 1/13 of the data which will be completed this season. Each week the number of games will grow, and with it the number of interconnected data points. That will cause the equation to begin to produce the results it is designed to produce.
|
|
|
|
|
|
Heisman Winner [108390]
TigerPulse: 100%
Posts: 64974
Joined: 2/25/06
|
"but everyone knows that." no they don't and that's why
Sep 5, 2013, 12:10 PM
|
|
this thing has blown up on the tigernet.
i'd much rather him release all his assumptions & weights so the theory, not the math & outcome, can be hashed over. that would be interesting.
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
That would be extremely esoteric.***
Sep 5, 2013, 12:11 PM
|
|
|
|
|
|
|
|
Heisman Winner [108390]
TigerPulse: 100%
Posts: 64974
Joined: 2/25/06
|
i think you'd have a better chance of bringing people along
Sep 5, 2013, 12:13 PM
|
|
than what's going on now.
|
|
|
|
|
|
All-In [29049]
TigerPulse: 100%
Posts: 36110
Joined: 8/28/00
|
Probably
Sep 5, 2013, 12:59 PM
|
|
I'd argue a better presentation/explanation would be great.
However it doesn't change the simple fact that at this point in the season his rankings are VERY fluid and really only serve as interesting fodder at best.
|
|
|
|
|
|
Heisman Winner [108390]
TigerPulse: 100%
Posts: 64974
Joined: 2/25/06
|
people blame the math when they shouldn't
Sep 5, 2013, 1:15 PM
|
|
i blame the people & you're right, but it also doesn't change the simple fact it's tainted by human bias no matter the point in the season.
kind of silly to argue whether 1+1=2 when you should argue why the sec's modifier is so high or why or why not MOV should be included & how much weight it should carry IF it is.
i'd rather have the human polls, scrap the math & hidden bias, then decide on a best number of teams playoff bound.
we're moving there and here you & i are discussing the real issue.
just happy we're talking football again.
|
|
|
|
|
|
All-In [29049]
TigerPulse: 100%
Posts: 36110
Joined: 8/28/00
|
Absolutely
Sep 5, 2013, 1:35 PM
|
|
I think these sort of rankings are interesting, but admittedly I'm a math nerd.
However, if it were up to me, polls, computer rankings, etc would have no place in determining who plays for the title. The solution is painfully simple, but for some reason folks are too caught up in the pursuit of the dollars to see it.
Win your conference = go to playoff.
Playoffs then ranked by some poll or the other. (Keep human wow factor, and big OOC SOS involved)
First round of playoff at home team of higher seed.
Boom, easy.
|
|
|
|
|
|
Heisman Winner [108390]
TigerPulse: 100%
Posts: 64974
Joined: 2/25/06
|
kiss...
Sep 5, 2013, 1:39 PM
|
|
don't take that the wrong way.
|
|
|
|
|
|
All-In [25021]
TigerPulse: 100%
Posts: 42972
Joined: 7/31/10
|
That's why it's flawed from inception. He includes a SOS
Sep 5, 2013, 12:45 PM
[ in reply to It's simply an equation ] |
|
based on last year and his numbers are wrong. Take Alabama for instance. Their schedule based on last year's records of this year's opponents would be #104 in the country. If based on last year's final rankings of this year's opponents, Bama's SOS would be #55. He has them at #34. Any formula, however well vetted through the years,is only as good as the accuracy of the inputted data. Sagarin clearly has an agenda and a bias from the start.
|
|
|
|
|
|
All-In [26968]
TigerPulse: 96%
Posts: 44823
Joined: 7/6/10
|
It is not clear he has an agenda.
Sep 5, 2013, 12:48 PM
|
|
It is clear that his accounting for strength of schedule is different from how you account for it (however that is), but that doesn't necessarily mean he has an "agenda". You'd have to come up with a lot more evidence for that.
|
|
|
|
|
|
All-In [29049]
TigerPulse: 100%
Posts: 36110
Joined: 8/28/00
|
Not really
Sep 5, 2013, 12:58 PM
[ in reply to That's why it's flawed from inception. He includes a SOS ] |
|
It may well be flawed (I don't know, as I"m not privy to the formula), but it is purely and 100% only an attempt to quantify a subjective ranking. Of course what inputs you choose to use and how you weight them will determine the outcome. During the first few weeks his rankings will move around a great deal because the "old" data falls out and the new data moves in.
It solidifies into what he is attempting to get. It may well be flawed in concept (quantifying the purely subjective), but it's really aimed at predictive behavior. As such it is a tool and nothing more.
His SOS calculations may be flawed, but we don't know that because he doesn't release his calculations.
|
|
|
|
|
|
All-In [25021]
TigerPulse: 100%
Posts: 42972
Joined: 7/31/10
|
38-35. I don't need any guess work, predictions, attempts,
Sep 5, 2013, 1:04 PM
|
|
concepts... subjective or not. There is no conceivable explanation for GA #5 and Clemson #11 after Saturday night,at least not one with any credibility.
|
|
|
|
|
|
Orange Blooded [4787]
TigerPulse: 100%
Posts: 9986
Joined: 6/12/10
|
i really don't like math !
Sep 5, 2013, 1:10 PM
|
|
i'll just sit back and watch this play out.
|
|
|
|
|
|
Mascot [22]
TigerPulse: 100%
Posts: 13
Joined: 9/9/12
|
Not sure if I'm reading this right but,
Sep 5, 2013, 1:23 PM
|
|
Tennessee's strength of schedule is ranked 198th (arguably one of the, if not the toughest schedule period IMHO) and Rice's schedule is 2nd? Is that what this is saying?
|
|
|
|
|
|
All-In [29049]
TigerPulse: 100%
Posts: 36110
Joined: 8/28/00
|
Their SOS is only who they have played so far
Sep 5, 2013, 1:37 PM
|
|
In this ranking it is only Austin Peay
|
|
|
|
|
|
All-Conference [442]
TigerPulse: 97%
Posts: 187
Joined: 11/30/98
|
He uses Bayesian analysis
Sep 5, 2013, 1:30 PM
|
|
Bayesian analyses is used in Engineering and statistics.
In the absence of actual data (this year) he uses generic data (last years ranking). As more actual data is gained, the weighting of the analysis shifts more to the actual data and less for the generic data. Bayesian analysis would say Clemson beating Georgia is a random event that does not predict future results. As more games are played (by all the teams), this years actual results count for more and it becomes more representative of reality.
5 games from now it may show that Georgia sucks. If SC destroys Georgia you can expect Sagarin to put Clemson even lower. Conversely if Georgia destroys SC then Clemson will move up qite a bit.
|
|
|
|
|
|
Legend [19352]
TigerPulse: 92%
Posts: 22266
Joined: 4/25/04
|
Re: Uh whuuuuut.....
Sep 5, 2013, 2:36 PM
|
|
I hate to do this but Prod is 100% correct in this thread.
|
|
|
|
|
|
Replies: 69
| visibility 1
|
|
|
50
