- My Forums
- Tiger Rant
- LSU Recruiting
- SEC Rant
- Saints Talk
- Pelicans Talk
- More Sports Board
- Coaching Changes
- Fantasy Sports
- Golf Board
- Soccer Board
- O-T Lounge
- Tech Board
- Home/Garden Board
- Outdoor Board
- Health/Fitness Board
- Movie/TV Board
- Book Board
- Music Board
- Political Talk
- Money Talk
- Fark Board
- Gaming Board
- Travel Board
- Food/Drink Board
- Ticket Exchange
- TD Help Board
Customize My Forums- View All Forums
- Show Left Links
- Topic Sort Options
- Trending Topics
- Recent Topics
- Active Topics
Started By
Message
re: lsumatt computer poll released
Posted on 11/22/11 at 10:24 am to BhamTigah
Posted on 11/22/11 at 10:24 am to BhamTigah
quote:
This SOS calculation is better than most. I've always thought that for the BCS rankings, only games against top 30 teams should be considered in SOS rankins, unless you lose to a team outside the top 30, in which case it should hit you pretty hard.
The problem with this method is that is if Team A and Team B both play the 1,2,3,4,5 rated teams, but then Team A plays teams 120,119,118,117,116 and Team B plays teams 35,34,33,32,31 - both teams would have identical SOS scores. You'd have to move the arbitrary 30 way up the scale to make the system work, and at that point why have the arbitrary number at all?
Also, a team's wins and losses against their schedule should mean little when it comes to determining how difficult the schedule was. What matters is your opponents records, and your opponent's opponents records, and so on and so forth.
1
Posted on 11/22/11 at 10:34 am to Grilled Bald Eagle
quote:
The problem with this method is that is if Team A and Team B both play the 1,2,3,4,5 rated teams, but then Team A plays teams 120,119,118,117,116 and Team B plays teams 35,34,33,32,31 - both teams would have identical SOS scores.
Not a problem for me. Both teams played 5 teams that should have given them problems. Set the number at 50 if you prefer. Any higher than that and you're getting into really horrible teams.
Another option is to add back a quality win component like they had in the early years. I know this isn't how it is calculated, but a team should get MUCH more credit for beating #1 and #120 that for beating #60 and #61.
Posted on 11/22/11 at 11:12 am to lsumatt
thanks for all you do, matt .. 
Posted on 11/22/11 at 11:19 am to lsumatt
Damn fine work son.
Posted on 11/22/11 at 11:30 am to BhamTigah
quote:
Think about it this way:
1. The only purpose of the BCS is to determine the top 2 teams.
2. If you are in consideration as a top two team, you shouldn't have a problem beating a team outside the top 30, regardless of whether they are #31 or #120.
Top 30 is an arbitrary #. Maybe it should be 40 or whatever. The point is that after a certain point, you shouldn't get any more credit for beating 1 bad team over another,
The problem with this logic is that there's a HUGE difference between playing the #30 team and the #120 team. This year the #30 team could be teams like Baylor, Florida, Auburn and Iowa State while the #120 teams are on the level of Troy, Western Kentucky and Kent State. Teams like LSU can sleep walk through games against the #120 team but not against the #30 team. Oklahoma State would still be undefeated if they had played the #120 team last Friday night.
Posted on 11/22/11 at 11:33 am to BhamTigah
quote:
Not a problem for me. Both teams played 5 teams that should have given them problems. Set the number at 50 if you prefer. Any higher than that and you're getting into really horrible teams.
It can't be a black or white thing. You can't give one team credit for beating the #50 team and give another team no credit for beating the #51 team.
Posted on 11/22/11 at 11:37 am to lsumatt
quote:
The above curve can be described by a “sigmoid” equation:
I now have a far clearer understanding of the derivation of the term "sigmoidoscopy." And I don't think it's just based on the shape of the curve!!
Good job. Now just use your math to keep LSU in the BCSCG.
Posted on 11/22/11 at 12:16 pm to BhamTigah
quote:
Not a problem for me. Both teams played 5 teams that should have given them problems. Set the number at 50 if you prefer. Any higher than that and you're getting into really horrible teams.
So you're seriously telling me that a team going 10-0 against the following schedules should be rated identically?
Team A: wins over 1,2,3,4,5,116,117,118,119,120
Team B: wins over 1,2,3,4,5,51,52,53,54,55
You're using arguments like "should have given them problems" and "any higher and you're getting into some really horrible teams", which are arguments that should be taken into account by human polls. Computer polls exist so that you can look at results objectively and with no bias.
Posted on 11/22/11 at 12:26 pm to Grilled Bald Eagle
quote:
So you're seriously telling me that a team going 10-0 against the following schedules should be rated identically?
Team A: wins over 1,2,3,4,5,116,117,118,119,120
Team B: wins over 1,2,3,4,5,51,52,53,54,55
Yes. My original statement of top 30 was probably too low, but I would stand behind top 50. If you claim to be the best, should #50 really be an issue?
As I said, perhaps the better method would be to keep SOS as is but add back a component for quality wins. A team should be highly rewarded for beating elite teams, not so much for beating average teams they were expected to beat.
Posted on 11/22/11 at 12:28 pm to BhamTigah
quote:Very good point.
This SOS calculation is better than most. I've always thought that for the BCS rankings, only games against top 30 teams should be considered in SOS rankins, unless you lose to a team outside the top 30, in which case it should hit you pretty hard.
Posted on 11/22/11 at 12:28 pm to lsumatt
You sure do know your math.
I like it.
I like it.
Posted on 11/22/11 at 12:30 pm to Grilled Bald Eagle
You can make an argument against any SOS calculation. Which schedule would you say is harder:
Team A plays 55, 5, 54, 4, 53, 3, 52, 2, 51, 1
Team B plays 116, 4, 3, 117, 2, 5, 1, 118, 119, 120
Team A plays 55, 5, 54, 4, 53, 3, 52, 2, 51, 1
Team B plays 116, 4, 3, 117, 2, 5, 1, 118, 119, 120
Posted on 11/22/11 at 12:32 pm to BhamTigah
quote:
Yes. My original statement of top 30 was probably too low, but I would stand behind top 50. If you claim to be the best, should #50 really be an issue?
you are arguing from a pure black and white point of view. It is far more realistic to admit one of the starting premises from the OP.
Whenever two teams play on the same field there is some chance of an upset. His sigmoidal curve visually represents that probability and his equations model/calculate the same.
Yes, the chance is reduced in cases of a mismatch but it will never fall to zero and the argument that a much better team will never lose to a much worse team has been refuted as recently as this last weekend (OkS/ISU, OU/Baylor)
Big upsets are obviously a big part of college football and your assumption that they never happen is outright wrong
Posted on 11/22/11 at 12:33 pm to BhamTigah
quote:
Yes. My original statement of top 30 was probably too low, but I would stand behind top 50. If you claim to be the best, should #50 really be an issue?
Of course #50 shouldn't be of issue to a top team. The problem is that a ratings system has to calculate the ratings of EVERY team, not just that of the top 25 teams. Because of that, you have to count every game against every team, or else the results aren't useful.
Also, I think you're confusing SOS rating and power ratings - SOS doesn't count a team's wins and losses at all, only the wins and losses of a team's opponents (and those of their opponents and so on). Power ratings are largely dependant on wins and losses, game locations, and (for many of the systems) the scores of those game.
Posted on 11/22/11 at 12:39 pm to CubsFanBudMan
quote:
Which schedule would you say is harder:
Team A plays 55, 5, 54, 4, 53, 3, 52, 2, 51, 1
Team B plays 116, 4, 3, 117, 2, 5, 1, 118, 119, 120
Team A's schedule in that case would be tougher. Using the average of 110 rating systems:
Team A plays: LSU, Alabama, Oklahoma St., Oklahoma, Stanford, Louisiana Tech, Tennessee, Washington, UCLA, and South Florida
Team B plays: LSU, Alabama, Oklahoma St., Oklahoma, Stanford, Tulane, New Mexico, Memphis, Florida Atlantic, and Akron
The real issue is that it doesn't matter what I or anyone else thinks of how hard a schedule is. Math has no bias. Feed it good numbers and you'll have good, solid, reliable results.
Posted on 11/22/11 at 12:40 pm to lsumatt
All those equations reminds me of my three year old's bath time when he puts all of his foam letters and numbers on the side of the tub! 
great work Matt!
great work Matt!
Posted on 11/22/11 at 12:44 pm to Grilled Bald Eagle
quote:
Team A's schedule in that case would be tougher. Using the average of 110 rating systems:
So you'd rather play #2, 5, and 1 in consecutive weeks with 118, 119, and 120 following than have a middle of the road team in between your elite teams?
Posted on 11/22/11 at 12:48 pm to lsumatt
Serious question matt. What happens if Oklahoma beats Okie State. Who gets shoe horned in ahead of #2 Alabama then? 
This post was edited on 11/22/11 at 12:49 pm
Posted on 11/22/11 at 12:51 pm to CubsFanBudMan
quote:
So you'd rather play #2, 5, and 1 in consecutive weeks with 118, 119, and 120 following than have a middle of the road team in between your elite teams?
Do you have any evidence - "everyone knows it" doesn't count as evidence, by the way - that playing tough teams back to back is more likely to lead to a loss than playing the same teams separated by a week? Because if you do, it should be reflected in the formula. If you don't, it shouldn't.
And the idea that playing the #51 team and the #120 team is the same is mind-boggling. lsumatt, what's the expected win % for #1 v. #51 against #1 v. #120
Posted on 11/22/11 at 12:56 pm to CubsFanBudMan
quote:
So you'd rather play #2, 5, and 1 in consecutive weeks with 118, 119, and 120 following than have a middle of the road team in between your elite teams?
Do you have a reasoned, mathematical method to account for this sort of thing in computing SOS? If not, its irrelevant to the situation.
Page 5 of 6
Popular
Back to top
Follow TigerDroppings for LSU Football News