Scheduling

[jeffgt14]

They would be 6-5 in games against teams not named Alabama, which is a winning record.

This is my favorite stat of the whole thread.

 

[smathis30]

lol at this guy legitimately arguing .500 is a winning record.

6-5 outside the game is a winning record. 6 wins is also a very significant number for college football and its dumb to ignore it. College basketball seeding involves looking at how you played against top 50 and top 100 teams. They are rather arbitrary cutoffs but they are important cutoffs.

 

[smathis30]

This is my favorite stat of the whole thread.

im not the one arbitrarily choosing 7 wins.

 

[jeffgt14]

Note to self. Only cherry pick stats that shine a better light on Bama.

 

[ramblinwreck1378]

6-5 outside the game is a winning record. 6 wins is also a very significant number for college football and its dumb to ignore it. College basketball seeding involves looking at how you played against top 50 and top 100 teams. They are rather arbitrary cutoffs but they are important cutoffs.

I'm not arguing 6 is insignificant. I'm saying 6-6 is not a winning record. You can't just remove L's from someone's schedule and then say they have a winning record.

 

[smathis30]

I'm not arguing 6 is insignificant. I'm saying 6-6 is not a winning record. You can't just remove L's from someone's schedule and then say they have a winning record.

yeah but even with 6 wins alabama still would have the best resume of all the teams

 

[Supersizethatorder-mutt]

I don't know if anyone has already said this or not, but IMO criteria or not, I would be willing to bet that when it came right down to it, the committee chose the teams they WANTED to be there.

 

[RonJohn]

its important because teams like south Florida who go 9-2 but not beat a single bowl elgible team get brought back to earth. Football doesn't have enough games to really do transsitive property like basketball can, so comparing schedules and results like that is a far better metric. Strength of schedule is useful, but doesn't tell the whole picture (e.g. results)

So I guess you don't know how it is calculated either? That means that your all important measurement is something that you don't even understand. SOS does take into account the teams you play. USF is 90th in SOS. Easy to understand, easy to apply. SOR is something made up that isn't even explained well. They could just throw darts at a poster to pick them and you wouldn't be able to tell if it fits some formula or not.

 

[RonJohn]

This is my favorite stat of the whole thread.

But six wins only counts as a winning record if you played against Alabama. If you played OSU or Wisconsin or USC you need to have 9 wins before you are considered a winning team.

 

[RonJohn]

But six wins only counts as a winning record if you played against Alabama. If you played OSU or Wisconsin or USC you need to have 9 wins before you are considered a winning team.

If a team were to play Alabama and OSU, then they would count as a winning team against Alabama, but not OSU because losses against Alabama don't count if you are looking at Alabama, but losses against Alabama and OSU both count if you are looking at OSU. It is very clear logic that the committee has been consistent about in the entire 102 years of the playoffs.

 

[smathis30]

If a team were to play Alabama and OSU, then they would count as a winning team against Alabama, but not OSU because losses against Alabama don't count if you are looking at Alabama, but losses against Alabama and OSU both count if you are looking at OSU. It is very clear logic that the committee has been consistent about in the entire 102 years of the playoffs.

I originally had it so it was "team" and not just alabama but Ohio state and wisconsin both didn't play any teams with 6 wins, except Purdue, which was included in the original 3-1 rankings. If you include g5, wisconsin is 5-1, but Alabama is 7-1. Ohio moves to 5-2. But hey people trying to pull at hairs to make the SEC look bad on a GT forum isn't out of the norm. Alabama is still the best against bowl elgible teams, so SoS is irrelevant as bama has the best SoR

 

[RonJohn]

Alabama is still the best against bowl elgible teams, so SoS is irrelevant as bama has the best SoR

I have to point out again that SoR can't be explained other than general best against their record. What is taken into account? How are each of the items factored against each other? How can it be verified that whatever rankings are listed actually meet the formula vs being arbitrarily modified to meet the needs of ESPN?

 

[smathis30]

I have to point out again that SoR can't be explained other than general best against their record. What is taken into account? How are each of the items factored against each other? How can it be verified that whatever rankings are listed actually meet the formula vs being arbitrarily modified to meet the needs of ESPN?

Takes into account the strength of schedule and the infield performance of the team and compares it to how a pretty good (average top 25 team) would do in it. Going 10-2 with 3 tough games and 9 very easy games is expected. It basically takes out outliers that distort SoS and normalizes it to a projected win total compared to actual win total that can be actually useful in comparing different teams schedules. SoS is useful, but like ELO style ratings, is succeptbile to outliers that can make it look way worse than it actually is. (Wisconsins SoS before and after their game aginady tOSU jumped over 10 spots) ESPN isn't the only one that does it. ELo style rankings do similar things in a meaningful manner but still have their flaws. SoS is in general a better metric than SoS because it allows for comparisons that mean something and takes into account wins and losses

 

[RonJohn]

Takes into account the strength of schedule and the infield performance of the team and compares it to how a pretty good (average top 25 team) would do in it.

In other words, no technical description of how it is done. If OSU was actually number 4 based on the algorithm and Alabama was number 7 based on the algorithm, there is no way to know that ESPN didn't just show the wrong number. You only gave ESPN's description of SoR, not anything that a statistician could look at to see if it makes any sense. Even if it did make some statistically sense, there is no guarantee that any average top 25 team would have done against any schedule. How does it factor in injuries or weather or turf-vs-grass or any of a long list of factors? As far as I am concerned it is an arbitrary formula using arbitrary variables with arbitrary factors to those variables. If the result doesn't fit ESPN's narrative, they can just change the factors since nobody actually knows what the variables or the factors are.

 

[smathis30]

In other words, no technical description of how it is done. If OSU was actually number 4 based on the algorithm and Alabama was number 7 based on the algorithm, there is no way to know that ESPN didn't just show the wrong number. You only gave ESPN's description of SoR, not anything that a statistician could look at to see if it makes any sense. Even if it did make some statistically sense, there is no guarantee that any average top 25 team would have done against any schedule. How does it factor in injuries or weather or turf-vs-grass or any of a long list of factors? As far as I am concerned it is an arbitrary formula using arbitrary variables with arbitrary factors to those variables. If the result doesn't fit ESPN's narrative, they can just change the factors since nobody actually knows what the variables or the factors are.

1. ESPN gets the average offenseive and defensive ranking for each team. Likely FPI. Its propietary information so its not public how they calculate it exactly, but they state what factors are involved on their website

2. ESPN goes through a teams schedule with location of game in mind and compares win probability from that average top 25 team to the team on the schedule.

3. Add up the game my game probabilities.

4. Compare each teams wins to the expected wins of a top 25 team.

5. Add the difference to how many wins you have

6. Rank your team

I do something similar to my own computer formulas. Like i said, ELO style rankings do the same thing and it takes like 3 seconds to google it. It doesn't account for turf/grass or all that but its better than strength of schedule. its a single metric. you can look at other metrics (which greatly value Alabama over the other teams in consideration)
These are my ELO style rankings for the top 25 teams. Its an ELO style ranking + teams record (wisconsin would add 12.01 to their record). Its the average score of a game where the points for a game are pts/ttlpointsingame*(131-opp rank) if you win a -1*ptswinningteam/ptstotal*opprank. Def is way more generous towards undefeated teams than not.

NAME rank 2 POINTS
CLEMSON 1 67.74
UGA 2 63.94
WISCONSIN 3 59.11
BAMA 4 57.34
tOSU 5 55.45
PSU 6 53.71
UCF 7 52.17
OU 8 51.10
ND 9 50.32
MIAMI 10 48.29
WASHINGTON 11 48.02
USC 12 47.66
AUBURN 13 46.21
MSU 14 39.66
STANFORD 15 39.44
TCU 16 38.01
VT 17 37.27
NW 18 35.70
WASHINGTON STATE 19 33.55
MU 20 33.04
LSU 21 32.88
MEMPHIS 22 32.67
MISS STATE 23 31.44
IOWA 24 31.19
BC 25 30.90

Here is my other ranking system which doesn't take winning or losing into consideration. Just on the field results and game location. Its a lot closer to what the committe had. It overvalues the Big12, which im fixing next year. But it gives an idea. In short, Bama and Ohio State are both good but alabama has a slight edge. Wisconsin is good but not great. (ignore hyperlinks they are all to my excel document)

OU 206.60 1
CLEMSON 203.39 2
BAMA 202.99 3
tOSU 201.63 4
UGA 198.42 5
WASHINGTON 197.02 6
ND 196.93 7
PSU 194.58 8
OK STATE 194.30 9
WISCONSIN 193.20 10
TCU 192.67 11
AUBURN 191.98 12
UCF 190.23 13
USC 189.41 14
LOUISVILLE 189.00 15
STANFORD 186.84 16
WSU 184.32 17
WAKE FOREST 181.32 18
MEMPHIS 181.01 19
OREGON 180.36 20
MIAMI 180.24 21
ARIZONA 180.03 22
VT 179.73 23
BOISE STATE 179.31 24
TEXAS 178.65 25

If you still believe in the ESPN bias here is a non-outlet post about it
https://www.sbnation.com/college-fo...otball-strength-of-schedule-rankings-2017-sos

 

[RonJohn]

1. ESPN gets the average offenseive and defensive ranking for each team. Likely FPI. Its propietary information so its not public how they calculate it exactly, but they state what factors are involved on their website

2. ESPN goes through a teams schedule with location of game in mind and compares win probability from that average top 25 team to the team on the schedule.

3. Add up the game my game probabilities.

4. Compare each teams wins to the expected wins of a top 25 team.

5. Add the difference to how many wins you have

6. Rank your team

1. OK so use rankings(which by the way use factors for things such as how much run defense/offense counts compared to passing, or how much 3rd down defense counts)
2. What level of top 25 team? From which year? Compare to number 12, or a specific team, or just average the probabilities of every top 25 team? There are many variables that can be factored in many different ways at the whim of ESPN.
3,4,5,6 How to the preseason predictions for wins/losses per team compare to actual wins/losses per team? How well does the conference win % midway through the season actually predict the conference winners? How many times per year does a team with an in game win % prediction of less than 20% actually end up winning the game? However you are willing to let such a system that isn't perfect decide who gets in the playoffs. I would rather just let the computer decide who the MNC is and award the trophy now.

If you still believe in the ESPN bias here is a non-outlet post about it
https://www.sbnation.com/college-fo...otball-strength-of-schedule-rankings-2017-sos

You are totally misunderstanding my statements and questions. How ESPN calculates the SoR is totally unknown. It cannot be checked by outside entities and verified. They can change the results at will if they desire. They can change the factors of variables at will to get different results if they desire. Using it as a basis for declaring who the best team or teams are is basically the same as simply deferring to ESPN's whims.

EDIT: By the way you are just guessing at how they do the calculations, because they don't even publish what you displayed. They simply say it predicts how a top 25 team would do against your schedule.

 

[smathis30]

1. OK so use rankings(which by the way use factors for things such as how much run defense/offense counts compared to passing, or how much 3rd down defense counts)
2. What level of top 25 team? From which year? Compare to number 12, or a specific team, or just average the probabilities of every top 25 team? There are many variables that can be factored in many different ways at the whim of ESPN.
3,4,5,6 How to the preseason predictions for wins/losses per team compare to actual wins/losses per team? How well does the conference win % midway through the season actually predict the conference winners? How many times per year does a team with an in game win % prediction of less than 20% actually end up winning the game? However you are willing to let such a system that isn't perfect decide who gets in the playoffs. I would rather just let the computer decide who the MNC is and award the trophy now.



You are totally misunderstanding my statements and questions. How ESPN calculates the SoR is totally unknown. It cannot be checked by outside entities and verified. They can change the results at will if they desire. They can change the factors of variables at will to get different results if they desire. Using it as a basis for declaring who the best team or teams are is basically the same as simply deferring to ESPN's whims.

EDIT: By the way you are just guessing at how they do the calculations, because they don't even publish what you displayed. They simply say it predicts how a top 25 team would do against your schedule.

They can change it, but doesn't change the fact that other people have the same results. Which is why I just posted another source that has similar results. It's called data verification. No need to think it's faked if it can be verified to some degree elsewhere. Scientific method in action

And yes I am guessing how they do it based off what they say but it's how I would do it

 

[dressedcheeseside]

yeah but even with 6 wins alabama still would have the best resume of all the teams

At this point I’m not sure who you’re trying to convince, yourself or the message board.

 

[RonJohn]

They can change it, but doesn't change the fact that other people have the same results. Which is why I just posted another source that has similar results. It's called data verification. No need to think it's faked if it can be verified to some degree elsewhere. Scientific method in action

Then let's just award the trophy to Clemson now. There isn't any point in playing the games. Or while we are at it, let's just skip the entire season and award the trophy to the top rated pre-season team. The NCAA knows how to run championships and they do it for every single sport except FBS football. Money is the only reason that FBS works the way it does. That is going to really hurt the NCAA's case about amateurism when it finally ends up in court. The whole business/gambling/trash talk aspect(s) of NCAA "athletics" is what bothers me.

 

[smathis30]

At this point I’m not sure who you’re trying to convince, yourself or the message board.

Person who made the list of team records included 6 win teams for Wisconsin and not Alabama. I was making sure the statistics were fair and intentionally unbiased