Strength of Schedule (SoS) Debate 2005/2006

[BuckeyeNation27]

Originally Posted by OregonBuckeye

Ohio State-#4 SOS

If you think we had the 4th hardest schedule in the country last year then you're kidding yourself. We played Texas, and the only other tough game was penn state, only because it was at penn state.

[OregonBuckeye]

Originally Posted by BuckeyeNation27

If you think we had the 4th hardest schedule in the country last year then you're kidding yourself. We played Texas, and the only other tough game was penn state, only because it was at penn state.

And just who had tougher schedules than us? Iowa, Minnesota, Penn St., scUM, Texas, and Northwestern are all quality teams. The bad ones were SDSU, Miami, Indiana, and Illinois. Michigan St. is a tough one because they went 5-6 but were still pretty talented. And don't give me any SEC stuff. They make eachother's records look good by playing the Citadel's of college football.

[BuckeyeNation27]

riiiiiiight

[OregonBuckeye]

Originally Posted by BuckeyeNation27

riiiiiiight

Thanks for answering my question.

[BuckeyeNation27]

No problem

[jwinslow]

Originally Posted by OregonBuckeye

And don't give me any SEC stuff. They make eachother's records look good by playing the Citadel's of college football.

'don't give me any Big10 stuff. They make each other's records look good by playing the mac-conference and IU/Ill.' Do you buy that argument? Or do outliers not trash an otherwise strong schedule?

[OregonBuckeye]

I can tell I'm going to take the heat off of NDC for starting this little debate.

Big-10 SOS:

#3 Michigan
#4 Ohio St.
#6 Northwestern
#13 Minnesota
#26 Penn St.
#34 Iowa
#37 Illinois
#46 Purdue
#58 Michigan St.
#61 Wisconsin
#63 Indiana
Average: 31.9 or 32

SEC SOS:

#8 Arkansas
#14 Tennessee
#15 Georgia
#24 South Carolina
#27 Florida
#30 LSU
#33 Auburn
#50 Alabama
#62 Mississippi St.
#65 Kentucky
#74 Vanderbilt
#82 Mississippi
Average: 40.3 or 40

I just want to know why strength of shedule doesn't mean anything and some of the teams that had tougher shedules than Ohio St.

[SanClementeBuck]

I don't agree that SOS is meaningless. Mebby at season's start when a fb team doesn't have a few games under its belt to guage the team's effectiveness. With game 6 or 7 into season, SOS is one of the few measurements that level the playing field. Watch UWVa this year...or Louisville recently: patsy schedules the team rides higher in standings, when they don't deserve a mention.

[Buckeyeskickbuttocks]

I've argued this SOS stuff on message boards since about 2000... In that time I have learned that the "major" teams SOS are all about the same. And, what's the import anyway? Indiana's "weak-ass" schedule sure hasn't helped them.

There are some outliers - I mean, Auburn's 03 Schedule, IIRC, was in the neighborhood of 68, that's not too stiff. But then, how was Auburn's schedule signifigantly different from anyone elses that year? I mean, yeah, they took a game and played nobody state.... but, they did also play LSU, UGA, Florida, Bama, Tenny and such....

Anyway, SOS is simply something for us to bicker about, if you ask me. There is some "relative to itself" validity, but pound for pound, USC's schedule is the same as Ohio State's schedule is the same as Notre Dame's schedule.

Indiana's schedule is the same as Vandy's schedule is the same as Duke's schedule.

the discussion should be about "tiers" not about raw SOS rank.

[BB73]

Originally Posted by Buckeyeskickbuttocks

There are some outliers - I mean, Auburn's 03 Schedule, IIRC, was in the neighborhood of 68, that's not too stiff. But then, how was Auburn's schedule signifigantly different from anyone elses that year? I mean, yeah, they took a game and played nobody state.... but, they did also play LSU, UGA, Florida, Bama, Tenny and such....

I'm guessing you meant to talk about Auburn's schedule in 2004, when they went 13-0 and played the Citadel, LA-Monroe and La-Tech. They played LSU, Georgia, Alabama, and Tennesse twice, but not Florida that year.

[MililaniBuckeye]

The only real time that SOS means much is when you're looking at two similar programs (i.e., all-time powers like Ohio State and Notre Dame) and there is a significant disparity in their rankings (as was this past season, with us at #4 and ND at #52).

[methomps]

Originally Posted by MililaniBuckeye

The only real time that SOS means much is when you're looking at two similar programs (i.e., all-time powers like Ohio State and Notre Dame) and there is a significant disparity in their rankings (as was this past season, with us at #4 and ND at #52).

The only real time that SOS means much is when it helps one's argument.

[MililaniBuckeye]

Originally Posted by methomps

The only real time that SOS means much is when it helps one's argument.

Yeah, well, that, too...

[Buckeyeskickbuttocks]

Originally Posted by BB73

I'm guessing you meant to talk about Auburn's schedule in 2004, when they went 13-0 and played the Citadel, LA-Monroe and La-Tech. They played LSU, Georgia, Alabama, and Tennesse twice, but not Florida that year.

correct

[Steve19]

Isn't this a debate about computer ratings in general?

Most of the computer ratings predict the winner of a contest about 75%-85% of the time. That means that the models don't capture all of the factors that cause winning and losing. Here's some factors that can't be captured by a computer model:

1. Luck. Call it what you will but all teams have randomness in their performance. Lucky plays, a slipping safety on a long pass, a fly that goes into the eye of a quarterback...chance occurences can mean winning or losing.
2. Situational context/influences. Models capture home and away advantages, but they can't capture the extra motivation of a rivalry game or the effect on a team if its best player's mother passes away. If its your down day and their up day and if you're computer ratings are similar, then you will probably lose.
3. Growth curves. Most models these days are Bayesian and that means they use prior probability and the week's performance to calculate the posterior probabilities. Such models can adjust but a team that gets back an injured player or a suspended quarterback coming back into a team can present the model with growing performance that cannot be captured in one or two games.

There are other non-random sources of variation that might cause inaccuracy. This, however, does not detract from the fact that computer models provide an objective measure of team performance.

By random chance, flipping a coin will get us a 50% success rate in choosing winning teams in football games. Computer models improve that success rate by 50% to 70% and generally beat all of the so-called experts we see on the websites.

Over the long run of a season, strength of schedule ratings are an excellent unbiased reflection of the difficulty of opponents that a team faced during the year.