A flaw in strategy

[skibum]

G

Here you go man:

View attachment 24382

I use Football Outsiders F/+ as metric for performance and only look at P5 teams because of the schedules. This graph looks a litle different every year, but the conclusions you can draw from it are the same. If you want a chance at the playoffs better have a top 30 recruiting class. Outside a recruiting class rank of 30 good luck telling me anything.

Recruiting is important. My point here is to say that after you get past the classes filled with the blue-chipers and into the 'dark-matter' that are three and two star recruits, the recruiting ranks from the recruiting services aren't equal to the 'true' recruiting rank. Colorado, OSU, WSU, Utah, Arizona are all given five year average recruiting class ranks between ~40 and 60 by these services. From the data above I can't see how one can say we are doing any better or worse than these programs in recruiting.

Damnit, you weren't supposed to actually do the work, because now I'm going to have to work to prove you wrong (or be forced to admit you're right).

Could you link me to the source of your raw data and / or tell me where you compiled it from?

(I'm going to try and expand the data set a little bit (add more years), and run a slightly different analysis than a straight up regression - I also may use season end ELO for the performance axis rather than F/+, but they should theoretically track somewhat close).

Won't be able to touch anything till late tonight, but I am willing to wrong...

 

[lawdogg]

I was told there would be no math.

 

[jkppkjjkp]

I was told there would be no math.

I guess you got lucky, that's like showing up to a party and saying "I was told there would be no booze," math is everywhere and it's freaking awesome!

 

[lvbuff]

BRING BACK THE WISH BONE!

 

[4BienemyTDs]

BRING BACK THE WISH BONE!

The Wishbone is gone, but the concepts can be recycled and creatively used in a dual threat offense that would be effective in the PAC 12

https://www.sbnation.com/college-football/2017/11/24/16694490/ucf-usf-touchdown-2017

 

[WarBuff]

BRING BACK THE WISH BONE!

I would like to see the I-bone return occasionally. That had a little more flexibility.

 

[4BienemyTDs]

I would like to see the I-bone return occasionally. That had a little more flexibility.

I advocated for the Full House backfield, which inverts the old wishbone into a modern pistol version. Provides for so much flexibility and pretty much looks like an I-Bone just in a more balanced format. We have run it a few times in each game and you could be so creative out of it either running, screen game, or play action into long passes, and it would be easier to have a similar formation used on each play because then the defense has less idea what you are doing. If you change formations, and then have similar tendencies out of those formations, you are not ahead of the defense. Pulling linemen is another way to confuse the defense.

 

[WarBuff]

I advocated for the Full House backfield, which inverts the old wishbone into a modern pistol version. Provides for so much flexibility and pretty much looks like an I-Bone just in a more balanced format. We have run it a few times in each game and you could be so creative out of it either running, screen game, or play action into long passes, and it would be easier to have a similar formation used on each play because then the defense has less idea what you are doing. If you change formations, and then have similar tendencies out of those formations, you are not ahead of the defense. Pulling linemen is another way to confuse the defense.

“Confuse the defense”? Let’s not get too carried away here! Why do that when we can run the QB into the backs of the OL repeatedly for minimal gains.

 

[8Jah]

G

Damnit, you weren't supposed to actually do the work, because now I'm going to have to work to prove you wrong (or be forced to admit you're right).

Could you link me to the source of your raw data and / or tell me where you compiled it from?

(I'm going to try and expand the data set a little bit (add more years), and run a slightly different analysis than a straight up regression - I also may use season end ELO for the performance axis rather than F/+, but they should theoretically track somewhat close).

Won't be able to touch anything till late tonight, but I am willing to wrong...

Sorry. Was out of pocket for few days.

The 5-year average class rankings come from:

https://www.sbnation.com/college-fo...tball-recruiting-rankings-2017-multiple-years

The f/+ ratings you can get anywhere. The direct source is football outsiders:

http://www.footballoutsiders.com/stats/fplus

I used the ratings from two weeks ago. The updated ones came out earlier this week...pretty much the same but Wisconsin moved from #6 to #4 and there were some other small movements.

This is kindof fun and I am going to do some other things too.

 

[skibum]

Sorry. Was out of pocket for few days.

The 5-year average class rankings come from:

https://www.sbnation.com/college-fo...tball-recruiting-rankings-2017-multiple-years

The f/+ ratings you can get anywhere. The direct source is football outsiders:

http://www.footballoutsiders.com/stats/fplus

I used the ratings from two weeks ago. The updated ones came out earlier this week...pretty much the same but Wisconsin moved from #6 to #4 and there were some other small movements.

This is kindof fun and I am going to do some other things too.

Thanks. I'm heading out of the country for 10 days, and I'm not bringing my laptop with me (F that). I'll pick it back up when I get back.

(one thing I've been thinking about is instead of doing an "average rank over 4 or 5 years is doing a sum of stars - or better yet the sum of the numerical rating 3.4 vs 3.7, etc for the past 4 or 5, or even 4.5 years - assuming the data is easy to pull together. There's also a grouped graph analysis that I can't remember what it's called, but it does a better job imo of handling outliers (both good and bad).

Anyway, won't return to it until the 11th at the earliest.

 

[8Jah]

Thanks. I'm heading out of the country for 10 days, and I'm not bringing my laptop with me (F that). I'll pick it back up when I get back.

(one thing I've been thinking about is instead of doing an "average rank over 4 or 5 years is doing a sum of stars - or better yet the sum of the numerical rating 3.4 vs 3.7, etc for the past 4 or 5, or even 4.5 years - assuming the data is easy to pull together. There's also a grouped graph analysis that I can't remember what it's called, but it does a better job imo of handling outliers (both good and bad).

Anyway, won't return to it until the 11th at the earliest.

I've wanted to do sum of stars across all the recruiting services as well but don't want to pay for the data. Wish AB could archive this. Another thing you run into is what value to assign all the NRs in lower ranked recruiting classes.