Worst Coach Rankings

Kyle,
If you want to do this right, you need to do a regression. I don't know what factors will be significant, but here is a guess at some you could include:

coach all time W/L
national championships
conference championships
times ranked in top 10
times ranked in top 25

Improvement Factors
coach winning W/L at current school
school all time W/L without coach
school all time W/L before coach
school all time W/L after coach
record vs top 25
record vs top 10
record vs top 25 - before coach
record vs top 10 - before coach

Trending Factors along with SOS Correction
Last Year W/L
Two Years ago W/L
Three Years ago W/L
Strength of Schedule (last 3 yrs)
Opponents Strentgth of Schedule (last 3 yrs)

Just for fun
Recruiting Rankings - Current Team (average of last 7 years)
Stadium Size
Annual Salary (might be a better indicator of a good coach than you'd think)
 
kyle, have you thought about becoming a minister?

you would probably be better at it than this. yes, i know you hate religion, still this is embarrasing for you!
 
gth816f said:
Uh...what sport relies on computers for anything, besides that they are 1/3 of the BCS formula(which has never come into play since they changed it to only be 1/3)?
Click to expand...

Since we're talking about football coaches...that's what I was referring to.

The fact that computers are even a part of the process at all IMO is a problem

This game is played by people and people should be the only contributing factor as to who plays where, who's ranked what, etc

The fact that we have someone on here developing a computer program to determine which coaches are better than others and vice versa is ridiculous.
 
ncjacket said:
What the HELL are you talking about? gth made a comment asking what sport relies on computers, Kyle replied with sabremetrics. I simply said baseball doesn't rely on them, it's just a way some people measure things. Kyle then referred to Moneyball, which is a point, but only Oakland really pays that much attention to it. I don't know what you have to do with any of this.

Oh, and find one instance of me editing a post to change the meaning....just one.
Click to expand...

My bad slick. The only part that was referring to yourself was me THANKING YOU.

All the other stuff was directed at Kyle....didn't mean to get your feathers ruffled
 
TechPhi97 said:
It looks like the formula you developed is to take the previous winning percentage (at the school, I guess) and compare the coach's winning percentage to that number, then multiple by 100. Is that right?

If so, I wouldn't really call it a model, I'd just call it a comparison. I guess you can call it a model, but I think of models as being predictive rather than backward-looking.

My first thought was that Bill Lewis would've looked great in your rankings, which kind of invalidate the ranking system. I wouldn't call the list "worst coaches", because a coach is usually evaluated on a body of work. Maybe "recent coaching success rates" or something like that.

By the way, the A's don't use all of the evaluation analytics like they used to, because once the book was published they lost any edge that they gained by doing it.
Click to expand...

That's not the model.

First of all, let me explain what the rankings are actually measuring. The rankings are a measure of how much pressure is on the coaches to win. It's NOT a measure of their legacy. Maybe "worst coaches" is a bad title.

Here is the exact method:

Linearly interpolated the win percentage over the tenure of the coach's career. Obtained the expected value using an exponential distribution function with lamba = 1. This was called the expected winning percentage (EWP).

This winning percentage over the previous 2y years was subtracted from the EWP, where y is the length of the coach's tenure in years. This was multipled by 1000 to obtain the points. Additional points were added for national championships and for first and second year coaches.
 
Geori said:
Kyle,
If you want to do this right, you need to do a regression. I don't know what factors will be significant, but here is a guess at some you could include:

coach all time W/L
national championships
conference championships
times ranked in top 10
times ranked in top 25

Improvement Factors
coach winning W/L at current school
school all time W/L without coach
school all time W/L before coach
school all time W/L after coach
record vs top 25
record vs top 10
record vs top 25 - before coach
record vs top 10 - before coach

Trending Factors along with SOS Correction
Last Year W/L
Two Years ago W/L
Three Years ago W/L
Strength of Schedule (last 3 yrs)
Opponents Strentgth of Schedule (last 3 yrs)

Just for fun
Recruiting Rankings - Current Team (average of last 7 years)
Stadium Size
Annual Salary (might be a better indicator of a good coach than you'd think)
Click to expand...

I guess I am assuming SOS on average doesn't change much between coaches so that can be neglected.

Also, all time winning percentage is a bad indicator because coaches are usually just compared to previous eras.

Good ideas about recruiting rankings and stadium size. I could even include ticket sales.
 
Oh God, why can't the season just begin so we stop having stupid arguments of why someone's rankings suck...

Oh wait, that'll happen anyway.
 
Xihix said:
Oh God, why can't the season just begin so we stop having stupid arguments of why someone's rankings suck...

Oh wait, that'll happen anyway.
Click to expand...

The funny thing is nobody has really pointed out why the rankings are that bad. People just don't like me so they just bash anything I say.

Look at the top 10 worst and top 10 best and tell me that is not at least somewhat accurate. If you take out the 1st and 2nd year coaches I think it is very accurate.
 
I just think there's a real life factor lost with these types of rankings. I don't care either way, I don't really mind you as much as other people.
 
Xihix said:
I just think there's a real life factor lost with these types of rankings. I don't care either way, I don't really mind you as much as other people.
Click to expand...

I think I just labeled it wrong. It's not really the "worst coaches", but rather coaches that are under pressure to win.
 
GTKyle said:
That's not the model.

First of all, let me explain what the rankings are actually measuring. The rankings are a measure of how much pressure is on the coaches to win. It's NOT a measure of their legacy. Maybe "worst coaches" is a bad title.

Here is the exact method:

Linearly interpolated the win percentage over the tenure of the coach's career. Obtained the expected value using an exponential distribution function with lamba = 1. This was called the expected winning percentage (EWP).

This winning percentage over the previous 2y years was subtracted from the EWP, where y is the length of the coach's tenure in years. This was multipled by 1000 to obtain the points. Additional points were added for national championships and for first and second year coaches.
Click to expand...

Not to make this a math argument, but why linear interpolation? That is used to fill in gaps in data, right?

Think about modeling it by using regression, like a previous poster said. Then you'll effectively have a target winning percentage that you can compare the actual WP to in order to make a comparison.
 
TechPhi97 said:
Not to make this a math argument, but why linear interpolation? That is used to fill in gaps in data, right?

Think about modeling it by using regression, like a previous poster said. Then you'll effectively have a target winning percentage that you can compare the actual WP to in order to make a comparison.
Click to expand...

That's fine. I don't mind talking about the math behind it because it's at least constructive. I linear interpolated because my probability density function is continuous and my data is discrete. If you know of a way to generate a variable-size stochastic vector with similar properties I am all ears. I will look into regression. Thanks for the input.
 
I did point out that there are at least some question marks about the 1st/2nd year coaches being at the top and bottom. But, enough of that.

Referring to including stadium size is an interesting point. This is not really related, but I think it is interesting: I was talking with the Assistant AD at Clemson and he was telling me that there is a direct correlation b/w how well Clemson Football does one year (relative to South Carolina) and their average SAT score for all entering students at Clemson the next year. He said over the past ten years it is pretty much always the case--more wins = higher score. His theory is that you have a limited pool of South Carolina high school kids that do not have an allegiance to either school and at least a good portion of them consider the experience (going to games, watching a better team, etc.) in their decision. Increased applications = higher SAT score. Just think it's interesting.

And, I'm fairly new here and have no feeling as to whether I like you or not. I was only commenting on the model, not you.
 
I did point out that there are at least some question marks about the 1st/2nd year coaches being at the top and bottom. But, enough of that.

Referring to including stadium size is an interesting point. This is not really related, but I think it is interesting: I was talking with the Assistant AD at Clemson and he was telling me that there is a direct correlation b/w how well Clemson Football does one year (relative to South Carolina) and their average SAT score for all entering students at Clemson the next year. He said over the past ten years it is pretty much always the case--more wins = higher score. His theory is that you have a limited pool of South Carolina high school kids that do not have an allegiance to either school and at least a good portion of them consider the experience (going to games, watching a better team, etc.) in their decision. Increased applications = higher SAT score. Just think it's interesting.

And, I'm fairly new here and have no feeling as to whether I like you or not. I was only commenting on the model, not you.
Click to expand...

I've heard multiple times that when a school wins a championship in football or basketball (I always feel as if this is said during the final four, so maybe it is more correlated with basketball), the number of applicants increases by a fair amount.
 
As another aside, I was told by the AD at Clemson that although they felt like spending $100k on plasma screen tv's and XBox's for the locker room and weight room was a waste of money, it would drastically improve their image. It's funny how recruits, as a whole, can change the choices of athletic programs so much. I know this is a useless bit of info, but interesting nonetheless.
 
GTKyle said:
That's fine. I don't mind talking about the math behind it because it's at least constructive. I linear interpolated because my probability density function is continuous and my data is discrete. If you know of a way to generate a variable-size stochastic vector with similar properties I am all ears. I will look into regression. Thanks for the input.
Click to expand...

Regression is just a different technique. You would look for varialbes that are significantly correlated to winning percentage. Once you had all the variables and coefficients, you would then use that as an "expected" winning % for a coach, and you could compare actuals to get the variance from expected.

I'm not familiar with the methods you're using, because it's been 15 years since my IE classes where we did that stuff.
 
GTKyle said:
I think I just labeled it wrong. It's not really the "worst coaches", but rather coaches that are under pressure to win.
Click to expand...


Now that's funny!!!

Attention everyone, Attention! GTKyle's been hard at work in the lab

And the product of his due diligence is this Coaching Formula he's created

Not only does it rank coaches....

His Formula that He has Derived also determines "who's under pressure"

Awesome Job Kyle, I'm sure they'll be waiting eagerly to hear from you to let them know that they are now under pressure due to your Mathematical/Computing Genious.
 
You must log in or register to reply here.
Back
Top