Should we replace with Zeke with Marcel Reece?

[waldoputty]

Lmao I'll take him to come be our fullback

me too

 

[T-RO]

New stat is - ANRY/A - Adjusted Net Rushing Yards per Attempt...
I bet the correlation to winning is much greater than 45% when a team's ANRY/A is better than their opponents.......

You do understand that a correlation of 50% in this type of discussion is no real correlation at all?

The team with a higher ANRY/A wins 45%?!? LOL.

45% would be an INVERSE correlation, proving the opposite.


It's becoming apparent that you and Waldo are true masochists. I do hand it to you guys...you're both good at taking a pounding!

 

[Idgit]

Actually sarcasm is denoted by a drawl in tone. It's the equivalent of doing that. If you deadpan sarcasm then many people see that as backbiting.

Intelligence has little to do with it. Autism spectrum and similar disorders which range in intelligence from both ends by definition struggle to pick it up. There are all different forms of nuance and many of them are completely subjective. It's less discerning than you think.

Is it bad that I can't tell if you're serious or not?

 

[Proof]

Actually sarcasm is denoted by a drawl in tone. It's the equivalent of doing that. If you deadpan sarcasm then many people see that as backbiting.

Intelligence has little to do with it. Autism spectrum and similar disorders which range in intelligence from both ends by definition struggle to pick it up. There are all different forms of nuance and many of them are completely subjective. It's less discerning than you think.

 

[dmax]

You do understand that a correlation of 50% in this type of discussion is no real correlation at all?

The team with a higher ANRY/A wins 45%?!? LOL.

45% would be an INVERSE correlation, proving the opposite.


It's becoming apparent that you and Waldo are true masochists. I do hand it to you guys...you're both good at taking a pounding!

Howdy! Never posted but now is as good a time as any. I am a professor that teaches statistics. A couple of things from this and the previous thread have interested me.

First, correlations can range from -1 to 1. So a correlation or .45 is not an inverse relationship. That would be -.45. Correlation of .45 is simply moderately correlated.

Second, one of the basic precepts of statistics is that correlation is not causation. That means that just because two variables are highly correlated, does not mean that one causes the other. Correlation is useful in prediction. Therefore, Adam can certainly say that his passing stat can predict winners at a high rate, but he cannot say that having a high passing efficiency causes winning. To determine causation you have to use something like an autoregrssive or error correction model.

Don't know if this is useful, but thought I'd make the point.

 

[FuzzyLumpkins]

Is it bad that I can't tell if you're serious or not?

Bad is a relative term. I dunno. Is it?

Nah seriously I think the lack of social cues like tone on the interwebs leads to a large part of the ugliness that we see.

 

[Nightman]

You do understand that a correlation of 50% in this type of discussion is no real correlation at all?

The team with a higher ANRY/A wins 45%?!? LOL.

45% would be an INVERSE correlation, proving the opposite.


It's becoming apparent that you and Waldo are true masochists. I do hand it to you guys...you're both good at taking a pounding!

those are your boys Adam's numbers for YPC .......keep up the cheerleading though, it looks good on you

 

[Nightman]

It's becoming apparent that you and Waldo are true masochists. I do hand it to you guys...you're both good at taking a pounding!

And you are a proud member of the Running Back deniers club.......after 2014 and the start to this season one group looks really stupid but keep coming back for more

 

[waldoputty]

Howdy! Never posted but now is as good a time as any. I am a professor that teaches statistics. A couple of things from this and the previous thread have interested me.

First, correlations can range from -1 to 1. So a correlation or .45 is not an inverse relationship. That would be -.45. Correlation of .45 is simply moderately correlated.

Second, one of the basic precepts of statistics is that correlation is not causation. That means that just because two variables are highly correlated, does not mean that one causes the other. Correlation is useful in prediction. Therefore, Adam can certainly say that his passing stat can predict winners at a high rate, but he cannot say that having a high passing efficiency causes winning. To determine causation you have to use something like an autoregrssive or error correction model.

Don't know if this is useful, but thought I'd make the point.

That sounds good to me.

Care to weigh in on the "rushing well does not matter" argument in light of defense formations that take away the run such as 8-in-the-box?
I could supply the details if you have not seen it - this is from another thread:


To me, running ability for a RB is defined by qualities like what I mentioned above such as:
1. breakaway speed - measurable
2. wiggle - mostly measurable
3. vision - mostly measurable but who is going to let you
4. toughness - hard to measure
5. ability to flatten the defender - somewhat measurable in terms of weight, weight distribution
6. decision making - hard to measure. this does improve at the beginning of a career while learning to play at NFL speed
7. probably left out a couple
8. Injuries can obviously change a RB's ability

Running ability with a team needs to include other things like OL's ability to block, QB/WR/OL ability to pass to make the defense less focused on the run. These also don't vary quarter to quarter except in the case of change in personnel and injuries.

A few other things from the other thread:

1) I mentioned this EXTREME example to make an illustrative case in the other thread of how I view rushing efficiency being used in a regression.
Goal: Evaluate whether water is important for grass to grow
Measurables: water (what you vary), grass growth
Not measuring: whether the grass is exposed to sunlight (e.g. 50% of grass kept in dark box, 50% is kept outside exposed to sun light during the day) - obviously this is the extreme analogy for defensive formation etc.
Regression result - no correlation
Mistaken analysis - water does not affect grass growth
Correct analysis - not sure if water affects grass growth because whether there is sunlight introduces a huge error in the statistical exercise

2) I made a 'defender density' argument that compares the ability to stop the run using 8-in-a-box vs. stopping the run with 8DBs. The density of defenders in the box is much higher (can be up to 10X) than defenders covering WRs running all over the field. Same reason why it is tougher to pass in the red zone.

3) I do not remember the more complex statistical models that well from many years ago, but googling did find the "errors in variable" model that accounts for errors in the independent variables. I do recall this is a complex regression subject from a graduate econometrics class. This equates things like 8-in-a-box as errors in the model. Without trying to get through all the complex theory, if the error is substantial, then the results are unpredictable, particularly for nonlinear complex models.

 

[dogunwo]

lol humor me please...

That's Gavin Escobar

 

[waldoputty]

That's Gavin Escobar

Thanks. Fuzzy was too busy humoring me...

 

[T-RO]

And you are a proud member of the Running Back deniers club.......after 2014 and the start to this season one group looks really stupid but keep coming back for more

Not a bone inside me--and nothing that I've posted--denies Zeke, his impact and his value as a legit first round pick.

Your sliding off the road again.

 

[waldoputty]

Not a bone inside me--and nothing that I've posted--denies Zeke, his impact and his value as a legit first round pick.

Your sliding off the road again.

so if you do not deny Zeke's impact, you do not deny Zeke's running ability is important to winning then...

 

[Nightman]

Not a bone inside me--and nothing that I've posted--denies Zeke, his impact and his value as a legit first round pick.

Your sliding off the road again.

Then they are going to fire you as Cheer captain for Team Adam

 

[waldoputty]

Bad is a relative term. I dunno. Is it?

Nah seriously I think the lack of social cues like tone on the interwebs leads to a large part of the ugliness that we see.

fuzzy, are you really escobar?

 

[dmax]

That sounds good to me.

Care to weigh in on the "rushing well does not matter" argument in light of defense formations that take away the run such as 8-in-the-box?
I could supply the details if you have not seen it - this is from another thread:

I read most all of the other thread. Honestly, I think both sides have some good points. For those that say rushing YPC does not matter, I think they are correct that it is not as good a predictor of winning in hindsight. The passing efficiency metric that Adam is using is great at saying who should have won the game, much better than YPC. But it uses past statistics so cannot predict who will win next week. And I think that should be the objective. I don't care if a stat says "yep, 10 out of 12 games last week, the team with the higher passing efficiency won the game." What I care about is a measure that says "Dallas will beat Pittsburgh because they have a greater passing efficiency differential for the year."

We don't have anything like that or Vegas would be mad at us! So the goal is to develop that stat. In light of that, the running is important crowd is correct in saying that there is not a measurable stat to evaluate rushing efficiency. Using correlation to discount rushing statistics invites spurious results.

As far as your list of attributes of a good rusher, yes that sounds good. Problem is how to measure them all. Same thing with measuring if a team is good at rushing.

I have some ideas and would love to pursue a more detailed model involving causation. It will have to be done at a later date though as I am actually writing an exam over regressions for tomorrow. Perhaps several people on here would like to join me and pool our resources in the offseason to see what we can come up with?

 

[xwalker]

Don't shoot the messenger.

Some have suggested that rushing well does not matter when it comes to winning, and only passing efficiency matters.
So if you believe this, we can replace Zeke with Marcel Reece.

Who is Marcel Reece?
He is a top rated 4-time pro-bowl FB formerly with Oakland but is currently available.
He is also 250lb and runs a 4.42 40, so bigger than Zeke.
As a blocker, Reece can generate power as a runner and blocker, with low center of gravity and is natural at firing out low and hard.
He is also a top tier route runner with 4.42 40 speed, so faster than Zeke.
What is not to like...

Why? Perhaps we can save on salary cap $ and we can even trade Zeke for at least one 1st round pick or even a DE that we desperately need!
That is if you truly believe that running the ball well does not matter to winning (based on silly statistical analysis using data that is very flawed...)

What do you guys think would happen if we were to replace Zeke with Marcel?



Seriously, Marcel may not be a bad pick up... given the receiving threat as FB...

They might as well sign me since it does not matter if you have a good RB or a terrible one.

 

[FuzzyLumpkins]

fuzzy, are you really escobar?

I am the walrus.

 

[waldoputty]

I read most all of the other thread. Honestly, I think both sides have some good points. For those that say rushing YPC does not matter, I think they are correct that it is not as good a predictor of winning in hindsight. The passing efficiency metric that Adam is using is great at saying who should have won the game, much better than YPC. But it uses past statistics so cannot predict who will win next week. And I think that should be the objective. I don't care if a stat says "yep, 10 out of 12 games last week, the team with the higher passing efficiency won the game." What I care about is a measure that says "Dallas will beat Pittsburgh because they have a greater passing efficiency differential for the year."

We don't have anything like that or Vegas would be mad at us! So the goal is to develop that stat. In light of that, the running is important crowd is correct in saying that there is not a measurable stat to evaluate rushing efficiency. Using correlation to discount rushing statistics invites spurious results.

As far as your list of attributes of a good rusher, yes that sounds good. Problem is how to measure them all. Same thing with measuring if a team is good at rushing.

I have some ideas and would love to pursue a more detailed model involving causation. It will have to be done at a later date though as I am actually writing an exam over regressions for tomorrow. Perhaps several people on here would like to join me and pool our resources in the offseason to see what we can come up with?

I think everyone agrees with YPC is not a good stat.
Fundamentally, I dont know how one strips from rushing ability enough of the external factors such as defensive formation.
But I suppose you already have read this over and over again already.
But the next argument is going to be over what you meant by:
"Using correlation to discount rushing statistics invites spurious results"

 

[waldoputty]

They might as well sign me since it does not matter if you have a good RB or a terrible one.

Man, not sure 500k per year is worth the brain damage risk...

He made an absolute claim - running well does not matter .
When tested using the 2 ends of the curve, he says no fair.

When I am wrong or bad at something, I admit it.
Like my poor attempt at sarcasm