QB who can pass to victory - a myth?

[Rock-A-Bye Beasley]

14 minutes ago, Hapless Bills Fan said:

 

Um, I think you might have skipped a couple lectures. There's nothing in Statistics 101 that says a premise is flawed because a demonstrated correlation does not address causation.

 

Sure, correlation is not causation.  The question I was asking was not asserting causation.  I was simply asking: what is the outcome when [EVENT] occurs?  To put it in Statisics 101 terms, I was testing the null hypotheses that passing 35 or more times per game is not correlated or is positively correlated with winning.  Passing 35 or more times per game is negatively correlated with winning.  The null hypotheses are rejected. 

Now, one can then ask a follow-on question such as: are high numbers of passing attempts by a team correlated with the team trailing in score?  One would have to define some sort of threshold for how much of a score deficit and at what stage of the game.  And you might be correct.

 

Or you might be surprised - while 300+ yard passing games used to have a slight negative correlation with losing, that's changed of recent years to neutral or a slight positive correlation.  I actually wasn't expecting this. 

 

I think it may be stronger correlation with losing because throwing lots of passes doesn't necessarily mean one is gaining lots of yards.  Looking at 300 yd games is looking at games where a team has passed a lot of times successfully.

 

The premise is flawed BECAUSE you’re not asserting causation. Why should I care in that case?

 

Besides, even if I’m accepting everything I want a top-tier QB anyway. So that’s my answer. 

 

[Hapless Bills Fan]

28 minutes ago, LSHMEAB said:

We're really going here Hap? 

I have immense respect for you and everything you do for this board; but I take exception to the premise of this thread.

 

We're really going here and you can respect it and me or don't.  Before we go there though, tell me what you see as the premise of this thread?  Because at least a couple of other people seem to have missed it.

 

28 minutes ago, LSHMEAB said:

If the have the right QB and yes, the right OC/weapons, passing nets you far more than the Bills are currently achieving. 

 

And I wouldn't argue with that.

 

But given this, I would look very carefully at the impact of a rush game.

 

28 minutes ago, LSHMEAB said:

The problem is not running less, passing more, running more, passing less; The problem is the efficacy of these things. They are just not getting it done. I don't know about Devin. He doesn't look like some game breaker. He's not bad, but he's not the kind of electrifying player you'd like to see considering his role. He's no Alvin Kamara. It wasn't a great pick IMO nor have any of the offensive picks.

 

Where are the rising stars on offense? Where are the studs? That's the problem. It's not playcalling. It's personnel. I believe that they will address this in the offseason, but they d*** well better get it right this time. I have my doubts.

 

I think Allen should be better than THIS, even if I understand the excuses/reasoning. But they really don't have "playmakers." Beane needs to put up or shut up this offseason in regards to the talent he acquires.

 

I really didn't do this to make some big point about the current Bills offense.  We need to score more points.  We need to be better.

 

But I also didn't expect it to have the result it did.  I expected it to be more neutral, or I expected it to change as I worked my way down the passing attempts.

Given this result, if I were running analytics for a football team I would at least deep dive into the impact of a run/pass balance on game outcomes.

 

And yes, of course you have a point that in the final assessment, it's not just what you dial up, but whether or not it works that matters.

[Hapless Bills Fan]

8 minutes ago, Chemical said:

The premise is flawed BECAUSE you’re not asserting causation. Why should I care in that case?

 

If you don't care, why read the thread?

 

If you don't understand that statistics is not about asserting causation, you possibly did more than just miss some lectures.

 

Quote

Besides, even if I’m accepting everything I want a top-tier QB anyway. So that’s my answer.

 

I want a pony.  And  dammit, a ticket to Belize.

[Augie]

6 minutes ago, Chemical said:

 

The premise is flawed BECAUSE you’re not asserting causation. Why should I care in that case?

 

Besides, even if I’m accepting everything I want a top-tier QB anyway. So that’s my answer. 

 

 

I want a lake house in the mountains and a place in Florida for the winters. You get what you get, can you live with it? I’m managing so far. You OK? 

[Hapless Bills Fan]

47 minutes ago, Shaw66 said:

I just eye-balled Brady's first few years.  Looks like he was around .500 when he threw over 35.  Which means his winning percentage was better when he threw 35 or under.  

 

Damn.  First years he played regularly, 7W-4L at 35 or over passes so same ~2-1 split as Brees, Brady, Rodgers, Wilson have now.  He only had 3 such games his first year though.

[Augie]

2 minutes ago, Hapless Bills Fan said:

 

 

I want a pony.  And  dammit, a ticket to Belize.

 

 

You made me look greedy! You may be more wise than I ever expected! 

[Shaw66]

2 minutes ago, Hapless Bills Fan said:

 

Damn.  First years he played regularly, 7W-4L at 35 or over passes so same ~2-1 split as Brees, Brady, Rodgers, Wilson have now.  He only had 3 such games his first year though.

I did over 35.  Seems like he wins every game with exactly 35 attempts. 

 

As for causation, no one can anything with simple analysis.  It's complicated.  But I gave a logical explanation about why balance works best.  I proposed a theory.  And your data is consistent with the theory, so that's something. 

[Rock-A-Bye Beasley]

19 minutes ago, Hapless Bills Fan said:

 

If you don't care, why read the thread?

 

If you don't understand that statistics is not about asserting causation, you possibly did more than just miss some lectures.

 

 

I want a pony.  And  dammit, a ticket to Belize.

 

You keep saying I don’t understand which is not true. But good attempt at insulting me a second time with the same BS about missing lectures. 

 

I’m not trying to be rude, just pointing out this is pointless if it’s not proving anything. 

 

 

[SlimShady'sSpaceForce]

Based on what I’ve seen Josh has 4 comebacks and 5 game winning drives (right?)

 

the problem is the first 3 quarters 

 

Make that 

 

5 comebacks, 7 game-winning drives

https://www.pro-football-reference.com/players/A/AlleJo02.htm

 

Edited November 13, 2019 by SlimShady'sGhost

[oldmanfan]

2 minutes ago, Chemical said:

 

You keep saying I don’t understand which is not true. But good attempt at insulting me a second time with the same BS about missing lectures. 

 

I’m not trying to be rude, just pointing out this is pointless if it’s not proving anything. 

 

 

It’s not pointless.  It is an interesting data set that, if this were a scientific endeavor, would suggest further experimentation and analysis.  Does it definitively prove or disprove anything?  No.

[Hapless Bills Fan]

3 minutes ago, oldmanfan said:

It’s not pointless.  It is an interesting data set that, if this were a scientific endeavor, would suggest further experimentation and analysis.  Does it definitively prove or disprove anything?  No.

 

someone who gets it.

[Mr. WEO]

1 hour ago, Hapless Bills Fan said:

There's a lot of talk about needing to find a QB who can win the game with his arm.

 

Now some would say I'm a football Australopithecine.  I love to watch a good defensive battle.  And I think nothing helps a young QB more than the knowldge his team can get a first down with their feet any time.  I think a balanced run-pass attack is critical for most QB's success.  But I'm told I should get with the era of Modern Football where passing must predominate.

 

Being a simple Hapless Fan, it occurred to me to ask a simple question:  What is the W-L outcome, actually, when a team's QB passes for most of the plays in a game?

 

I'll put the bottom line right up front here, then give you the details: 62% of the time, You Lose. 2% of the time, you Tie. 

 

Smoke That, Sports Fans: When the QB is slinging the rock a lot, almost 2/3 of the time your team ain't gonna win.

 

Now I know what some of you are probably thinking.  If I looked at most of the QB in the league, maybe that skewed the data.  The top passers win more, the bottom passers lose more.  But that isn't what I saw.  It was pretty consistent through ALL the QB.  Exceptions were Mr GOAT Tom Brady (7W, 1L).  Other exceptions who won 2x as much as they lost were Aaron Rodgers, Russ Wilson, Drew Brees.  Limited data on some positive newbies who won more than L but didn't have many high passing attempt games were Josh Allen, Jacoby Brisset, Mason Rudolph and Desean Watson.  Mahomes and LJax even split.

 

So, Cro Magnon Football Fans: the Australopithecine here wants to raise the question, is the QB who can carry the team to victory with a blistering passing attack a myth?  There seem to be about 4 of 'em in the league right now.

 

OK, here's what I done:

 

Criterion: The league average #plays per PFR is 63.  So I defined "most of the plays" as "more than half" and chose >= 55% of the ave. # of plays as my cutoff.  That is 35 or more passes per game. 

 

Method: I looked at the top 34 QB based on passing attempts as of Week 10, then excluded the best (Tom Brady, 7-1) and the worst (Andy Dalton, 0-8) [that's a common statistical practice, to prevent outliers from skewing the data set].  I used the QB game stats in Pro Football Reference, sorted by pass attempts, and tallyed W-L-T.  Total games were 105.

 

 

 

 

Isn the point of this analysis to assert that a team doesn't need a talented passer at QB?  That high passing attempts alone, regardless of whether the QB is very good at throwing them, is a negative predictor of success?  That you don't need a top passer to win a lot games?

 

25 of 32 teams (the good and the bad) throw for more than 55% of their Offensive snaps.   Also, lots of football has been played before 2019...

 

First down with their feet "anytime"?  No such team.  Tossing out outlier Baltimore (11 per game), the next best is under 8 poor game rushing.  Baltimore has 12 per game--near the bottom.  25 teams have more than 11 passing first downs per game.

 

Bottom line: if your D is bad, you're gonna throw a lot.

 

 

 

 

[Rock-A-Bye Beasley]

2 minutes ago, oldmanfan said:

It’s not pointless.  It is an interesting data set that, if this were a scientific endeavor, would suggest further experimentation and analysis.  Does it definitively prove or disprove anything?  No.

 

“What is the W-L outcome, actually, when a team's QB passes for most of the plays in a game?”

 

that question is pointless. 

[Hapless Bills Fan]

30 minutes ago, Augie said:

You made me look greedy! You may be more wise than I ever expected! 

 

I'm not sure but what a lake house in the mountains might be less expensive than a pony.
You can always rent it out when you're not using it. 

 

9 minutes ago, Chemical said:

You keep saying I don’t understand which is not true. But good attempt at insulting me a second time with the same BS about missing lectures. 

I’m not trying to be rude, just pointing out this is pointless if it’s not proving anything.

 

Except of course, that you keep demonstrating it is true.  It's OK to not understand - lots of people today don't "get" how science works, hypothesis testing, statistics and so forth, but it's useful to know what you don't know.  If I wanted to insult you, you would most unambiguously know it.  Against my ethics here damn the luck.

 

This is how it works. 

 

1) Someone has an idea - "Huh, everyone keeps talking about how important passing is in the NFL and it's a passing-centered league.  I wonder what the actual W-L outcome is when teams pass more than they run?  How could I look at that?"

2) A bit of thought given to what would be a reasonable test - "Hmmm, league average is 63 offensive plays per game.  Let's just look at what happens when teams pass a bit more than they run, maybe 10% more.  55% pass plays.  What would that be?  25 passes.  I'm going to guess that teams win more than they lose when they pass more than 25 times. (that's called the hypothesis)

3) Collect the data and look at the result.  Huh.  Hypothesis rejected  - teams actually don't win 2/3 of the time when they pass more than 25 times

4) Think about why that might be and do further studies to confirm or reject those reasons

There is also 2a) examine the test used and consider if the methodology was valid, but that's not what you're doing.

 

I mean, you can not care and all that, it's fine, but "pointless unless proving anything" or "premise flawed if it doesn't show causation" is just a fundamental lack of understanding of how science and engineering work and how statistics support them.  Sorry for the lecture, helping people understand science is important to me in real life. 

[Hapless Bills Fan]

16 minutes ago, Mr. WEO said:

Isn the point of this analysis to assert that a team doesn't need a talented passer at QB?  That high passing attempts alone, regardless of whether the QB is very good at throwing them, is a negative predictor of success?  That you don't need a top passer to win a lot games?

 

25 of 32 teams (the good and the bad) throw for more than 55% of their Offensive snaps.   Also, lots of football has been played before 2019...

 

First down with their feet "anytime"?  No such team.  Tossing out outlier Baltimore (11 per game), the next best is under 8 poor game rushing.  Baltimore has 12 per game--near the bottom.  25 teams have more than 11 passing first downs per game.

 

Bottom line: if your D is bad, you're gonna throw a lot.

 

No, not at all.  You need a talented QB to win, I'm sure of it.  But there seems to be this idea that teams that pass a lot, win a lot; that a talented QB (with some talent around him on OL and WR) should be able to win on his arm, without much support from a run game. 

 

That doesn't seem to be true so far this year, is the point. 

 

The overall offensive snaps thing averages R/P balance over all games.  Looking at individual games and the balance there vs outcome, is a different question.

[oldmanfan]

26 minutes ago, Chemical said:

 

“What is the W-L outcome, actually, when a team's QB passes for most of the plays in a game?”

 

that question is pointless. 

No.  It’s a question.  The answer makes you go hmmmm.  Wonder what that might mean?  Maybe look a little deeper.

 

A lot of Nobel prize winners started by asking a simple question,  then going hmmmm.

[Rock-A-Bye Beasley]

11 minutes ago, Hapless Bills Fan said:

 

I'm not sure but what a lake house in the mountains might be less expensive than a pony.
You can always rent it out when you're not using it. 

 

 

Except of course, that you keep demonstrating it is true.  It's OK to not understand - lots of people today don't "get" how science works, hypothesis testing, statistics and so forth, but it's useful to know what you don't know.  If I wanted to insult you, you would most unambiguously know it.  Against my ethics here damn the luck.

 

This is how it works. 

 

1) Someone has an idea - "Huh, everyone keeps talking about how important passing is in the NFL and it's a passing-centered league.  I wonder what the actual W-L outcome is when teams pass more than they run?  How could I look at that?"

2) A bit of thought given to what would be a reasonable test - "Hmmm, league average is 63 offensive plays per game.  Let's just look at what happens when teams pass a bit more than they run, maybe 10% more.  55% pass plays.  What would that be?  25 passes.  I'm going to guess that teams win more than they lose when they pass more than 25 times. (that's called the hypothesis)

3) Collect the data and look at the result.  Huh.  Hypothesis rejected  - teams actually don't win 2/3 of the time when they pass more than 25 times

4) Think about why that might be and do further studies to confirm or reject those reasons

There is also 2a) examine the test used and consider if the methodology was valid, but that's not what you're doing.

 

I mean, you can not care and all that, it's fine, but "pointless unless proving anything" or "premise flawed if it doesn't show causation" is just a fundamental lack of understanding of how science and engineering work and how statistics support them.  Sorry for the lecture, helping people understand science is important to me in real life. 

 

Pointless drivel. The initial question is so useless that all your “scientific” analysis is a waste of time.

(I have multiple college degrees btw so don’t waste your time)

 

” I wonder what the actual W-L outcome is when teams pass more than they run? “

 

Once you answer that question then what? What does it tell you if it’s only a correlation and a loose one at that?

 

 

[Nextmanup]

Winning with the arm of your QB doesn't mean "throw all the time" so in that sense, this experiment is flawed before it begins.

 

You are also going to retrieve tons of "bad" data from teams that are getting blown out and have to go into pass mode early and stay there for large portions of the game.

 

It's kind of a foolish experiment.

 

I think we can all agree that your chances of winning are better if you have a good  throwing passer.

 

Don't take my word for it.  have a look at the final 4 teams each year for the last decade and report back on whether or not they had a good passing QB.

 

 

[oldmanfan]

5 minutes ago, Chemical said:

 

Pointless drivel. The initial question is so useless that all your “scientific” analysis is a waste of time.

(I have multiple college degrees btw so don’t waste your time)

 

” I wonder what the actual W-L outcome is when teams pass more than they run? “

 

Once you answer that question then what? What does it tell you if it’s only a correlation and a loose one at that?

 

 

 It tells you to look deeper into the question.  Again, it is not meant to provide a definitive answer.

 

I’d be curious what your degrees are in and whether you’ve conducted research.  

5 minutes ago, Nextmanup said:

Winning with the arm of your QB doesn't mean "throw all the time" so in that sense, this experiment is flawed before it begins.

 

You are also going to retrieve tons of "bad" data from teams that are getting blown out and have to go into pass mode early and stay there for large portions of the game.

 

It's kind of a foolish experiment.

 

I think we can all agree that your chances of winning are better if you have a good  throwing passer.

 

Don't take my word for it.  have a look at the final 4 teams each year for the last decade and report back on whether or not they had a good passing QB.

 

 

You need a good passer.  But you might also benefit from a good runner.

[Jerome007]

25 minutes ago, oldmanfan said:

You need a good passer.  But you might also benefit from a good runner.

A great offensive line makes both suddenly look way better. A lot of pieces matter in the puzzle. Yet all teams have string and weak points. Good coaches adjust to their own and their opponents'. Once in a while, a team is stacked at some positions, but even then, it's most often close fights, and Any Given Sunday applies.