Sneak Peek of 2nd Down Play-Calling Study

[Arch Stanton]

Those are all fair points. My point is just that the only information we do have is based on the result of what happens, not what was intended to happen perhaps. We dont actually know playcalls or audibles that are associated with each formation, in each situation given game situation and opponent.

This has been the endless defensive argument, guys blaming corners/linebackers and safeties for making what appears to be a mistake only to find out someone else was at fault.

Just like last year when people were blaming romo miscommunications with owens....everyone claimed one thing or another and had a negative opinion of romo.....yet when I watched laufenberg break down game film of romo on those plays the fault didnt lie with him.....

We simply dont know what we dont know. Now we can debate it to death, but in the end we dont have the correct info to make assumptions like people are unacceptable.

but to each his own...I will go away now.

Agreed. There are so many coaches that post here I wonder why they're not play calling in the NFL.

 

[dallasfaniac]

Guys,
Whether or not the other coordinators show this trend is irrelevant. The point is that Garrett does, and it is a mistake. If all 31 other coordinators do the same thing, it does not make it correct for Garrett. It is not his "fault" that they might make the mistake as one of you said, but it is his fault that he displays the trend.

This right here is all I need to know about this "article". Pseudoscience FTW.

 

[Arch Stanton]

Guys,
Whether or not the other coordinators show this trend is irrelevant. The point is that Garrett does, and it is a mistake. If all 31 other coordinators do the same thing, it does not make it correct for Garrett. It is not his "fault" that they might make the mistake as one of you said, but it is his fault that he displays the trend.

And to those claiming the stats are skewed because of audibles: the Cowboys are not a team that checks a lot. When they do, it is almost always a "kill" call in which they called two plays in the huddle. Thus, there would be no incentive to make a dummy kill call. The defense doesn't know either play called, so no "kill" calls are faked because doing so would create no competitive advantage.

Are you Jack Reilly in disguise?

 

[Hoofbite]

That's the single most ridiculous thing that will be said in this thread.

Any one who take some stats would know that it will actually be worthless if we don't have the stats from the rest of the league to compare this to

Please.

It would take far too much ****ing time and everyone knows it. It would go unheard if Garrett was a deviation and the only way people would even acknowledge the work is if it was complementary of the man.

It would be worthless. Mainly because of time but because of other reasons as well. Oh, did I mention that people don't care about other teams? That's a big reason as well.

By all means, I beg you to take the time and figure out these numbers for the other 31 teams. Please, I'd love to know if Garrett is the rule or the exception.

 

[Jongb35]

This right here is all I need to know about this "article". Pseudoscience FTW.

Which part of my study is pseudoscience? How would the other coordinators' trends affect the validity of Garrett's own play-calling trends? I am unclear as to where you think I may have gone wrong. Please enlighten.

 

[dallasfaniac]

Which part of my study is pseudoscience? How would the other coordinators' trends affect the validity of Garrett's own play-calling trends? I am unclear as to where you think I may have gone wrong. Please enlighten.

Let's say you are 6'-2". I am 6'-4". In my testing, you are short.

 

[Jongb35]

Let's say you are 6'-2". I am 6'-4". In my testing, you are short.

That metaphor isn't representative of my stats though. The definition of 'short' is of course relative, while the percentage of runs on second down is not.

If my claim was that Garrett is a 'bad' play-caller, your analogy would fit, but I am only claiming that his percentage of runs on 2nd down contains a large discrepancy that is dependent on the 1st down play-call.

 

[Chocolate Lab]

Let's say you are 6'-2". I am 6'-4". In my testing, you are short.

Or let's say that Sean Payton had substantially the same pattern.

I would be more interested in a breakdown of why we failed to score TDs in the red zone... Or if there was something different about the Green Bay and Minnesota games when we barely scored at all. Seems to me those were the real problems with our offense.

 

[Jongb35]

Or let's say that Sean Payton had substantially the same pattern.

I would be more interested in a breakdown of why we failed to score TDs in the red zone... Or if there was something different about the Green Bay and Minnesota games when we barely scored at all. Seems to me those were the real problems with our offense.

Perhaps I should have consulted you about which stats you would be most interested in uncovering?

 

[21Savage]

Perhaps I should have consulted you about which stats you would be most interested in uncovering?

You asked for constructive and we're giving it to you.

Good job and the work done? But like others have said, we can draw any true meaning or conclusions from them without know how he compares to the rest of his peers.

We ranked 2nd in the league in yards and broke our 07 record. So it's obvious that moving the ball (whether via predictably running or passing on 2nd down) wasn't the issue.

And the height comparison metaphor was perfect because you're saying that whether or not the rest of the league is doing the same thing doesn't absolve Garrett. Well, it does. Because your assumption through out the thread is that predictability is a bad thing for him. Well, what if the more predictable offenses are the better ones?

And just so you know. Doing the stats for the whole league will be too painstaking a process to undertake without any concrete gains on your end. So I don't expect you to do so. We're just taking your stats for what they are right now..............

 

[Jongb35]

You asked for constructive and we're giving it to you.

Good job and the work done? But like others have said, we can draw any true meaning or conclusions from them without know how he compares to the rest of his peers.

We ranked 2nd in the league in yards and broke our 07 record. So it's obvious that moving the ball (whether via predictably running or passing on 2nd down) wasn't the issue.

And the height comparison metaphor was perfect because you're saying that whether or not the rest of the league is doing the same thing doesn't absolve Garrett. Well, it does. Because your assumption through out the thread is that predictability is a bad thing for him. Well, what if the more predictable offenses are the better ones?

And just so you know. Doing the stats for the whole league will be too painstaking a process to undertake without any concrete gains on your end. So I don't expect you to do so. We're just taking your stats for what they are right now..............

How could a predictable offense possibly be better than an unpredictable one?

And I still do not agree that the trends of the other coordinators has any bearing on Garrett's play-calls because it is not a zero-sum scenario. When an offense does well, the opposing defense suffers. Opposing offenses can both do well, though, or both do poorly. The success or failures of one offense does not affect another. Coordinators can be collectively good or bad because the success of one does not affect another.

 

[dallasfaniac]

I am not going to keep discussing this with you, because it's obvious you won't understand.

Garrett could have created a computer to randomly select plays out of a playbook and it is statistically possible that it calls the exact plays he did this year. Heck, it's possible that if it had 1 run and 1 pass play to select, that it selected the run 100% of the plays following a pass during the season. Looking at the statistics you might think that the computer was predictable, yet the flaw was in the observation.

You suggest that his predicatability is "unacceptable" and even though the most successfully offenses might also be predicatable, they are all failing. I would suggest that if the best offenses also are predictable, that perhaps they are only predictable when looking at the stats after the fact.

 

[21Savage]

How could a predictable offense possibly be better than an unpredictable one?

And I still do not agree that the trends of the other coordinators has any bearing on Garrett's play-calls because it is not a zero-sum scenario. When an offense does well, the opposing defense suffers. Opposing offenses can both do well, though, or both do poorly. The success or failures of one offense does not affect another. Coordinators can be collectively good or bad because the success of one does not affect another.

We ranked 2nd in the league in yardage and set a franchise record in yards gained. That's how! Why wouldn't teams have stopped us short of that if we were that predictable?

As for your second paragraph question, riddle me this: How does it affect your stats if the top 5 offenses in the league were just as predictable as Garrett's and the bottom 5 were the unpredictable ones?

 

[Jongb35]

I am not going to keep discussing this with you, because it's obvious you won't understand.

Garrett could have created a computer to randomly select plays out of a playbook and it is statistically possible that it calls the exact plays he did this year. Heck, it's possible that if it had 1 run and 1 pass play to select, that it selected the run 100% of the plays following a pass during the season. Looking at the statistics you might think that the computer was predictable, yet the flaw was in the observation.

You suggest that his predicatability is "unacceptable" and even though the most successfully offenses might also be predicatable, they are all failing. I would suggest that if the best offenses also are predictable, that perhaps they are only predictable when looking at the stats after the fact.

While it is statistically possible, it is not statistically probable. Not even close. If a computer were to randomly select plays on second down, even based on the down and distance, they would be nowhere near the percentages we see with Garrett.

You can I don't get it, but unfortunately I think I am the only one who does. The possibility of something is not a reason to believe it. Yes, it is possible a computer could select 100 percent runs, but it is also possible that the universe was created by a cyclops. You don't believe the latter just because it is possible, so why believe the former?

I think the best question to ask is if we were d-coordinators, would this info help us defend the Cowboys. And I think it would. If the Cowboys passed on first down for five yards and I knew they are three times as likely to run as pass, simply because of the previous play call, that information would be helpful, regardless of what other play-callers are doing.

Tell me if I am wrong, but how would that information not be helpful?

 

[dallasfaniac]

I lied, I can't leave work yet so I have some more but reserve the right to run out of this conversation at any point.

My main reason for comparison is this: If Garrett called running plays 80% of the time after a pass but everyone else in the NFL called running plays 100% of the time, by comparison Garrett is very unpredictable.

Sean Payton could be considered unpredictable on kickoffs due to the few onside kick calls, yet he probably only calls an onside less than 5% of the time.

So yes, league-wide comparison can be very valid when shedding light on Garrett's predictability. The fact that you don't care about the comparison and think his predictability is "unacceptable" suggests that the data analysis might not be the most accurate and unbiased. Most times, you can make data say what you want it to say...

 

[Jongb35]

I think I know where the confusion lies. Looking at the stats of other coordinators on their 2nd down calls WOULD be important if what I was concerned with was just 2nd down run/pass percentage. That number is unimportant to me though.

What is important is the difference in run/pass percentage after a run or a pass on 1st down. The further the difference, the more predictable the play calls. If Garrett runs on 60% of 2nd down plays between 3-7 yards, for example, we would want this number to be around 60% both after runs and after passes the previous play.

If the runs on 90% of those plays, we would want that number to be around 90% after both runs and passes on 1st down. So, the percentage means nothing, but the difference between the percentages based on the previous play. The latter is a stat that is independent of other coordinators, while the former (percentage of runs on 2nd down in general) would be relevant to other play-callers.

 

[Jongb35]

I lied, I can't leave work yet so I have some more but reserve the right to run out of this conversation at any point.

My main reason for comparison is this: If Garrett called running plays 80% of the time after a pass but everyone else in the NFL called running plays 100% of the time, by comparison Garrett is very unpredictable.

Sean Payton could be considered unpredictable on kickoffs due to the few onside kick calls, yet he probably only calls an onside less than 5% of the time.

So yes, league-wide comparison can be very valid when shedding light on Garrett's predictability. The fact that you don't care about the comparison and think his predictability is "unacceptable" suggests that the data analysis might not be the most accurate and unbiased. Most times, you can make data say what you want it to say...

I see what you are saying here, and I agree with that. The label of Garrett as predictable may be dependent on other play-callers, but the fact that he COULD even out the discrepancy doesn't depend on the calls of others. In that manner, he COULD become less 'predictable,' or at least more random.

Good point though. Nice convo.

 

[dallasfaniac]

If a computer were to randomly select plays on second down, even based on the down and distance, they would be nowhere near the percentages we see with Garrett.

How many times are we talking here? 2? 3? I just flipped a coin 4 times and it hit heads once. That's pretty predictable coin I have here.

What do Garrett's stats look like since 2008? How do they look from game to game? How do they look when facing different defensive alignments? How do they look when having the lead in the 4th quarter? How do they look when behind?

Using variables of my own choosing, I can make it look very unpredictable.

 

[Jongb35]

How many times are we talking here? 2? 3? I just flipped a coin 4 times and it hit heads once. That's pretty predictable coin I have here.

What do Garrett's stats look like since 2008? How do they look from game to game? How do they look when facing different defensive alignments? How do they look when having the lead in the 4th quarter? How do they look when behind?

Using variables of my own choosing, I can make it look very unpredictable.

I agree, but I think my variable are vague enough that the conclusions are valid. Further, no matter what you may think, I don't hate Garrett as a coordinator. Thus, I don't get the numbers and then form them into a way that suits my bias, because I attempt to have no bias (I know ZERO bias is impossible, but I do not go into a study TRYING to get the numbers to come out one way or another, nor do I try to present them in a way that I think is deceiving).

 

[dallasfaniac]

I am not saying Garrett isn't predictable, I am merely saying that the predictability might be overstated.

I'd imagine most team runs on second and long regardless of the preceding play. The best QB last year had a 70% completion percentage. I don't know the stats, but I would imagine that a larger percentage of rushing plays gain positive yards. Knowing that, how many teams would run more on 2nd downs after minimal gain on 1st just to make sure 3rd was a managable down?