4th down? GO FOR IT!

[Bronco Billy]

I'm not going to pretend that I can provide an analysis as thorough as the one linked to by the OP and I'm not sure why either of us running some half-baked version using flawed assumptions and numbers would either confirm or discredit their simulations.

 

I was simply pointing out the folly in saying, "What a load of crap, these numbers disprove it", when "these numbers" are based on rounding things up or down, in both cases in your favor (I already pointed out the punting yardage, but you also undersell the pts per first down bit because teams average 22 pts a game and 20 first downs, which, while not a huge difference does happen to be the 2nd of two data points that you rounded towards making your point sound better), but also failing to take into account the other effects going for it on 4th would have on both how many more yards the team going for it could gain but also how many less the opponent would by being deprived the ball.

 

Obviously I can't pretend to know what either of those 2nd factors would have on the equation, but I do know that you have to count them in somehow, because they certainly exist. But, in your quick dismissal of these guys findings, you didn't bother to.

Exactly. Bronco. Both yours and theirs are basically simulations. Yours is just based on flawed or very limited data. So I don't see where you can come off saying "you'll take a simulation over actual data?"

 

Because there is no "actual data". There is no reliable sample size of NFL teams actually going for it on 4th down far more often than they currently do and whether or not that's a winning strategy. So, that also does away with the "don't you think that if it worked, they'd be doing it?" argument, because we simply don't know. All we have are computer simulations run by geeks because we don't actually have any real-life data that says one way or the other.

 

I'm just prepared to think these guys are onto something because of other ways we have a flawed perception of what is risky and what is not. Few actually understand odds. That, they're less about the likelihood of something happening and more about the pay-off vs the likelihood. Take the Tebow 2 pt thing. Now, I agree that, even if he's 51%, that doesn't automatically mean you should always go for 2 (not saying that's not the case either, mind you, I just don't know). However, if you're down by 7 at the very end of the game and score a TD, that's a great effing time. Because you're trading less than 50/50 chance of winning the game (PATs convert at just below 99%, and you multiply that times the 50% chance you have of winning in OT) for a 51% chance of going home right then and there as the winner.

 

Of course, if a coach goes for it there, he's "a gunslinger, living by the seat of his pants". Why? Because he took the more statistically sound road? Seems like the more conservative approach, no?

 

So, this is just another example. We have a built-in notion of what makes sense and what doesn't. What's risky and what's not. But it's based entirely on what we're used to seeing. So, when someone suggests otherwise, we rush to come up with the first half-baked, poorly supported argument we can find to show how wrong they are.

 

 

I actually was hoping you'd support your assertions since I found them well founded and valid. I'm disappointed you chose not to. I would have liked to have heard your support for your position other than just a feeling.

 

And the faults in computer modeling are well known. It doesn't mean they don't have value, it simply means they have to be recognized as providing inaccurate outcomes that some people rely upon to make emphatic conclusions.

[detlef]

This explanation does not involve any attempt to quantify the risk/reward. It's based upon football theory. Pretty sure I made that clear in the initial sentence.

 

My point is, your focusing entirely on "giving" the opponent the ball in good field position and ignoring the fact that you're also "taking" it by virtue of not forfeiting possession of the ball. Which is happening, by your assumption 3x in your favor and 2x in theirs.

 

The second part was bringing it back to your first detailed post. Which, of course, if you're prepared to admit was based on flawed and incomplete data as well as sketchy assumptions and therefor baseless, I'll be happy to move on from and not question any further.

[Bronco Billy]

My point is, your focusing entirely on "giving" the opponent the ball in good field position and ignoring the fact that you're also "taking" it by virtue of not forfeiting possession of the ball. Which is happening, by your assumption 3x in your favor and 2x in theirs.

 

The data supports that assumption. Perhaps you can show otherwise?

 

The second part was bringing it back to your first detailed post. Which, of course, if you're prepared to admit was based on flawed and incomplete data as well as sketchy assumptions and therefor baseless, I'll be happy to move on from and not question any further.

 

 

Again, I thought I had done that. You must have missed that. Still hoping you'd do something to support your position. I'd like to see it and then discuss.

[detlef]

I actually was hoping you'd support your assertions since I found them well founded and valid. I'm disappointed you chose not to. I would have liked to have heard your support for your position other than just a feeling.

 

And the faults in computer modeling are well known. It doesn't mean they don't have value, it simply means they have to be recognized as providing inaccurate outcomes that some people rely upon to make emphatic conclusions.

 

Do you honestly think either of us has the time, acumen, access to data, or otherwise to truly make a compelling argument why going for it more often on 4th down makes sense?

 

Sorry, but the best I could do is what you did. Cobble together some data, either skew it in my favor like you did or not and draw some conclusions that made me feel good about my point. And I'm not exactly sure why that's any better than saying, "If I had to guess".

[Bronco Billy]

Do you honestly think either of us has the time, acumen, access to data, or otherwise to truly make a compelling argument why going for it more often on 4th down makes sense?

 

Sorry, but the best I could do is what you did. Cobble together some data, either skew it in my favor like you did or not and draw some conclusions that made me feel good about my point. And I'm not exactly sure why that's any better than saying, "If I had to guess".

 

 

Too bad.

[detlef]

Too bad.

 

Dude, it's pretty simple. I heard two arguments on the topic. One presented by the geeks in that article, and one contrary to that given by you.

 

No sooner did I finish reading your post, but I noticed several flaws in your data as well as factors that you completely missed. I made the bold assumption that a bunch of computer geeks who spent some time on this likely had done a more complete job than you did and figured they had a better chance of being on to something than you did.

 

Your follow-up posts have been mercifully lacking more bad data but that has simply been replaced with absolutes like "is well known" or "is just plain bad strategy", so you're not really making arguments. You're just stating your opinion in absolute terms. And, lacking my own super computer and team of nerds, I don't know if I could bring any more than that to the table.

[stevegrab]

For a moment, I thought Mularkey had read this post before coaching tonight . . .

 

 

I loved how Mayock kept blabbering on about stuff endlessly as the Jags offense was going back out on the field for fourth down. Then finally recognizes it "but look who's on the field." Well duh dumbass, I'm watching the game so I noticed that. WTF are you doing staring at pictures on a wall?

[stevegrab]

In college when you are coaching a team like Oregon that can seemingly score at will, it's a whole lot easier to take chances. You miss on a couple of 4ths downs? So what? You beat Arizona 45-21 instead of 60-10. There isn't that wide a differential between teams in the pros. And if you go for it on 4th down and miss regularly - and by regularly, we're talking 40% of the time - you're going to take some losses you shouldn't. Those kinds of losses are well remembered and will affect your evaluation and coaching future.

 

 

That was exactly how I took it when I read it, and tried to make that point. In the NFL teams are not so vastly superior that they can take that risk.

[detlef]

That was exactly how I took it when I read it, and tried to make that point. In the NFL teams are not so vastly superior that they can take that risk.

 

But even that breakdown on the Oregon game given by BB was misstated, and not unlikely intentionally. The premise is not based on "so what if we miss a few 4th downs, we'll just win by 20 instead of 40", because that assumes it's a bad strategy but one you can afford because you're so much better.

 

No, the premise is based on "so what if we miss a few 4th downs, all in all we'll convert enough that it will be a net positive." Which means, you're expecting to win by more because of this, not just doing it for the hell of it because you're good enough to get away with it.

 

Now, that may ore may not be true, but if you phrase it the way BB did, you're basically assuming it's a bad idea and then going from there.

[stevegrab]

Of course, if a coach goes for it there, he's "a gunslinger, living by the seat of his pants". Why? Because he took the more statistically sound road? Seems like the more conservative approach, no?

 

 

You're leaving out one very important factor, going for 2 after scoring a TD late in the game with Tebow, 49% of the time you'll miss and lose the game. No chance at OT where the team winning the toss still has some advantage and anything is possible.

 

 

Also, I'd suggest that unless we know what goes into the computer simulations of the geeks (as stated before I didn't read it all) its hard to know what is or isn't flawed about their assumptions. As somebody that works with computers all the time, simulations and programs are only as valid as they data they are given.

 

I'm not doubting that this works for Oregon, I do doubt that it would work in the NFL, or more precisely that if it were applied to the NFL, you'd see these teams not going for it on 4th down that much more than they currently do.

[detlef]

You're leaving out one very important factor, going for 2 after scoring a TD late in the game with Tebow, 49% of the time you'll miss and lose the game. No chance at OT where the team winning the toss still has some advantage and anything is possible.

 

 

Also, I'd suggest that unless we know what goes into the computer simulations of the geeks (as stated before I didn't read it all) its hard to know what is or isn't flawed about their assumptions. As somebody that works with computers all the time, simulations and programs are only as valid as they data they are given.

 

I'm not doubting that this works for Oregon, I do doubt that it would work in the NFL, or more precisely that if it were applied to the NFL, you'd see these teams not going for it on 4th down that much more than they currently do.

 

I'm doing nothing of the sort. I fully realize that I have a 49% chance of losing the game right there and then. However, I also realize that I have a less than 50% chance of winning if I kick the PAT. After all, each team has spent the last 60 minutes proving that it's anyone's game, so it would be folly to assume, barring any extenuating circumstances (like they built a huge lead, then lost their starting QB and haven't done dick since then while we scored 20 in the 2nd half to tie or something like that), that it's anything but a 50/50 chance in OT. However, I still have to convert something that, while hughly likely, is not certain, because PATs work 99% of the time. So, the net chance if I go that was is 49.5% (.99x.50)

 

In other words, I have a choice between a path that provides me a 51% chance or one that gives me a 49.5% chance. Why would I choose the lesser of the two?

 

As far as the computer simulation vs BB's. Yes, for all I know, theirs is for chight. But I know his is. So, I'll go with the one that has a chance of being accurate over the one that I can see with my own two eyes is flawed in many obvious ways.

Edited November 9, 2012 by detlef

[Bronco Billy]

I'm doing nothing of the sort. I fully realize that I have a 49% chance of losing the game right there and then. However, I also realize that I have a less than 50% chance of winning if I kick the PAT. After all, each team has spent the last 60 minutes proving that it's anyone's game, so it would be folly to assume, barring any extenuating circumstances (like they built a huge lead, then lost their starting QB and haven't done dick since then while we scored 20 in the 2nd half to tie or something like that), that it's anything but a 50/50 chance in OT. However, I still have to convert something that, while hughly likely, is not certain, because PATs work 99% of the time. So, the net chance if I go that was is 49.5% (.99x.50)

 

In other words, I have a choice between a path that provides me a 51% chance or one that gives me a 49.5% chance. Why would I choose the lesser of the two?

 

As far as the computer simulation vs BB's. Yes, for all I know, theirs is for chight. But I know his is. So, I'll go with the one that has a chance of being accurate over the one that I can see with my own two eyes is flawed in many obvious ways.

 

 

Well then put forth something rational yourself rather than repeated blathering of the same tortured points. You made some good points, but you refuse to run with them at all but rather just gleefully continue on in attack mode. My old man used to have a saying applicable to you. "Either #### or get off the pot."

 

If an NFL coach thought they could get a competitive advantage by doing something like this, someone would have tried it long ago, and then if it worked everyone in the league would copy it. It's not like the NFL is 5 years old and guys are just figuring out how to coach this newfangled game of football. There's been some pretty damned smart people coaching through the life of the league, and not one of them has tried this as part of their game scheming. That's says a lot more to me than my number crunching - faulty as you think it may be - or any computer simulation of games.

Edited November 9, 2012 by Bronco Billy

[stevegrab]

I'm doing nothing of the sort. I fully realize that I have a 49% chance of losing the game right there and then. However, I also realize that I have a less than 50% chance of winning if I kick the PAT. After all, each team has spent the last 60 minutes proving that it's anyone's game, so it would be folly to assume, barring any extenuating circumstances (like they built a huge lead, then lost their starting QB and haven't done dick since then while we scored 20 in the 2nd half to tie or something like that), that it's anything but a 50/50 chance in OT. However, I still have to convert something that, while hughly likely, is not certain, because PATs work 99% of the time. So, the net chance if I go that was is 49.5% (.99x.50)

 

In other words, I have a choice between a path that provides me a 51% chance or one that gives me a 49.5% chance. Why would I choose the lesser of the two?

 

As far as the computer simulation vs BB's. Yes, for all I know, theirs is for chight. But I know his is. So, I'll go with the one that has a chance of being accurate over the one that I can see with my own two eyes is flawed in many obvious ways.

 

 

Ok, I see your point on the less than 50% chance of winning if you attempt the XP. But I don't think coaches or most people look at it that way. PAT is nearly automatic, and the game continues and you have a 50-50 chance of winning at that point. Even if a team believed they had a 75% of converting the 2P conversion, I don't think they'd consider that giving them the better chance to win, even with the revised OT rules (where the winner of the coin toss isn't as likely to win).

 

Sometimes statistics and probability just don't translate to making real world decisions. You're taking that 51% action on one single play, where kicking PAT and going into OT has a million variables that could affect the outcome. Basically go for 2 and fail, you lost because of that. Kick PAT and still lose in OT, there were many factors involved. And that is probably why coaches will not do it, even with that slightly higher probability success.

[detlef]

Well then put forth something rational yourself rather than repeated blathering of the same tortured points. You made some good points, but you refuse to run with them at all but rather just gleefully continue on in attack mode. My old man used to have a saying applicable to you. "Either #### or get off the pot."

 

If an NFL coach thought they could get a competitive advantage by doing something like this, someone would have tried it long ago, and then if it worked everyone in the league would copy it. It's not like the NFL is 5 years old and guys are just figuring out how to coach this newfangled game of football. There's been some pretty damned smart people coaching through the life of the league, and not one of them has tried this as part of their game scheming. That's says a lot more to me than my number crunching - faulty as you think it may be - or any computer simulation of games.

 

Fine, here goes:

1) My primary point, all along, is that the article made enough sense to me that it's going to take more than the very faulty argument you put forth to convince me it's wrong. That, they may be right, or they may be wrong, but it would be virtually impossible for their methodology to be any worse thought out or intentionally skewed as yours was. And I believe I've pointed out those mis-steps, errors, and intellectually disingenuous arguments thoroughly. At the same time, I'm not going to pretend that I can bring any data or credible analysis beyond what these guys have done to convince anyone who wasn't convinced by the article. That doesn't mean I can't point our where arguments that dispute the article are lacking.

 

2) That much if this may be based on our own flawed definition of what is risky and what is not. That, rather than look at actual probabilities, we go by what we think should be true, whether or not it actually is. Our dear Steve has given us a fine example of this below:

Ok, I see your point on the less than 50% chance of winning if you attempt the XP. But I don't think coaches or most people look at it that way. PAT is nearly automatic, and the game continues and you have a 50-50 chance of winning at that point. Even if a team believed they had a 75% of converting the 2P conversion, I don't think they'd consider that giving them the better chance to win, even with the revised OT rules (where the winner of the coin toss isn't as likely to win).

 

How anyone could argue that you should or even would forgo a 75% chance of winning a game for a 50% (actually slightly less) illustrates this completely.

 

3) You're basically saying that every good idea in football has already been thought of and tried. Up until 1980, it was basically about pound the rock, pound the rock, throw it down field. Then Bill Walsh said, "What if we just throw it more often but 5 yards at a time?" Seemed to work. Up until about 10 years ago, you went no-huddle basically only when your back was against the wall and you didn't have time to huddle up. Then guys started noticing that they moved the ball so effectively in those situations, so they just started doing it more often. Why is this so different? Again, are you implying every good idea has already been thought of and tried?

[stethant]

And the faults in computer modeling are well known. It doesn't mean they don't have value, it simply means they have to be recognized as providing inaccurate outcomes that some people rely upon to make emphatic conclusions.

 

And this is different from your posts how exactly? One is a computer model with its inherent problems, considered by academics and AccuScore. The other is you presenting an argument off the cuff in a FFL website forum.

 

Now, if you're a modeler and this is your area then my bad. Otherwise, you may wish to consider the insights of people who have thought about this alot more than you or I. I think it would be very refreshing if a team actually tried this. When a team punts, the blame goes to players. When a team goes for it on 4th down, the blame goes to the coach. No surprise why we don't see more of this, despite its advantages.

[nelsosi]

There was an article written for an actuarial magazine on this topic a few years ago. Well worth a read if you're interested in the stats behind it:

http://www.contingencies.org/septoct08/fourth.pdf

[Bronco Billy]

Otherwise, you may wish to consider the insights of people who have thought about this alot more than you or I.

 

I believe that I have stated several times that I have done exactly that. I consider the insights of decades of NFL coaches more valuable than you, me, or the number crunching computer modeling geeks - whom I am pretty certain have little experience as NFL HCs. But I could be wrong.

[tazinib1]

Bucs just went for it...in the red zone ready to score. Nuff said.

[MikesVikes]

I think Jeff Fisher saw this posting today.

[tazinib1]

yep

[detlef]

I believe that I have stated several times that I have done exactly that. I consider the insights of decades of NFL coaches more valuable than you, me, or the number crunching computer modeling geeks - whom I am pretty certain have little experience as NFL HCs. But I could be wrong.

 

Absent any meaningful amount of real-life data (because no team has actually tried this at the level being suggested or even close to that), how could the coaches "consider" this without employing computers, number crunchers, and the like? Throw some grass in the air and see which way it blows?

 

So, you're going back to your apparent assertion that every good idea in the NFL has already been thought about and done. Which, history shows is silly. In fact, the NFL has a very reliable track record of everyone doing the same thing until someone challenges the paradigm, unveils something new that works, which in turn becomes more and more common. See: West coast offense and no-huddle offense at times other than the end of the half. Why is this the one new idea that we're going to just accept, "if it made sense, they'd be doing it" argument?

[Big John]

And PIT fails on it