Further Thoughts on Fourth-and-2

Pretty sure it's backwards. Had the same thought.

No, I think this is right.

Going for it means there's a 66% chance to win.
Punting means there's a 34% (or more/less) chance to *lose*.

So if the Colts' odds of scoring after a punt are 30%, then that's a 70% chance for the Patriots to win and punting is the right decision. If the Colts' odds of scoring are 40%, then that's 60% for a Patriot victory and then punting would be the wrong decision.

Every "stat nerd" I've read over the last two days has had nearly the *exact* same opinion as Barnwell on this matter; it was a very close call, but Belichick probably made the correct decision.

Easily could plug simple model into Excel and run the Crystal Ball (or @risk) simulators to see results.

Would be interested to see that.

You are absolutely right. If he was going to go for it on fourth down, he should have known this before the third down play. He could have directed his offensive coordinator to call a running play that would have made it a shorter fourth down.

Another mistake by Belicheck I'd add in was the seemingly non aggressive play calling on the drive that led to the Pats final field goal. I was saying at the time that there was too much time to be running clock and that the Pats should aggresively try and make it a 17 point game.

That's what I've been thinking too. I have no problem with the decision to go for it, but why not do it from the punt formation when you've got the element of surprise in your favor?

Now you're really mucking things up.

I don't know what the data on fake punts is, and it probably suffers from small sample size, but I'd assume the trickeration would not carry a lot of weight--especially after a timeout. (Now if they had Randall Cunningham at QB, a shotgun snap and QB-punt might have landed Indy on their own 5.)

I sided with BB in his choice--as a Colts fan I was a lot less worried about a punt than a potential conversion. But taking the ball out of the hands of your highly regarded QB and trusting it to a punter who played QB in HS, or the equivalent, or relying on a direct snap to a protecton back that your team practiced all of six times last week, does not sound like a great idea to me.

Of course a 2-yard pass pattern when you need exactly 2 yards doesn't leave a lot of room for error.

Oh hear hear - that would've been pure genius! Try to pull a guy offside, playclock at one, drop back and kick it away... Didn't Cassel do it last year? I think it was a snowy game...

"In poker when you look at the cards in your hand and the cards on the table you can calculate the odds of winning, but this isn't poker, and we can't apply the same logic."

actually your odds in (hold 'em) poker are approximations as well, since there at least 12 (five other players, two burned) cards and as many as 20 cards that are out of the deck, that you have to assume are in play when you make your calculations.

"Since we can't judge whether the process was good or not we either have to judge the result or not judge at all."

Isn't this the same as all the talking heads saying, "it didn't work, so it was wrong?"

Whether or not the odds can be calculated shouldn't factor in at all. What they are getting at is whether or not the decision made sense given what was known at the time. In other words, as long as going for it can't be shown to be the wrong decision, you can't claim it's wrong simply based on result.

Look at this at the extremes. Say a team A is leading team B, and team A has the ball with five seconds left in the game. Pretend team A's QB takes the snap and starts to kneel and for some reason throws the ball over his head, at which point team B scores and wins.

In that case, just because the outcome was bad, doesn't mean kneeling was the wrong decision. Given the facts at the time (lead with 5 seconds left) the kneel play is justifiable. The same thing applies to Belichick. Given what he knew at the time for his team and the opponent, the decision made sense. You can't clearly say that the odds were in favor the other way, so regardless of outcome it was a good decision. (Or at least an OK one.)

Not only is it amazing that someone believes that Barnwell's statement doesn't apply to "football," but it is absolutely astonishing that you used half a page to defend this belief. Ummm, no.

It's better to be lucky than good. I will grant you that.

We can never know the "real" odds of anything (other than on theoretical questions). I will also grant you that.

We can use results to try to determine what the odds might have been in any given situation. That's what this site does. It compiles statistics, analyzes them, and often tries to make predictions about what such analysis says about the risk profile of any past, present or future decision.

That said...

If someone is fairly flipping a fair coin and says to me, "I will give you $10,000 if you pick heads or tails correctly, and you will pay me $1,000 if you choose incorrectly," I will take that bet any time.

If it comes up tails and I said heads, I may *hate* the result, and I may really, really wish the result was different, but I absolutely and totally made the right decision before the result was known.

As for Belichick, he has won consistently like no other coach/team builder in recent history. It's getting to the point that we cannot reasonably attribute this to luck. He is "smarter" than his competitors. The yin/yang of this is that his competitors are "stupider" than he is. How is this possible? Well, I would posit that all his competitors are subject to cognitive biases in evaluating players and in-game decisions, most notably the CYA bias of being deathly afraid of looking stupid to the public. Of course, their desperate desire to avoid looking stupid *condemns* them to looking stupid when someone like Belichick (who is clearly not burdened by the same fear) absolutely schools them.

At what point is everyone going to realize that this is why Belichick wins? If he did the conventional thing every time, drafted according to Kiper's big board, designed every fourth down play the way Norv Turner would, *he would not have been so consistently successful.* It is literally impossible (save extreme luck) to so consistently and overwhelmingly defeat your opponents unless *you are doing something different than your opponents, all of them.* He wins because *you* are stupid.

For my own two cents, this was a close call. It may not have been "right," but I have always thought more coaches should go for it on fourth down. Had the game been tied or the Patriots up by three or less, then it would be different. Clearly, you must punt there. But if I'm the coach in this situation, I want to win it on my own terms. I have one of the best and most experienced offenses in the NFL. This is not a playoff game. I've proven I can outplay the Colts on their home field. I don't even need to win this game. But I will put it on my offense to make these two yards. I will take the risk that *even if it fails,* my defense can still prevent a TD from 30 yards away. I am trying to keep this in my own hands, not in Manning's. I don't know if I would have had the stones to make this decision, but I would *know* I made the right decision the next day after listening the hordes of the ignoramus hoi polloi condemning it. I would *know* that their idiocy is why I win. I would never be able to win without it.

This comment is patently false. If you look at the tape you see he lands at the 28. If they had not given him forward progress they would have spotted it at the 28, not the 29.5. The announcer was incorrect.

By that logic, if he had the ball in his hands but was bobbling it while standing on the 30, then dove backwards (still bobbling it) and landed on his back on the 25 with the ball firmly in his grasp, are you saying the placement should be the 30? I think the NFL rule book would disagree.

Fwd progress starts with possession of the ball. With RBs it's a little more obvious that they have it rammed into their gut--they have the ball, so it's simpler to eyeball where they cease moving forward.

Possession is the first element of a catch, and then position. Think of all the out of bounds "catches" you've seen where the guy does a great job grabbing the ball but it's not an official catch because he's OB. Same thing here--when he came down to the field of play with full possession, it was the 29.5 not the 30.

If the pass had not been rushed it might have been more accurate and catchable--his feet were beyond the 30 before he went up for it. Had he had firm control rather than bobbling it, you'd be correct. But the bobble was the deal killer here.

I don't understand why, if we accept Barnwell's numbers (which by the way I think are good) we would think this is a close decision. Don't you think Manning has a better than 1-in-3 chance of scoring of TD with 70 yards to go, 2 minutes left, and a TO? (Especially given the fact that we just allocated an 85% chance to Manning scoring a TD with 30 yards to go in the same scenario).

I keep thinking of the last two meaningful times the defense had to stop one drive in the fourth quarter, 2006 and 2007, and I was really glad to see brady with the ball

Economist understand statistics = fail

Sad sad day for our economy, if you are in a position of decision making power or teaching the youth of our nation.

The future implications is absolutely modeled into the formula.

65% complete fourth down (estimated NFL average for 4 n 2)
85% chance indy scores TD from the 29 (Indy great offense)
35% chance indy score TD from own 25 (Indy again great offense)

expected outcome:

70% chance of winning if you go for it
64% chance of winning if you punt

see awesome calculator:

http://belichick-decision.heroku.com/

Don't hate the model, hate the game

But by punting, he puts all his eggs in the "defend for ~65 yards" basket. All of the things you mention that could go wrong on a 65-yard drive could still go wrong on a 30-yard drive, but if the offense can advance the ball six feet, they don't have to go wrong.
The fact that the numbers were close reinforces your opinion that punting was better, but I'll bet if you originally thought going for it was the better option, you'd see them as backing that up. What they say to me is that neither decision is definitively "better," and while I'm happy to sit here on Tuesday morning and say yeah, they should have gone for it, if I happened to be calling plays for the Patriots that night, I'd probably have sent out the punt team.

Exactly what he says here. That 60% estimate sounds WAY!, WAY! high.

my thoughts exactly (maximizing winning).

In fact, I'd say the mistake in the second half was Welker calling the timeout on first and 10. I think timeouts are largely undervalued (not nearly as much as in college basketball, though), and I would guess the Patriots in 1st and 15 near midfield aren't in that bad of a position - of course that assumes that instead of Welker calling the time out, he'd have done nothing and the Pats would have taken a delay-of-game penalty. I guess the only catch is that time outs are valuable primarily when you are losing; coaching aggressively might imply that you should be less concerned with using time outs strategically late in the game and more concerned with using them as the need arises.

I can understand the time out before first and fourth down, if only because at that point the goal was to convert and close out the game, and a 1st and 15 in that situation is probably harder to accept than a 1st and 10. (Certainly a 4th and 7 is much worse than a 4th and 2; the last time out had to be used.)

maximizing your chance of winning vs minimizing your chance of losing is pretty meaningless. It's a zero-sum game. If you minimize your chances of losing, your chances of winning goes is affected inversely (yes I know there are ties). Either variable you choose to look at (% to win or % to lose), it's really the same thing.

now, I'm thinking you probably meant something like Belichick was trying to play the aggressor which is inherently advantageous. I think there was no way to do that in this case. Even though the Pats had the ball, I think at that point in the game, the Colts defense was clearly the aggressive unit. The Pats were just trying to hold on regardless of the personnel on the field.

I'm not buying this explanation. The potential for botched execution and the choice of play are all inherent in the decision to go for it. As someone else mentioned, Belichick didn't decide to call an "average 4th-and-2 play." He decided to call *that* play, in *that* situation, after giving Indy a cue that something was up by mucking around with the timeout, and with the knowledge that he had just removed his own ability to challenge any near misses even though he was calling routes that gave no room for error. When people say it was a bad decision they aren't talking about going for it in some random 4th-and-2 situation that exists in a statistical vacuum, they're specifically referring to 4th-and-2 from the -28 against the Colts in a game with large playoff implications where Peyton Manning has been rather inconsistent.

Maybe I'm just biased against Hoodie (who I will nonetheless say is obviously an exceptional coach) but to argue that you have to split the playcall out from the decision to go for it is ludicrous. Deciding to go for it without doing anything to set up for 4th-and-2 was a bad decision and using that play made it worse. People are tying themselves in knots to avoid that simple acknowledgement.

Yes well part of the reason that Belichick is a better coach and has won more games than Cowher is because he managed to amass more games with 11+ point leads than Cowher did.

Interesting perspective--in my mind, until yesterday, I thought BB always had a 4-down strategy in mind. With a 12 our of 16 success rate vs Indy on 4th down attempts since 2001, I just assumed he always considered it.

I guess he didn't, by his own admission. That really surprised me, as I thought he was a guy who plotted out every plausible scenario ahead of time.

12 of 16 includes a 4th and 15 last year with Matt Cassell getting picked off. If you include just Brady's makeable 4th down attempts, that's 11/15 or 80%. I like those odds.

I was under the impression that they had considered it prior to fourth down? Although the quote below could be taken either way, I guess.

From ESPN.com:

"We had a little miscommunication on that. Once some of the guys on the punt team started out, then the guys on offense they started to come off," Belichick said, noting the decision had been made on third down that they wouldn't punt. "That wasn't cleanly handled. I'll take responsibility for that."

Just to restate a bit: we need to judge on process, not on outcome. Some commenters have suggested that process is beyond what we know and, therefore, we cannot judge it. Not true. Process does not just mean 4th down. Process means the whole sequence of downs. In other words, the 3rd down call is important to our understanding of whether or not going for it on 4th is “correct”.

Many of us here agree that the 3rd down call was problematic. Some even question the 1st down call (see #21).

We would have preferred a run on 3rd down. That has the major advantage of keeping the clock moving (either forcing the Colts to call timeout or running the clock down to the two-minute warning). It has other possible advantages including Brady (and his OL!) not being subjected to Indy’s pass rush. As the Patriots appeared to be running well against Indy, that might also be something to take into account.

We also think the call on 3rd down was not a good one (see #24), though we may be on shakier ground.

The process suggested by the 3rd down call is one that many of us think was faulty. It follows therefore that the 4th down call was the result of a bad process, not a good one.

One screenshot I thought was VERY interesting during the game. Following a Colts TD, Belichick was seen talking with his defensive unit, even while the Pats offense was on the field. It made me think that he doesn't actually call most of his offensive plays, but rather delegates that process to someone he trusts.

In the instance of the controversial 4th down play, Brady has been quoted as saying that he had pushed for that particular play, as it was something he and Faulk had practiced a lot.

All of which makes me think that the particular play calls were not initiated by BB. Obviously, as the head coach, he has to take responsibility for the playcalling as well as the processes that led to the playcalling. But it did seem striking to me (as opposed to someone like Norv or Jim Zorn, who would call their own plays).

great post!!!!

A few thoughts to consider:

Models are built like road maps. You can add details, but it will never be an actual road. K.I.S.S is one of the most important things to remember when modeling. Keep it simple, stupid!

reason: once you factor in too much qualitative data, you need to quantify it to get numerical results. This quanitification of non-numerical data is ripe w/ assumptions and biases and changing variables. You can "manipulate" these numbers for any outcome you want. In fact it is easy to do, once the model is set up. Many times it is done unintentionally.

I like the simplitic approach...or road map making approach, you can see the direction and get from point A to point B, but you wouldn't want to buy your dream home from looking at a road map (limitations to the model)

Here's your explanation: http://www.advancednflstats.com/2009/09/4th-down-study-part-3.html

-------------
Sports talk radio and sports message boards are the killing fields of intellectual discourse.

Can FO please start deleting posts like this, they are a waste of everybody's time.

I don't think they could sell out and ignore the big play here, I certainly wouldn't have as D coordinator.

I haven't seen one good reason posted so far as to why NE would have a lower than average chance of making it?

IND has a good DEF, NE has an even better OFF...
IND was defending NE well. really? 34 pts in a game is doing well? NE was having a good day on OFF.

And so on and so forth.

I also wouldn't be second guessing Belicheck's play calls. I think there is a lot more subtly in how much people are lining up against the pass vs the run than people realize. I am sure BB knows that a run is more likely than a pass to get at least 1 yrd on 3rd down. I am also sure he had his reasons for choosing a pass.

even after dropping the conversion rate under 50% in the calculator, the numbers still come out extremely close, within 2% or so, depending on your other numbers

a failed conversion =/= 100% lose rate

Wrong, Wrong, Wrong

How is this so hard to understand? I mean know wonder my quant finance program was 99% Chinese. They would get this, model it in Excel, run the monte carlo simulation 5 different ways within 10 minutes and report to you on the outcomes

and that is with no understanding of American Football or who this Ball Billichuck is.

Seriously, WTF, all this discussion and you missed the whole point.

it is not a zero-sum decision.

Wow, saddened by the American public. Unless of course you are from elsewhere, then I will be a bit happier w/ us Yankees not being so stupid

Put another way, if Belichick's so damn smart, why didn't he have his team punt so he could have several chances to prove his defensive brilliance?

Somehow, I doubt that proving his defensive brilliance was at the top of his mind.

Who takes seriously the dismissal of a complex argument with the sentence "Period."? Who says, "Oh, now I'm conviced. You just said 'Period.' Now I see the folly of my ways."

In this case it was particularly funny, because after you say "Period." you go on to elaborate at great length.

I would have said the exact opposite. Going for it gives you more opportunities to win. You can win by getting the first down. If you fail to get the first down, you can win by preventing the Colts from moving 30 yards and scoring. Punting, however, removes the opportunity of winning by getting the first down -- the only opportunity you have to "make decisions" to win is on defense. And with the way the Colts were playing on offense, you were likely to have about as many "decision points" defending a 30-yard field as you would have defending a 70-yard field.

Offense and defense are not the same concerning "decisions" or choice. Choice resides with the offense in football. Offense gets to determine defensive personnel, fer gawd's sake (this one facet is exploited a lot more than we really pay attention to). The best a defense can do is to limit offensive playcalling options by being great at something (yes, sometimes this "best" is phenomenal).

Defense is not about choice, but reaction (based, admittedly, on a lot of study regarding tendencies - but still basically reacting to what the offense does). Defensive players have to keep several things in mind and be ready to react (ladies and gentlemen and Patriots fans, I give you Andy Katzenmoyer!). Pass or run? Screen or draw or downfield throw? If the offense can't stop a pass rusher what does their protection look like and what plays can they call out of that formation? Yes, sometimes down and distance - and time left - dictate the choices the offense is likely to make, but the Patriots have a very, very good offense. Their choices on any one play are not so limited. But their choices on defense are, in fact, constrained.

A team does not maximize their choices/decisions by punting. They give over decision making power to the other team. A team maximizes their control by MAKING A CHOICE on offense.

I happen to think the whole set of downs process was flawed (I don't think the Pats were in four down mode), so going for it on 4th was not the optimal decision - in my opinion. But I also think that the flaw being criticized in #65 is not a real one. (Not that anyone's mind is being changed by all this foofaraw! Great conversation - mostly. I feel like I'm in a really intelligent bar after three drinks.)

Pat you say a lot of intelligent and interesting things, this was not one of them.

"I don't see anyone taking notice that they had a better chance of winning the game than losing it just by punting (>50%)."

if we're only using generalizations, then I would say that I would rather have the 99% chance at victory once we convert than the 50% chance after a punt. these numbers are completely wrong, by the way, and there are a number of posts already explaining why, but even using your logic the call to go for it is still defensible.

What he was risking was losing the game if he didn't convert. He also risks losing the game if he punts. The argument that going is it was incorrect because you're more likely to lose if you fail to convert is flawed, because it presupposes that you'll fail to convert.
A loss does mean HFA is almost impossible, but it means that no matter how you lose, so that's irrelevant. The other two statements boil down to punting to avoid criticism. If that's your argument, well, it's not necessarily wrong - as you point out, a team that punts is still more likely to win than lose - but I hope my team has a coach who makes decisions without worrying what they'll say in the papers the next day.*

*I should note that, because "my" team is the Lions, I will likely never know if they have such a coach.

Belichick was risking by going for it there just to gain a few percentage points in win probability.

A *few* percentage points? Dude, if they converted it in bounds, they won. That's a lot of percentage points.

After the Pats didn't make the 3rd down, every Colts fan worth their salt KNEW Belichick was going for it. I mean, we've watched this guy every year, and sometimes twice a year, for nearly a decade. We KNEW that he was going for it, because that's what he does.

As for how nervous we were, I bit off three fingernails between the end of the 3rd down play and the end of the 4th down play.

As for after the play, we really weren't thinking "We got this." It was more a "Let's see if Manning can earn that ridiculous contract he's about to get."

It wasn't over until Reggie scored. And even then, we were all sitting there going "Oh hell, it's going to be a Hail Mary with Randy Moss streaking toward the end zone, and if the Pats protection holds up for five, six seconds, he's going to drop a bomb right into his hands and we're screwed."

Basically, it wasn't until the clock hit 00:00 that we finally went "Holy crap we won".

For what its worth, the Colts radio play-by-play announcer and the colorman (former all-pro lineman Will Wolford) were downright GIDDY when the Pats lined up to run the fourth down play. Of course, they were even happier after the play failed, but prior to the play they were figuratively "licking their chops" for the chance to stuff the play in BB's face.

Ok. Thank you for the answers. My impression on TV was slightly different, but I was emotionally involved besides being a Giants fan. What a GAME it was!

nobody's saying he scores on more than 50% of his drives after a punt; they're saying that if he scores on more than 34%, that's one time in three, then going for it provided a better chance to win.

as to the third down/being outcoached...I don't think you'll get too many arguments, even from bill

My favorite post thus far. Well said.

A guy won state somewhere without ever punting. Its been done in high school, and people decided his results didn't translate up to higher talent levels without trying it.

Like the spread option in the NFL

I think "put their careers on the line" explains what you're describing, though. If the Patriots had punted and Manning had driven the Colts down the field to score the TD and win the game, nobody would have criticized Belichick for punting on 4th-and-2. (They might have criticized him for other things, of course.)

A coach who isn't 100% sure that the ownership/management is completely behind him might quite rationally (for his own self-interest) pick a non-optimal team strategy for any individual game, on the thinking that for his own career, it is better to lose conventionally than unconventionally, even if it lowers his overall winning percentage. The coach's self-interest and the team's interests are not always aligned, especially during an individual game.

Umm, for the same reason that baseball managers sacrifice bunt and issue intentional walks?

The Colts scored a touchdown 4/4 times within the 30 that game. 85% is roughly the point at which the likelihood of getting 4/4 exceeds 50%. 85% is probably the best guess that Belichick could've made.

54% seems wildly unsupported by the evidence from that game.