November 15, 2009

Fourth and two

With the New England Patriots leading the undefeated Indianapolis Colts 34-28 and 2:08 left on the clock, Patriots coach Bill Belichick decided to go for a first down on 4th and two yards to go at his own 28 yard line rather than punt and give Peyton Manning the ball back.

This was a violation of traditional coaching practice to always punt in that situation, but, after all, offenses, especially short passing offenses, are now much more reliable than when this tradition was invented decades ago. This also means the other team's offense is better, too. So, if you are Belichick, the question is, "Whose hands do I want on the ball: Tom Brady's or Peyton Manning's?"

The improvement in offenses is why you never see the "quick kick" (an unexpected punt on third down) anymore. As recently as the 1971 USC-UCLA, the Bruins surprised the Trojans by pitching out to halfback Greg Jones on third and long only to have him pull up and punt (using a sideleg topspin-inducing punting style he'd secretly practiced). With nobody on USC playing back to receive it, the ball finally rolled dead after almost 70 yards, pinning USC deep in their own territory. This helped UCLA pull out a 7-7 tie.

But there aren't many 7-7 ties anymore (at least not played on dry fields where a punt might roll a long ways), so nobody punts on third down anymore.

So, why should they punt on fourth down either?

A study by economist David Romer a few years ago argued in favor of teams going for the first down on fourth and two even on their own ten yard lines, even in the first half.

Unfortunately, according to Romer:
"Decisions to go for it on fourth down (that is, not to kick) are sufficiently rare, however, that they cannot be used to estimate the value of trying for a first down or touchdown. I therefore use the outcomes of third down plays instead."

In other words, what he actually discovered that it's a good idea to go for it on third and two at your own ten yard line instead of quick-kicking it away. But we already knew that you should go for the first down on third-and-two. The odds of you making it are pretty high because the defense is playing fairly back so that you don't score a 90-yard-touchdown on them. On 4th-and-two, however, the payoff to the defense from crowding the line to prevent a short gain is much higher, so Romer's analysis isn't worth too much.

A better way to analyze fourth and two decisions is from data collected on two-point conversion. In the NFL, a team that scores a touchdown gets to either kick a one-point conversion, or, from the two yard line, attempt a two-point conversion by running or passing the ball across the goal line. The defense doesn't play back at all because there is nothing they can give up worse than a two point conversion.

At present, the odds of making it from two yards out appear to be not over 50%, because no NFL team regularly tries a two point conversion. The universal default in the NFL is the one point conversion, which has nearly a 100% success rate, putting the expected value of kicking the conversion at just a little under 1.00 points. Hence, the success rate on two-point conversions can't be much over 50 percent because NFL teams today only go for two when there is some strategic reason to do so.

However, the success rate has been going up. The NYT reported after the 2006 season:
In each of the last two seasons, N.F.L. teams have made slightly more than half of their 2-point conversions, up from less than 40 percent in the late 1990s.

Eventually, an NFL offensive juggernaut might start going for two after each touchdown, but that hasn't happened yet. Coaches would rather have their players lose the game than the coach lose the game.

One thing to keep in mind, though, when apply two-point conversion rates to fourth-and-two rates is that two point conversions are usually attempted when the offense is hitting on all eight cylinders, while 4th and two attempts are made when the offense is sputtering.

So, all this theorizing is interesting, but you still have to execute on the football field, which the Patriots did not: Brady hit Kevin Faulk, running a pattern where he was coming back toward the line of scrimmage for a three yard gain, but Faulk juggled the ball and didn't grab it firmly until he was only a yard past the line of scrimmage, turning the ball over to the Colts.

Not surprisingly, Peyton Manning marched them 29 yards for the winning touchdown.

My published articles are archived at iSteve.com -- Steve Sailer

42 comments:

Anonymous said...

FYI, I know of at least one instance around 2002 where the Virginia Cavaliers did a "quick kiok" with quarterback Matt Schaub. It wasn't anything terribly important, but the practice isn't completely extinct.

eh said...

It was a bad decision and a quasi-insult to his defense.

Steve Sailer said...

In football games during 7th grade recess, quick-kicking was my favorite strategy. Both teams would be too disorganized to move the ball forward, so punt on first down and try to eventually get a safety to win 2-0.

But that's like telling your team in baseball never to swing. It's not the spirit of the 7th grade recess football game, so I usually got overruled.

Anonymous said...

Gregg easterbrook has been saying go for it on fourth for years on espn.com

Belichick made the right call- would have won the game right then and there

Harder to make a two-pointer than random 4th and 2 bc defense only has to defend twelve yards of territory

Great game tonight- New England is the better team though- would be favored by 5 on neutral field imo- Brady will beat Brees in the superbowl imo

Dan in DC

Bob said...

Football would be a lot more exciting if there were more 4th down plays and more attempts at 2-point conversions.

It'd be easy to tweak rules to favor more 2-point conversions, just move the line 20% closer to the end zone.

Not sure how to discourage punting just a little on the margin.

Jim Bowery said...

What would men do without sports to distract them?

I'm just sayin'...

Mr Lomez said...

"Harder to make a two-pointer than random 4th and 2 bc defense only has to defend twelve yards of territory

Yup. This method is just as flawed as using 3rd and 2's.

Good call by Belichik. The difference between getting stuffed on 4th down and punting is roughly 50 yards. Given Manning's history in these situations, that 50 yards is worth about a 15% increase in the likelihood of a win for the Pats. Compare that to the infinite increase in likelihood of a win if they convert on 4th down. It was an odds play. A good one, even if it didn't pay off.

green mamba said...

It was a terrible call by Belichick. If they had been near mid-field, ok, but at their own 28? Leading by less than 7 points?

Anonymous said...

football players are much more specialized these days. so unlikely that they even know how to kick - unlike rugby which takes all around developed skills.

Anonymous said...

Football would be a lot more exciting if
- in rugby you take the extra point from the imaginary perpendicular line from where you crossed the try (goal) line - not a preset hashmark - thus if you score in the corner its a tough extra point kick.

- rugby has only two subsitutions

- you actually have to 'touch the ball down (get it?)

- rugby has a continous clock

all of these rule changes would make football much more exciting - and there would be a hell of a lot more white backs as blacks would start dropping like flies after the first quarter or so

sykes.1 said...

The psychology of two point conversions, fourth and two, three and two, etc are very different, and this must affect the way the players respond. Also, the quality of the offensive and defensive play that leads up to these situations is also different. All in all, I don't see how you can extrapolate from one case to the others.

Usually Lurking said...

What would men do without sports to distract them?

Distract us from what?

Sports are not a distraction, they are an attraction.

Anonymous said...

It'd be easy to tweak rules to favor more 2-point conversions, just move the line 20% closer to the end zone.

Don't mess with football! God, I hate liberals.

Anonymous said...

Steve,

I see you remember UCLA HB Greg Jones. I wonder what ever happened to him? I recall reading that he intended to go to law school.

burger flipper said...

I think it was a good call. Punting would give you an extra 35-40 yards, not fifty.

Lost in all the jibber jabber I've seen on ESPN is the possibility the Patriots D might have stopped them from the 30.

Back of the envelope:

Punt:
25% chance the Colts can score a TD from around their 35.
Pats win 75% of the time

Go for it:
60% chance they convert it (probably conservative)
40% chance they can stop Manning if they don't
Pats win 76% of the time.

Marc B said...

Manning could have scored the winning TD from his own six yard line with over 2 minutes on the clock & three timeouts. That's why the coach felt it was so important to risk giving the Colts field position to keep it out of his possession.

It would have been a hailed as a bold and genius manner to secure the win had they converted. They should have done a QB sneak or a hand off for the two freaking yards, though.

The quick kick was used this weekend by the opposing team at my nephews high school playoff game. Took everybody by surprise, and added some excitement to an otherwise lopsided victory.

OhioStater said...

Belichick's play was very aggressive, but we are lucky this weekend offered a conservative alternative on display at Ohio Stadium. Iowa and Ohio State could have taken chances but both took the safe road leading to overtime and eventual Ohio State victory. Go Bucks!

Losses hurt more in college football, and I think New England will be fine. Maybe this call will help New England down the road as it communicates the message to the Pats defense "I don't trust you".

Danny said...

What would men do without sports to distract them?

Yeah you're right Bowery. If men didn't waste so much time on sports they would have colonized the galaxy by now and increased the solar system's "carrying capacity." No doubt you would have been reading Sailer's blog from Pluto or something.

What you fail to recognize is that your "techno-fetishism" is not really any different from obsessive male sports fans.

Moses said...

Belichick plays to win. This is exactly like his first Super Bowl victory. I remember an announcer (Madden, IIRC) saying he should run out the clock and go to OT. Instead, the Pats drove into field goal range and won the game.

The main factor in this call is the outcome if he succeeds. During the game, the Pats would still need to get into field goal range at a minimum to make the call worthwhile. In this case, it's equivalent to 4th and goal from the 2-yard line.

My "favorite" no guts call is by underdog coaches when they trail by 1-pt at the end of the game, having just scored a touchdown. The guts of the coach is what made Boise State's Fiesta Bowl win over Oklahoma so sweet.

Anonymous said...

"Coaches would rather have their players lose the game than the coach lose the game" - what a brilliant insight! I actually think Belichick was a fool for the fourth and two call, but he showed a lot of courage. He could easily have hid behind conventional wisdom and let responsibility for the loss go to the players.

Bill said...

Substituting 3rd down results for the missing 4th down results is obviously a major weakness of that paper. Substituting two point conversions for 4th and twos in general is a bad idea for the field length reasons already discussed.

In the published version of the paper (Journal of Political Economy, 2006, vol 114, no 2), this is discussed on pp 351-2. Outside the opponent's 17, the prob of converting a 3rd and 1 is 64%. Inside the 17 it is lower and apparently declining as you get closer to the goal line. The text is less clear than one would like, but it looks like 38% is a reasonable guesstimate at the 5 yard line. This is a big difference (64 v 38 coming from field length differences).

In the published version (but not, I think, in the working paper version on the web), Romer compares 4th down attempts to 3rd down attempts, here looking at almost the whole game instead of just the 1st quarter to get sample size up. He concludes (p 357):

"The point estimate corresponds to the probability
of success being one percentage point lower on fourth downs than on
third downs, which would have almost no impact on the analysis"

So, Romer's approximation is off by around 1% (+/- quite a bit presumably because of small sample size); whereas, Steve's is off by around 20%. All on Romer's accounting, of course.

Steve's hatred for this paper is one of the minor mysteries of this blog.


On 4th-and-two [relative to 3rd and two], however, the payoff to the defense from crowding the line to prevent a short gain is much higher, so Romer's analysis isn't worth too much.


The words much higher are doing a lot of work here. You have to consider the effect of crowding the line on breakaways/long passes. Using the part of the paper Steve seems not to hate, the value of stopping the opponent on the fifty yard line on 4th and short is about 2 points, on average (Figure 1 says this). Crowding the line drives the probability of getting these 2 expected points higher. It also drives the probability of losing 7 on a long TD higher. How much extra crowding you do depends on how the extra crowding affects these two probabilities --- Steve is assuming that you can drive down the prob of stopping on 4th and short a lot without giving up much in terms of enhanced big plays.

Jim O said...

As a Jet fan I hold no brief for Belichick, but this wasn't a bad decision at all. If they make it, (and the odds of making it were about 50%, agreed?), the Pats win. Period. If they don't make it, they don't definitely lose. A punt can be blocked or run back to the original line of scrimmage, or beyond. The Pats could stop them from scoring a TD from the 28. Not imposssible, for Crissakes. And a field goal would not have won for the Colts, who would have had to follow up with a successful onside kick, and try again for another score! And if that next score, it it were made, was not fated to be a touchdown, it's overtime, and the Pats can still win the toss, score, and win the game. And the way the Colts were moving the ball in the second half, who can say that the net 35-40 yards that a typical punt would have gained the Pats would have made any difference?

The vitriol coming out in reaction to this one just proves how much people hate Belichick, IMHO. Not that that's inappropriate.

mnl said...

The very fact that this is the topic of Steve's post--as well as, no doubt, numerous carpool discussion this morning--highlights the reason so few coaches go for it on 4th and short: no coach got ridiculed or fired for punting on fourth down. Or, as Steve put it... "Coaches would rather have their players lose the game than the coach lose the game."

But my understanding of Romer's statistical analysis was that it makes sense to go for it on fourth and short WHEN IN THE FIRST HALF of the game. At that point, you've got a great bit of the game yet in front of you, more room to correct if such a fourth and short play falters.

Belichik's style may be a bit "cowboy" and we can't fault him for being aggressive here yet laud him for the same aggressive behavior elsewhere and when it pans out. That's unfair. But just the same... it was an insult and unfair to his defense to not punt the ball and give them more room to stop the Colts.

Going for it on fourth and short in the FIRST HALF and in OPPOSING territory is wisely aggressive. Going for it on your own side of the field with two minutes left? Not so much.

rakeback said...

This will all be forgotten at the end of the year, because the eventual Super Bowl Champion is all that really matters. I think it will be Favre v Brady, with Favre riding off into the sunset with an MVP in the biggest game of the year.

Anonymous said...

I like this topic bc it's one of the few where i'm an expert- again great call by Belichick- you guys broke down the odds...some may disagree but people who REALLY know football don't- that said almost all of the pundits and experts (Dungy, harrison etc) said it was a mistake- I'm always shocked about how little these experts really know about the game.Case in point- Peter King's life revolves around football but he is absolutely worthless- Here are two of his worst predictions- Danny Wuerffel barely hanging on to an NFL spot with the Saints would throw for 4000 yards under Spurrier's Redskins...he, uh, didn't- I think his experience at QB was so bad he doubted his faith in God. He was sacked so many times against Dallas that the head trainer used a rape kit on him. Test came back negative but still.

Another of King's gems was saying the Browns should draft Ohio St QB Troy Smith with their first pick in the draft (3rd overall)and make him their QB of the future. Troy Smith was drafted in the 5th rd after the ravens reached and grabbed hime- he is better than expected- meaning he is a third string qb. You can not possibly know ANYTHING about the game of football and think Smith deserved a first pick- nothing. He is the premier football writer in the country btw- he's also a dumbass liberal but that is besides the point. He does write an interesting column bc it's so gossippy- he should be working for TMZ and not SI.

Dungy is a great man but average coach- Gruden had to get Tampa over the hump and they were extremely lucky to win the Super Bowl- New England was better and they were lucky they didn't have to go out to San Diego to play the Charger juggernaut after SD s### the bed in the final minutes aginst NE.

Dan in DC

rast said...

rugby has only two subsitutions ... all of these rule changes would make football much more exciting

Riiiight.

jody said...

this was perhaps the worst decision i have ever seen in 25 years of watching NFL football, and here is why. note that pretty much everybody, even the professional analysts, miss the reason why this was a shockingly bad decision. everybody, EVERBODY, needs to look backwards one patriots possession. forget, forget, FORGET about the now famous 4th and 2 call. go back one possession to this:

the patriots have the ball on the COLTS 18 yard line with a 10 point lead and 4 minutes to play. if the numbers, the percentages, support going for it on 4th down on your own 28, HOW ON EARTH can they not support going for it on 4th down when you have the ball on the OPPONENT'S 18 yard line? there is dramatically less downside here to missing the conversion! if the patriots scored a touchdown on that possession instead of a field goal, THE GAME WOULD HAVE ALREADY BEEN OVER AT THE 4 MINUTE MARK. that would have forced the colts to score 3 times instead of 2 times, which would have been almost impossible in 4 minutes.

everybody, EVERY PROFESSIONAL, who is paid hundreds of thousands of dollars to analyze this stuff, has totally, completely missed this part, the MOST IMPORTANT PART OF THE ENTIRE GAME.

why not "win the game right now" at the 4 minute mark. if you miss the conversion, the downside is SO MUCH LESS WORSE. turn the ball over on the opponent's 18, with a comfortable 4 minutes left and 3 full timeouts. why the total, TOTAL BS call of "winning the game right now" on your OWN 28 with the downside that if you do not convert, you "lose the game right now".

it's sheer idiocy that people are defending bill belichick here. the time to go for it on 4th down was the PREVIOUS possession.

jody said...

just so it is CRYSTAL CLEAR why this was perhaps the worst decision in at least a decade:

if the patriots were so sure, so certain that the colts were definitely going to score 2 touchdowns, or 14 points, on them at the end of the game, THEN A FIELD GOAL MAKES NO SENSE at 4th and 8 from the colt's 18 with 4 minutes to play. that only gives them a 13 point lead. that means THEY LOSE when peyton manning does his thing and the colts score 14 points on the patriots. so:

1) go for it on 4th and 8 and do not convert, 10 point lead, still lose to colts by 4 points due to manning magic producing 14 points on you in 4 minutes.

3) field goal on 4th and 8, 13 point lead, still lose to colts but lose by 1 point instead of by 3 points.

2) go for it on 4th and 8 and convert, take probable 17 point lead, defeat colts by either 3 or 7 points.

if the patriots wanted "to win the game right now", they would have tried to score a TOUCHDOWN at the 4 minute mark, going up 17 points, and forcing the colts into a near impossible situation of needing to score 3 times.

bill belichick should have gone for it on the PREVIOUS POSSESSION. instead he made the worst decision i have ever seen a great coach make.

the math DOES NOT support what he did. the statistics DO NOT support what he did. the numbers DO NOT support what he did.

Anonymous said...

Jody, he probably should have done both. On the earlier possession was it 4th and 10? 2? that makes a difference. I say probably bc going up 13 means the Colts have to score 2 tds which is much more difficult than 1fg and 1 td. Actually, I think Belichick made the right decision on both accounts. He is the best coach in the NFL and they will win the SB this year over the saints btw.

Jody originally I thought you were gonna say "he should have given it to Heath Evans!!!"

Dan in DC

DAJ said...

all of these rule changes would make football much more exciting - and there would be a hell of a lot more white backs as blacks would start dropping like flies after the first quarter or so.

The NFL is already sufficiently exciting as it is the most popular spectator sport in the U.S. Besides, why would you want to rig the game to benefit whites at the expense of blacks? Again, HBD will never catch on as long as people can sense underlying animus against the black race in more than a few of its proponents.

Anonymous said...

Question:

Why go with a pass play? Why not a run, a QB Draw, or a fake punt?

Passing seems the most risky of options.

Truth said...

we PLAY...to WIN...the GAME!
-Herman Edwards

You don't get anywhere in life without taking BIG risks; and yes, occasionally they don't work.

CJ said...

"I think it was a good call. Punting would give you an extra 35-40 yards, not fifty.

Lost in all the jibber jabber I've seen on ESPN is the possibility the Patriots D might have stopped them from the 30."


Good points, Burger Flipper -- your talents are wasted back there on the grill. However, your numbers could be somewhat off because there are still more permutations. Whenever a team punts, things can go wrong -- dropped snaps, blocked kicks, short punts, big runbacks, penalties. The percentage chance for Belichick's decision to go for it is IMO even better than you say.

And you know, I don't want to go out on a limb here, don't want to get too far outside the box, but I'm actually ready to entertain the notion that Bill Belichick knows more about football than I do. He may even have specific information about the New England Patriots that I don't have.

jody said...

"he probably should have done both."

no way. NOBODY EVER MAKES THAT CALL. and now we know why.

remember in the superbowl, when, during a field goal, a cardinals player ran over the steelers holder? and the steelers got to advance the ball, instead of getting a field goal. the reaction around america was "Oh my goodness, what a terrible call ref, who calls that, I've been watching football for 30 years and I've never even heard of that call."

well, that's because NOBODY IS STUPID ENOUGH TO DO THAT. we've never seen it, because it's so basic, so obvious, nobody does it. likewise, YOU DON'T GO FOR IT ON 4TH DOWN ON YOUR OWN 28 WITH ONLY A 6 POINT LEAD. YOU PUNT. YOU PUNT. YOU PUNT.

against the colts, YOU LOSE THE GAME if you miss the conversion!

please research and get back to me on all the times coaches made a decision like that. like the cardinals player running into the steelers holder during the superbowl, it is something we NEVER see, because you would have to be AN IDIOT to do that.

"I thought you were gonna say "he should have given it to Heath Evans!!!"

heath evans is on the saints. he played 6 games this year, scored 3 touchdowns, then got hurt. he's out for the year.

"Why go with a pass play? Why not a run, a QB Draw, or a fake punt?"

because they have one of the best quarterbacks ever, because they have two great receivers, because "the numbers support belichick!" (ok, that last part was not true.)

my question was, why not throw a pass, you know, 1 or 2 yards BEYOND the first down marker.

yet another thing the patriots utterly screwed up but which experts have completely overlooked. you are ALWAYS, ALWAYS supposed to throw it BEYOND where you need to go in situations like that. it's like intercepting the ball on 4th down at the end of the game when your team is ahead - YOU DON'T. you BAT THE BALL TO THE GROUND, so there is no opportunity for you to then FUMBLE like a moron and turn the ball over to the other team so they can win.

this is BASIC FOOTBALL STUFF. jumpin' jehoshaphat.

the patriots EVEN screwed up their timeouts! they called a timeout after the kickoff! WTF WAS THAT.

the defense of the patriots in this situation is ludicrous. they clearly, CLEARLY lost the game on ALL-TIME bad decision making.

Anonymous said...

Heath Evans is with the saints now? ouch major pwnage...

The rest of your rant doesn't make sense- yes don't run into the holder ok but what BB did made sense % wise....ehh forget it neither of us will come around

Dan in DC

Michael said...

"Decisions to go for it on fourth down (that is, not to kick) are sufficiently rare, however, that they cannot be used to estimate the value of trying for a first down or touchdown." -Romer

I don't see this. Surely if you looked at 20 years of data you could get a sample size big enough that it would at least be worth reporting.

And even if it was too small to draw confident conclusions, it would still be interesting and certainly more relevant than data on 3rd and 2.

Though to be honest Steve, I'm not sure how relevant the 2 point conversion data is to this issue, as in such a situation giving up 2 yards is the same as giving up 10 yards, giving the defense much more incentive to crowd the line than they would in an actual 4th Down Situation.

Steve Sailer said...

Right, while the 2-point conversion data would be relevant to the Colts-Patriot's fourth and two situation where the Colts didn't care whether they gave up a touchdown or a first down, it would underestimate the odds of making it on fourth and two in less climactic situations than that.

Anonymous said...

Steve wtf? i said that in the 4th post on here and you give credit to this "Michael" fella. I want ten dollars back

Robert said...

"Lost in all the jibber jabber I've seen on ESPN is the possibility the Patriots D might have stopped them from the 30."

So - maybe Belichik's decision showed that he had great confidence in his defense's ability to stop the Colt's from scoring from the 28 if the Pats didn't get the first down. Remember, they needed a touchdown, not just a field goal.

The decision, even though it may have been wrong, was not an insult but a compliment to his defense.

ben tillman said...

please research and get back to me on all the times coaches made a decision like that. like the cardinals player running into the steelers holder during the superbowl

Mackovic for Texas in the Big 12 championship game (successful). Switzer for the Cowboys vs. the Eagles (unsuccessful). That's all I can think of at the moment.

Truth said...

"Why go with a pass play? Why not a run, a QB Draw, or a fake punt?"

1) they don't have a good RB

2) It would have taken Tom Brady 15 seconds to run two yards.

3) You have to back the ball up 12 yards and put it in the hands of a non-offensive factor to gain two. Small chance.

luc schouten said...

Vince has played extremely well the past few weeks. I think this season will decide his future with the team and whether or not they will go another direction.