Editore"s Note
Tilting at Windmills

Email Newsletter icon, E-mail Newsletter icon, Email List icon, E-mail List icon Sign up for Free News & Updates

May 4, 2006
By: Kevin Drum

GOING FOR IT....Enough about politics. Let's talk about football.

I've been convinced for years that football coaches are way too conservative on fourth down: instead of punting or kicking field goals, they'd be better off going for it if they have only two or three yards to go for a first down. With the notable exception of Pete Carroll, though, mostly they don't. (And the only fourth down of Carroll's that anyone remembers is the one that went disastrously wrong in the Rose Bowl this year. But it's worth remembering that for the past five seasons, they mostly went right.)

Now, however, I'm happy to report (via Tyler Cowen) that David Romer of UC Berkeley has written an exhaustive analysis filled with sigma signs and subscripts that provides a quantitative answer to this burning question: exactly when should you go for it and when should you kick?

The solid line in the chart below provides the answer. At the 50-yard line, you should go for it if you have less than five yards to go. At the 40-yard line you should go for it if you have less than seven yards to go. At the 35-yard line you should go for it no matter what. Beyond the 33-yard line, as you get into field goal range, the value of kicking rises and the "critical value" necessary to go for it declines steeply (though it stays above four yards all the way to the goal line). The dashed line summarizes actual coaching decisions over the course of the study and shows that, on average, coaches go for it only if they're past midfield and have only about two yards to go. That's much too conservative.

Romer's analysis accounts for the probability of making a first down and then going on to score; the likely field position of your opponent depending on whether you kick or not; the likelihood of making a field goal; and a whole variety of other factors. Read the whole thing if you want to argue with him.

But the bottom line is simple: always go for it if you have less than three or four yards to go. Past midfield, you should go for it even in higher yardage situations until you get into field goal range. But even then, you should go for it if you have less then three or four yards to go.

In other words, Pete Carroll is a smart guy. It was LenDale White's fault that USC lost the Rose Bowl.


Kevin Drum 5:47 PM Permalink | Trackbacks | Comments (91)

Bookmark and Share
 
Comments

You've really crossed a line here, man. You are a sick, sick wonk.

Posted by: BarrettBrown on May 4, 2006 at 5:52 PM | PERMALINK

Greg Easterbrook will be happy to read this post!

Posted by: chris brandow on May 4, 2006 at 5:53 PM | PERMALINK

I believe the report uses NFL statistics and rules as opposed to the college game. While obviously similar, the college game has a few more variables to deal with, namely, the quality of a team's kicker, punter, and special teams. At the pro level, it's assumed that these units have a certain level of competency.

Posted by: Double B on May 4, 2006 at 5:56 PM | PERMALINK

Double B: Yeah, Romer's data was all from pro games. But since LenDale White's failed fourth down run in the Rose Bowl this year is still eating away at my vital organs, I figured I should work that in somehow.

Posted by: Kevin Drum on May 4, 2006 at 5:59 PM | PERMALINK

This completely fits into the psychology of investing, as well. People are demonstrably more afraid of loss than they are motivated by gain. You might call it the "bird in the hand" rule. So taking a field goal is never wrong, whereas going for it, especially in field goal range, is crazy scary.

Posted by: craigie on May 4, 2006 at 6:01 PM | PERMALINK

the one that went disastrously wrong in the Rose Bowl this year.


What was wrong about it?

Love,
The Longhorns

Posted by: ckelly on May 4, 2006 at 6:03 PM | PERMALINK

You can also do the math and show that going for the win is better than going for the tie in the closing seconds of a game if your chance of making the game-winning touchdown or two-point conversion is at least 50%, unless you are clearly more likely to win in overtime.

In general, I detest risk-averse strategies, whether they occur in sports, in politics, or in war. (Thought you could get away with some non-political stuff, did ya?)

Posted by: Anthony on May 4, 2006 at 6:03 PM | PERMALINK

It all well and good that USC has a good intuitive coach, but note that it took a guy from Cal to actually explain why it works.

Go Bears!

Posted by: Nonplussed on May 4, 2006 at 6:04 PM | PERMALINK

Which is why the Berkeley football team is nationally ranked year after year.

Posted by: Skeptic on May 4, 2006 at 6:08 PM | PERMALINK

Kevin, regarding that fateful play this past January: I wouldn't be so quick to judge LenDale White. Why did Carroll have Reggie Bush on the sidelines during that play. Sure, everyone in the stadium new it was going White on that play. However, with Bush, the best player in college football, in the back field, Texas would have needed somebody off of the line in case he went out for a short pass from Leinert. That may or may not have allowed White to power through that last 6 inches. No matter how happy I was with the outcome, I keep going back to that.

Posted by: BD in Texas on May 4, 2006 at 6:10 PM | PERMALINK

"It was LenDale White's fault that USC lost the Rose Bowl."

LenDale White didn't forget to send in Reggie Bush as a decoy.

Posted by: scouser on May 4, 2006 at 6:13 PM | PERMALINK

It was LenDale White's fault that USC lost the Rose Bowl.

Reggie Bush is certainly culpable given his incredibly poor judgement when attempting his unforgettable "lateral".

Posted by: Edo on May 4, 2006 at 6:13 PM | PERMALINK

The problme is that while we wonks may read this and lament the fact that more coaches don't go on 4th down, your average football fan perhaps is not so aware of this, and is more than happy to punish the coach come Monday morning for going for it on 4th and inches and then giving the ball up to the team that goes on a massive scoring drive to win the game.

Posted by: Alexander Wolfe on May 4, 2006 at 6:16 PM | PERMALINK

BD in Texas,

White was the short yardage and goalline back all year for the Trojans. Huff (the Texas safety) has admitted to cheating up to stop the play (Power "O"), which Texas had struggled with much of the game. I think the better call would have been a play action pass to the TE.

Posted by: Double B on May 4, 2006 at 6:17 PM | PERMALINK

I find it hard to blame anybody on the USC offense for the loss. How many punts did they make in the game? 1 or 2. They lateral was poor, Leinart's pick was poor, but they still lit Texas (a fantastic defense) up for 38 (they would have scored 60 on their own defense). What in god's name was the defensive game plan for USC? They didn't stop Young from running or THROWING. Receivers were wide open. The blame starts and ends with them.

Posted by: Double B on May 4, 2006 at 6:20 PM | PERMALINK

Who was at fault when Vince Young ran in that last touchdown?

Posted by: j on May 4, 2006 at 6:24 PM | PERMALINK

From the conclusion of the paper:

"Much economic analysis is built on the idea that the assumption that agents maximize simple objective functions leads to reasonably accurate descriptions of their behavior. This paper demonstrates that in a case where this hypothesis can be tested directly, it fails. Specifically, the paper shows that the behavior of National Football League teams on fourth downs departs systematically from the behavior that would maximize their chances of winning. The departure is large and overwhelmingly statistically significant, and it cannot be explained by rational risk aversion, information known to teams that is omitted from the analysis, or other complications. This is true even though the decisions are comparatively simple, the possibilities for learning and imitation are unusually large, the compensation for the coaches who make the decisions is extremely high, and the market for their services is intensively competitive. Despite these forces, the coaches who fail to make maximizing choices are not fired and replaced by ones who do. How can this be? "

Indeed how can it be?

I suspect it is the curse of conventional wisdom.

Generations of football coaches plodding along in the worn tracks of their mentors.

Really no different than when folks assume the religion, the politics, and the prejudices of their parents and friends.


Posted by: koreyel on May 4, 2006 at 6:26 PM | PERMALINK

There's a crucial flaw, of the usual "assume a can-opener" type, in Romer's assumptions. Because essentially no one tries for fourth-down rushes (or passes, for that matter) Romer uses data from third-down plays instead. There is obviously a lot more light under the lamp post, but there's really nothing to say that the distribution of play outcomes (or the distribution of plays chosen) would be the same on fourth down as on third. There's a pretty good argument to be made from goal-line stands that the chances of making a particular amount of yardage are smaller when it's known that a turnover or a score would result. (And that's before you get into fumbles and interceptions.)

That said, the notion that people depart from profit-maximizing behavior in favor of risk aversion or of simple rules that won't get them into trouble by making them stand out is pretty obvious.

Posted by: paul on May 4, 2006 at 6:27 PM | PERMALINK

I have no statistical evidence to back this up other than the fact that I watch way too much football every Sunday during the season, but... it seemed like NFL coaches went for it an awful lot last season, relatively speaking. Maybe they are already catching on.

Posted by: Rick on May 4, 2006 at 6:36 PM | PERMALINK

Anybody remember Sam Wyche? He had a similar philosophy in Cincy, at least when their offense was good. I remember him as an innovative coach who just couldn't keep himself from doing stupid things (like running it up against Glanville and the Oilers, and any number of public statements).

And White lost the game? A little more blocking on that play, a different call, or even just a little more defense and USC wins.

Posted by: mwg on May 4, 2006 at 6:37 PM | PERMALINK

Greg Easterbrook is TNR scum.

Posted by: FTW on May 4, 2006 at 6:44 PM | PERMALINK

Does this analysis take into account what the score is? Because that's a not-so-insignificant determination of how risk averse a coach should be.

Posted by: govols on May 4, 2006 at 6:47 PM | PERMALINK

What I find most fascinating about that graph is that at the one single point where real and ideal behavior coincide (ball on the opponent's 20 with ~4 yards to go) the only reason that they do coincide is because the collective coaches' instincts are all wrong.

That is, Romer's data indicates that the closer you get to the goal line inside the 32-yard line, the less often you ideally should go for it, while right around the 20, real-life coaches' efforts to go for it spiked.

Posted by: The Confidence Man on May 4, 2006 at 6:47 PM | PERMALINK

The paper has been around for about 4 years now. Granted it was up on NBER...

Bellman equation still gives me nightwares. Intertemproal utility optimization was never my strogest suit.

Posted by: clone12 on May 4, 2006 at 6:54 PM | PERMALINK

Check out foonote #4. He cites a paper written by Carroll, Palmer, and Thorp in 1998. :) I know who the first two are, but I don't know who Thorp is. I had no idea that Carson Palmer and Pete Carroll collaborated on econometrics papers.

Posted by: mrjauk on May 4, 2006 at 6:56 PM | PERMALINK

"Much economic analysis is built on the idea that the assumption that agents maximize simple objective functions leads to reasonably accurate descriptions of their behavior. This paper demonstrates that in a case where this hypothesis can be tested directly, it fails. Specifically, the paper shows that the behavior of National Football League teams on fourth downs departs systematically from the behavior that would maximize their chances of winning. The departure is large and overwhelmingly statistically significant, and it cannot be explained by rational risk aversion, information known to teams that is omitted from the analysis, or other complications. This is true even though the decisions are comparatively simple, the possibilities for learning and imitation are unusually large, the compensation for the coaches who make the decisions is extremely high, and the market for their services is intensively competitive. Despite these forces, the coaches who fail to make maximizing choices are not fired and replaced by ones who do. How can this be? "

Michael Lewis went over much this same ground in "Moneyball." The coaches who went by the conventional wisdom (i.e. hire a baseball player who looks like a baseball player) were more highly regarded, while the few coaches and GMs who ignored the CW to dive down into statistical analysis were roundly derided as kooks and eccentrics -- even after they started winning.

Or look at Iraq. Those of us who said from the beginning that it would be a fuck-up of epic proportions have been proven right, while the "sensible" moderates and liberals have been proven wrong -- and yet even now it's the "sensible" ones who are still employed as pundits and analysts, their failure swept discreetly under the rug. We were right, but apparently that doesn't matter.

All of which shows that in life, it's considered better to be wrong in the "right" way than right in the "wrong" way -- you'll be fine as long as you stay in the middle of the herd, but stray too far and you'll get chopped.

Posted by: Stefan on May 4, 2006 at 6:59 PM | PERMALINK

A team on its own 10 yard line should go for it if it has two and one-half yards to go? Come on.

Posted by: dogfacegeorge on May 4, 2006 at 6:59 PM | PERMALINK

Come on. According to the graph, a team on ITS OWN 11-yard line should go for it on fourth and 3.
If you fail to make the three yards, you hand the other team at least three points, and they only have to travel at most 13 yards for a touchdown. If you succeed, you still have to march approx. 60 yards for a chance at three points.
I find it hard to believe that going for it on fourth and three from the shadow of your own goal posts is routinely the optimal choice.

Posted by: John on May 4, 2006 at 7:09 PM | PERMALINK
Check out foonote #4. He cites a paper written by Carroll, Palmer, and Thorp in 1998. :) I know who the first two are, but I don't know who Thorp [SIC] is.

Dude! Jim Thorpe?! Arguably the greatest all- around athelete in 20th century? (So great he clearly managed to contribute to a paper after his death. Take that Ali!)

Posted by: Edo on May 4, 2006 at 7:10 PM | PERMALINK

"paul" above questions the validity of using third down plays to predict fourth down tries.

It's precisely the question that occurred to me. It is, however, addressed in the paper--see Section IV, starting on page 20 for his discussion.

Kevin, if you want to blog occasionally about the NFL, I'd gladly take you over Easterbrook, who descended into unreadable self-parody a while back. Anyway, Easterbrook's TMQ did have a good year or two, at the beginning.


Posted by: pfd on May 4, 2006 at 7:18 PM | PERMALINK

I quit watching football after reading the book by investigative reporter Dan Moldea, titled:

"Interference: How Organized Crime Influences Professional Football"

CHAPTER ONE
.

Posted by: VJ on May 4, 2006 at 7:20 PM | PERMALINK

I agree with Paul above, the real world implications change under the stress of an actual 4th down situation, especially when most coaches will be playing from a smaller playbook (thereby actually tipping the advantage to the defense). If it were possible to guarantee that a player's play would not be affected by the "all or nothing" nature of a 4th down and that the coaches would treat the down like any other, this would work. But in real life, I don't see it. Everybody tightens up.

And I think it was actually Vince Young's fault that USC lost the Rose Bowl.

Posted by: Doug-E-Fresh on May 4, 2006 at 7:24 PM | PERMALINK

Penn State under 79 yr old JoPA was perfect on 4th down conversions last season, going 11-1 for their efforts. You don't have to sell me on this.

Posted by: Bill From PA on May 4, 2006 at 7:24 PM | PERMALINK

In the Colts-Steelers playoff game last January, Weren't you excited when the Colts went for it on fourth down in their own end of the field? Regardless of the outcome, weren't you glad that they made it when quarterback Manning overruled Coach Dungy? The Colts did it again, and later the Steelers also did it. three first downs on fourth and long in unusual field position, and in the playoffs.

Posted by: republicrat on May 4, 2006 at 7:34 PM | PERMALINK

"A team on its own 10 yard line should go for it if it has two and one-half yards to go? Come on."

The reason why this is the case is because the 4th-and-punt from there gives opposing team such a favorable postion that you would be better off taking that gamble, as crazy as it sounds.

This result I believe is based on empirical evidence, so it's not some crazy econ theory that's coming up with that number.

Posted by: clone12 on May 4, 2006 at 7:38 PM | PERMALINK

off topic, but here is a good source on biofuels:

http://www.technologyreview.com/read_article.aspx?id=16765&ch=biztech

Also, follow the link to desalinition in Spain.

Why wait until KD starts moaning about fuel again? Follow the story everyday, or at least every month, at MIT's Technology Review! At the journal Science (check out the item on semiconducting single-walled carbon nanotubes on P. 554 of Vol 312, April 28, 2006; payment required if you want more than the abstract, unless you join AAAS), and at American Scientist (also some proprietary information, unless you join Sigma Zi.)

Posted by: republicrat on May 4, 2006 at 7:40 PM | PERMALINK

I think the data is bad, since almost every football coach usually punts or kicks the FG there is no relevant corresponding data for going for it on fourth down. The data for going for it on fourth down are all anamolies.

It reminds me of the baseball analysis that says you should not bunt a runner into scoring position. From what I read, when this strategy was tried it failed.

Posted by: ftw on May 4, 2006 at 7:54 PM | PERMALINK

I think if LenDale's family had only had a chance to live rent free in their own palatial house in San Deigo, USC would have won.

As it is, of course, Kevin, not only did the Trojans lose the Rose Bowl. They're going to have to pay back their slice of revenue from it. Oh yeah, and they're going to have to give up their Pac 10 title and all eleven wins from the season. Sorry to hear your Trojans went 0-12. Tough year.

But next year, by all means I hope they go for it on fourth and five on their own 15. That will be a great move for everyone else in the conference looking for a little payback.

Seriously, though, best post ever.

Posted by: kidkostar on May 4, 2006 at 7:55 PM | PERMALINK

Studies such as this also tend to make the assumption that there is a typical defense that a typical offense is running the play at. There isn't. Going for the first down against the best defense in the league carries different probabilities than going for the first down against the worst defense in the league, and the relative quality of the offense changes the probabilities as well.

Also, the score and time remaining have an impact on what choice maximizes the odds of winning, particularly depending on the type of players a team has. Say the offense for team A excels at running the ball, but is suspect in pass protection, and has a below average red zone, and run, defense, but forces an above average number of interceptions (fumble recoveries tend to be random events). Say team B runs the ball very well, and has a ferocious pass rush, as well as a very good defense overall.

With this match-up, it would be extremely unwise for team A to be aggressive in going for it in the shadow of their goal post early in the game, as the study recommends. The best chance of defeating a team like team B is for A to deny them an early lead, which tends to allow B to turn loose their pass rush against A's suspect pass protection, and then allows B to control the clock with their running game while in the lead. Conversely, team A's defense is not good in the red zone, but if the opposing team must put together a long drive with some passing, team A has an above average chance of getting some interceptions. Team A should not give Team B early opportunities to grab the lead without having to drive the ball down the field.

Having said all that, football coaches are way too conservative, particularly when it s fourth down, and the ball is somewhre between (approximately) the opponent's 33 and mid-field.

Posted by: Will Allen on May 4, 2006 at 7:58 PM | PERMALINK

more on biofuels:

http://www.technologyreview.com/read_article.aspx?id=16407

Principally about commercial investors.

OK, now back to your game.

Posted by: republicrat on May 4, 2006 at 8:08 PM | PERMALINK

But since I don't care at ALL about college football, how's it stack up in the pros?

Posted by: MNPundit on May 4, 2006 at 8:08 PM | PERMALINK

Oops, sorry. I just assumed that you were talking about college because thats what you always natter on about when it comes to football.

Um, go vikings? Blech, Twins rock.

Posted by: MNPundit on May 4, 2006 at 8:13 PM | PERMALINK

Yeah, but these curves only deal with winning or losing, not with the real issues, like the likelihood of keeping your job as coach.

To calculate that CYA curve, you'd have to subtract the amount of grief you'd get if things go wrong from the amount of credit you'd get if things go right.

It's all negative on fourth down, even with inches to go at your opponents 40.

Posted by: frankly0 on May 4, 2006 at 8:22 PM | PERMALINK

To understand the behavior of coaches one needs to understand not just their analysis of the game situation, but their analysis of their own life situation. Coaches are rarely fired for not running some medium play that could have had positive results. But they are frequently fired for running medium plays that actually have negative results

People are far more frequently punished for sins of commission not sins of omission. Suppose a coach goes for it in a 4th and 5 situation on the 50 yard line in the second quarter, his team fails to make the first down, and the opponent scores a touchdown on the ensuing possession and goes on to win the game by a narrow margin. The coach will be hammered by media and fans. Now suppose the coach fails to go for it in the same situation, the opponent does not score on the next possesion, but later scores a fourth quarter touchdown to win the game by the identical margin. No obvious coacing error.

One might say this is irrational behavior, but on the other hand a certain sense of justice might be involved. People are reluctant to punish on the basis of missed consequences of omitted actions, even if the hypothesized consequences are in fact highly probable.

Posted by: Dan Kervick on May 4, 2006 at 8:22 PM | PERMALINK

Who gives a crap.

All sports comes down to one thing:
Who can get away with doing steroids.

Posted by: Osama_Been_Forgotten on May 4, 2006 at 8:27 PM | PERMALINK

obf,

if we don't give a crap about armchair coaching, then the terrorists have won.

Posted by: mudwall jackson on May 4, 2006 at 8:39 PM | PERMALINK

Just can't let it go, can you, Kev? :-)

Posted by: grape_crush on May 4, 2006 at 8:45 PM | PERMALINK

I can't wait for his paper on pulling the starting pitcher in baseball.

Posted by: toast on May 4, 2006 at 8:51 PM | PERMALINK

Growing up, the football coaches I was around didn't strike me as the sort that would glean knowledge from a dissertation.
Otherwise, how could you explain this: Team scores touchdown, misses extra point. Team scores another touchdown, then goes for two.
Why is that poor strategy? Consider alternative B. Team scores two field goals, then a touchdown. Under alt B, the coach almost always goes for one. But in both cases, the team has 12 points before deciding whether to go for two. So in the coach's head, HOW the team got 12 points affects whether to go for two.

Posted by: Ron Zealot on May 4, 2006 at 9:18 PM | PERMALINK

The Steelers won the Super Bowl!

Smiling Hines Ward was MVP!

Undrafted Willy Parker set the record for longest touchdown run in a Super Bowl!

Wide receiver Antwaan Randle El threw a touchdown!

Big Ben is the youngest quarterback ever to win the Super Bowl!

Just sayin'...

Posted by: putnam on May 4, 2006 at 9:30 PM | PERMALINK

Parts of this result are new. However, the conclusion that you should always go for it on 4th down needing 4 yards or less for a 1st down was pointed out quite a while ago in "The Hidden Game of Football." http://www.amazon.com/gp/product/1894963237/qid=1146792168/sr=1-1/ref=sr_1_1/104-4375619-0518323?s=books&v=glance&n=283155

Pro and c oollage coaches ignored that book. I suppose they'll ignore this one, too.

Posted by: David on May 4, 2006 at 9:32 PM | PERMALINK

twas quite a triumph, twasn't it putnam? they're still walkin on air in the 'burgh

Posted by: mudwall jackson on May 4, 2006 at 9:35 PM | PERMALINK

"In other words, Pete Carroll is a smart guy. It was LenDale White's fault that USC lost the Rose Bowl."

No, Carroll had a month to get his team prepared for the biggest game of the decade and he didn't have a formation for this short yardage situation where the obvious thing to do is to put LenDale White and Reggie Bush in the backfield at the same time so the defense can't concentrate on White.

Further, Carroll failed to come up with anyway to stop Vince Young. He didn't have any body on defense quick enough to tackle Young. However, he _did_ have somebody on offense who had the physical skills to sack Young on that last drive by Texas. What Carroll should have done during that month off is have Reggie White work out on defense, to play the spy position following Vince Young around. A single successful blitz on that last drive by Texas would have won USC the national championship.

Posted by: Steve Sailer on May 4, 2006 at 9:36 PM | PERMALINK

Nobody's mentioned the graph's spike at the 35 yard line. And they should, regardless of how they feel about the overall argument.

Think about when a pro team has the ball on its opponent's 35 yard line. If it punts, the punt's probably going into the end zone. The other team gets the ball - a mere 15 yards away. Might as well go for it.

Similarly with the FG attempt. The ball will be spotted at the 42 or 43 yard line, which is where the other team will take over if you miss. Great field position for starting a drive. But that means the kick is 52-53 yards, which is a pretty long-odds kick for all but a few NFL placekickers. So a FG try means a slim chance of gaining 3 points, and an overwhelming likelihood of losing both the ball and 7-8 yards of field position. That pretty much sucks too, so again, you might as well go for it - even if you need a lot of yards for the first down.

I've been calling the zone from about the opponent's 32 out to the 38 or so the 'dead zone' for about 25 years now, for this reason. This zone was created in the 1970s when the goalposts got moved back to the back of the end zone. (Up until then, they'd been goalposts, with the crossbar suspended over the goal line. It was still there in January 1973 - the Redskins lost an early TD in Super Bowl VII on account of that crossbar.)

More coaches have been noticing this in recent years, and going for it in that part of the field, but usually with

Posted by: RT on May 4, 2006 at 9:43 PM | PERMALINK

So the strategy of football coaches, who as a rule, being manly men in a fiercely competitive meritocracy must be conservative, is criticized by, of all things, a UC Berkeley professor? And where was this first reported, the New York Times? I know more in my gut about football strategy than some namby-pamby numbers-obsessed left coast egghead.

Anyway, that aside, I have thought that coaches are too risk averse on 4th down, although the numbers of the study seem excessive at first glance. Some of the issues raised by the other commenters, like Paul's comment as that the average yardage gains of 3rd down plays is used, is a real, if unavoidable, flaw. It's not just the pressure of 4th down, it's the fact that it's a different situation, with different plays and different defensive strategies. You don't need to get any set amount of yardage on 3rd down. There may be ways to control for this but I doubt it.

A few more questions about the graph. Why a discontinuity at 33 yards? Why wouldn't this be more rounded? It's not like 50 yards (or any other distance) presents some fundamental limit on field goal attempts, even if it is the region in which the success drops fairly sharply. To the right of that the wave makes sense I think, at 20 yards a field goal is fairly easy but a TD is still hard, but at 10 the field goal isn't much easier, yet the TD is much easier. But why the waves to the left? Why go for it at the 30 more than your 40? Do that many punts go 60 yards for a touchback (the only reason I can think of)? And there's still another wave to the left of that.

My guess is the optimal numbers (for the hypothetically average conventional offense against the average conventional defense) are less than the study finds, but more than coaches go for. I guess I'm a centrist Footballicrat.

Posted by: ChiSox Fan in LA on May 4, 2006 at 9:59 PM | PERMALINK

Even if you accept that Carroll should have gone for it in a situation where keeping his opponents from scoring was much more important than having his own team score, the play call was flawed. The formation screamed power run up the middle (especially with the lack of Bush...seriously, at least make it look like you might use him), Texas knew it, and White never had a chance.

Posted by: Viserys on May 4, 2006 at 10:00 PM | PERMALINK

This was one of my biggest complaints about Mike Sherman, as well. In the infamous (to Packers fans) playoff game against the Eagles a few years back, when the Eagles converted on 4th and 26, the less remembered event is Sherman choosing not to go for it on 4th and 1 somewhere around the Eagles' 40, despite the fact that a 1st down would have ended the game, and that our offensive line had abused the Eagles' line all day. The punt got a good return and I think we only got 14 net yards. Long story short, I'm still bitter, and I'm glad Sherman's gone.

Posted by: ChiSox Fan in LA on May 4, 2006 at 10:06 PM | PERMALINK

"Otherwise, how could you explain this: Team scores touchdown, misses extra point. Team scores another touchdown, then goes for two.
Why is that poor strategy? Consider alternative B. Team scores two field goals, then a touchdown. Under alt B, the coach almost always goes for one. But in both cases, the team has 12 points before deciding whether to go for two. So in the coach's head, HOW the team got 12 points affects whether to go for two."

Well you *could* explain it this way:
- my kicker just missed a PAT, why the hell should I trust him to try this one?

versus;
- my kicker''s already made two FG's, why wouldn't I let him kick a PAT?

But i think you're right for the most part.

Oh, and Kevin, if you're reading this far into the thread....when USC has to vacate that one national championship they managed to win with ineligible players will it go to Oklahoma or Auburn? There's a sports topic worth blogging about!

Posted by: chaboard on May 4, 2006 at 10:20 PM | PERMALINK

How does he factor in the score of the game?

Posted by: mm on May 4, 2006 at 11:34 PM | PERMALINK

So did this include variables for injuries, wind, rain, kickers ability, field types, field conditions?
Drunk Fans? A hottie cheerleader?
A half Time Nipple Ring throwing star? =)

Posted by: Mach Tuck on May 4, 2006 at 11:40 PM | PERMALINK

They're breaking out the ice skates in hell... I agree with Steve Sailer completely.

At the risk of sounding suspiciously like a Bush apologist, I do think there are occasions where a coach needs to throw away the slide rule and go with his gut. Statistical models tend to break down in marginal cases, and they can't account for the momentum shift that a defensive stop and an easy score can give the other team.

No matter what the numbers say, you just don't go for it inside your own 30 unless you're losing in the last few minutes. The rewards are uncertain and the downside is a guaranteed 3-7 points for the other team.

Posted by: ajl on May 4, 2006 at 11:43 PM | PERMALINK

I think coaches should go for it more often on their side of the 50 when they need a score, but this theory will never be tested in the NFL because a few failed conversions in traditional punting downs will end a coaching job quick. Coaches alone get the blame if they go for it and fail in a situation that's not desperate. Everybody gets the blame if the team fails to get back into scoring position after punting in a typical punting situation.

And should you really ALWAYS go for it on 4th and 4? I don't think anyone would have enough confidence in this analysis to go for it in the first half of tie game with 4th and 4 at your own 5 yard line. That's just nuts.

Going for it even when it's 4th and 4 at the 50 in a tie game is a Tecmo Bowl move (or a Pete Carroll move).

By the way, Pete Carroll is a terrible coach. Put him in a situation where the talent on his team is not 5X better than the talent on the other team, and you see right away that he's crap.

Bill Belichek beats top teams with better talent than his at a scary clip. Just ask folks in Indianapolis. Or Pittsburgh.

Pete Carroll isn't fit to wash Belichek's jock.

Posted by: Sean on May 5, 2006 at 12:13 AM | PERMALINK

Yes, almost certainly the study is underrating the psychological effects of momentum. (If you don't believe in momentum on the football field, we probably shouldn't discuss football, certainly not in real life where I could hit you with something handy...)

Basically, not even a championship team scores on every drive, or even most drives really. Punting it away isn't a big deal, it's going to happen over and over. A punt is a good thing for a defense, but not on the order of forcing a turnover.

But when you go for it on fourth down (especially if it's not short yardage!), you're creating a clutch play. Not only that, but you're creating one where your reward (first down) has a much smaller impact on your team than your risk (turnover!) will have on theirs.

Basically, if you go for it on fourth down, and you blow it, the other team gets a huge morale boost, especially if you did so in a situation that sets up an easy score for their offense. That's the sort of mistake that causes trouble all game long - once a team gets their fire lit, they're a lot more likely to go on a tear.

There's your answer. Most coaches won't risk the game on the chance that they'll put seven on the board.

Posted by: Avatar on May 5, 2006 at 12:26 AM | PERMALINK

Kevin, you are WRONG.

It is Reggie Bush who will cost the Toejams the 05' and '04 seasons plus their single one-peat victory. Mark Sanchez will cost the Toejams the '06 season and beyond.

ALL YOUR VICTORYS ARE BELONG TO US.

Posted by: GastonGreen on May 5, 2006 at 12:26 AM | PERMALINK

Yeah, but under what conditions should a coach order the running back to lateral to his imaginery friend.

Posted by: toast on May 5, 2006 at 1:12 AM | PERMALINK

Yes, almost certainly the study is underrating the psychological effects of momentum.

I think, though, that a very good proportion of that momentum is actually built on the notion that going for it on fourth down is something extraordinary, rather than common and expected. If it were done routinely, I doubt that either team would react anything like the same way they do now.

Posted by: frankly0 on May 5, 2006 at 1:25 AM | PERMALINK

I guess I'm a centrist Footballicrat.

Brilliant.

Posted by: craigie on May 5, 2006 at 1:35 AM | PERMALINK

NO dissing Lendale White!

I'm a long time Titans fan. We're twitchy.

Bieng a politics junkie and a football junkie is tough..and then you find something like this.

Speaking of field goals, what will the Pats be like without Adam Vinatieri ? Y'all discuss - that is if anyone ever looks at this thread, again.

Posted by: T4TN on May 5, 2006 at 3:04 AM | PERMALINK

If it were done routinely, I doubt that either team would react anything like the same way they do now.

I dunno. Fumbles and interceptions are farily "routine," and yet the team gaining possession of the ball generally sees them as stupendous events; those kinds of turnovers certainly have plenty of power to change/magnify momentum.

By the same token, I can't see how turning the ball over on downs, no matter how commonplace, could fail to fire up the team that gains possession. (Okay, I suppose if the score isn't remotely close....)

Posted by: Rieux on May 5, 2006 at 3:36 AM | PERMALINK

As has been pointed out, going for it on 4th down deep in your own territory isn't necessarily a bad choice, because punting from deep in your own territory already makes an opponent's score a likely outcome. Punting from outside your own 25 might, in some ways, make more sense because you're going to push your opponents starting point (assuming an average return) beyond mid-field.

As for whoever wrote about the "lack" of wisdom in bunting a runner ahead, bunting is a poor strategy unless the batter is very bad or the opposing pitcher is very good. It might make sense to bunt in a 1-0 or 1-1 game, but it makes little sense to bunt in a 6-5 or 6-6 game, as you're just surrendering an out. In fact, teams that bunt relatively infrequently -- hoping instead to walk, put the ball in play, or hit a home run -- have been pretty successful.

One more point, no matter how long it may take to challenge the conventional wisdom, that "wisdom" can change incredibly quickly in the face of a successful challenge. I think I have this right...
Bangladesh (or Sri Lanka?) came up with a strategy of extremely aggressive batting in one-day cricket matches, with the aim of maximizing the number of runs rather than minimizing the number or wickets (the traditional cricket strategy)... because the match is only going to last one day! Now, almost everyone bats this way in one-day cricket.

P.S. the Pats are going to miss Vinatieri dearly.

Posted by: keith on May 5, 2006 at 5:31 AM | PERMALINK

I first ran into Mr. Romer's work in James Surowiecki's excellent book "The wisdom of Crowds." It has a whole section on 'group-think' and, besides this football example, considers mutual fund managers, NASA managers (vis--vis shutle disasters) et cetera.

All talk of Mr. Carroll aside, this quote from the book shows the author's keen perspicacity:

"[T]he one NFL coach who appears to have taken Romer's ideas seriously -- and perhaps even used them in games -- is New England Patriot's Bill Belichick...." Go Pats!

Posted by: jhm on May 5, 2006 at 7:02 AM | PERMALINK

Some of ya'll might have heard rumors that there's a horse race of some kind going on tomorrow in some little fly-over burg called Lexington, Kentucky...

Jeez.

Me, I'll be watching the Oaks this afternoon. Got my '84 (Swale) Derby glass, my 12 year old Bourbon, fresh mint...I'm set.

Run, baby, run.

Posted by: CFShep on May 5, 2006 at 7:44 AM | PERMALINK

OK, lets just take pro football.

What are the odds of converting 4th and 4. Or the odds, subsequently if lost, of the opposition making 10 yards to be in field goal range. How many games lost by # points or less.

The decision obviously depends on game position, time and score also. If giving the ball to the opposition in field goal range is high enough to lose the game . . .

I haven't played football, but there seem to be some pretty smart guys out there. They study odds. I would.

This strikes me as wrong!

Posted by: notthere on May 5, 2006 at 7:57 AM | PERMALINK

# = 3.
Most could figure that out, I guess.

Posted by: notthere on May 5, 2006 at 7:59 AM | PERMALINK

Dear Kevin: You may not see this comment, because I didn't see your entry until Friday AM East Coast time, but I want to raise a few points.
Prof. Romer's work is known to the coaching community. Bill Belichick read his initial paper the day he saw it referenced in the WSJ. NFL Flims has already produced and aired a piece in which the good professor airs his theory and various coaches give their opinions. They ranged from a big-time gambler like Bill Cowher to very cautious Marty Schottenheimer.
The coaching consensus boiled down to the following: When we get tenure like this guy has, we'll go for it on fourth down a lot more often.

Posted by: JMG on May 5, 2006 at 10:11 AM | PERMALINK

It was LenDale White's fault that USC lost the Rose Bowl.

Or rather, it was the Texas defense that won the Rose Bowl. Confusing to SC fans, I know, 'cause their team stopped playing defense after 2004.
.

Posted by: Grand Moff Texan on May 5, 2006 at 10:28 AM | PERMALINK

Steve Spurrier was going for it regularly before Pete Carroll

Posted by: rkf on May 5, 2006 at 11:06 AM | PERMALINK

Another vote for Texas defense...

...and Vince Young....

Posted by: Canid on May 5, 2006 at 12:08 PM | PERMALINK

Hope Texas fans enjoyed watching Vince Young in college. He won't be pretty to watch in the pros.

Akili Smith anyone?

Posted by: NSA Mole on May 5, 2006 at 12:17 PM | PERMALINK

"What are the odds of converting 4th and 4?"

How would you even measure this? And please don't say they keep stats on it. As teams aren't obligated to go for it on fourth down, coaches cherry pick the most ideal situations (e.g. go for it only when you're o-line is healthy, you're at home). What's the *real* conversion rate? What does Romer use?

Posted by: morairty on May 5, 2006 at 12:24 PM | PERMALINK

Michael Vick seems to be doing OK, NSA.
.

Posted by: Grand Moff Texan on May 5, 2006 at 12:47 PM | PERMALINK

How ya feelin' about rooting for
The University of Scared Coeds
Kevin,
all those women who file police rape reports year after year and amazingly the City Attorney does nothing.

but yeah, Carroll is a great coach who understood the 4th down thing well. It also helps that he recruits and retains the scum of the earth to play for him.

When your players tell the police "I own the police" after a bar fight and get away with it, you know they really do own the police.

Posted by: CalDem on May 5, 2006 at 4:20 PM | PERMALINK

Grand Moff,

Vince Young isn't Mike Vick. He's much slower and played in an even more remedial offense than Virginia Tech's (which is saying something). Vick still seems like he has a long ways to go 4 years later. I see Young in a similar situation.

The Texas defense was absolutely torched in that game. They didn't help at all.

Posted by: Double B on May 5, 2006 at 4:39 PM | PERMALINK

Back during the Steelers' amazing run (I'm a huge fan), one of the announcers said that half of all two-point conversions were made. Well, if that's the case, then the expected number of points a team would get going for a two-point converion is 0.5 * 2 = 1. However, the expected number of points for kicking the extra point is P[successful field goal] * 1 = P[successful field goal], which is slightly less than 1. In other words, going for two slighly increases a teams expected points. That's surprising considering how few teams elect to go for two.

Posted by: blank on May 5, 2006 at 4:52 PM | PERMALINK

I find it interesting that the line corresponding to yardage where 50% of teams actually go for it on 4th down roughly tracks the shape of the main curve, except with about four std. error's worth of conservativeness.

Posted by: Dennis on May 5, 2006 at 5:20 PM | PERMALINK

Just a quick comment on sacrifice bunts (bunts for a hit are obviously different):
Apart from pitchers, succeeding on a sacrifice bunt is never good--it always reduces your expected runs scored and your expected chance of winning in every situation possible.

However, attempting a sacrifice bunt can be good in certain situations for certain batters. In particular, it is beneficial to mix in a few sac bunt attempts to prevent the defense from playing the same way every time you come to bat. Even strong hitters should attempt to bunt once in a while early in the game.

Posted by: Andy on May 5, 2006 at 7:47 PM | PERMALINK

It seems obvious that teams should go for it when they are between midfield and the opponents' 35-yardline and don't have a long way to go for a first down, but I have a hard time believing the breakeven point is four yards or less when you're inside the 30. It also seems to me that you can't ignore the situation; if you're down 14 with 5:00 left, that's one thing, if you're up 12 that's quite another. And I absolutely agree; using 3rd down plays is a ridiculous mistake. To cite just the obvious, a coach may decide to run for it on 3rd and 3, because he knows he can always go for it on 4th down, but if he's facing 4th and 3 he's going to pass 95% of the time.

Posted by: Brainster on May 5, 2006 at 9:03 PM | PERMALINK

I don't know about you guys, but every football game I've been to, if there's less than 2 yards to a first down, there seems to be quite a cheer in the stands to go for it.
The fans are right!

Posted by: doug r on May 5, 2006 at 10:36 PM | PERMALINK

I don't agree with you about much, but I believe you're absolutely correct -- coaches are FAR too conservative on 4th down.

Posted by: mh on May 6, 2006 at 12:09 AM | PERMALINK

"Enough about politics. Let's talk about football."

Let's not.

Silly me, I thought this blog was about politics. Dude- I don't expect to see sports coverage on C-Span, or political talk shows on ESPN. So how come every third blog entry on this site lately is about football, hockey, baseball, etc.?

Stick to the subject please, and attend to your hobbies elsewhere. What's next, the latest news on home metal detecting, or the newest needlepoint patterns?

Yeah, yeah, free speech and all that- except, as far as I can tell, this isn't YOUR personal blog, it's the Washington Monthly's politics blog. And it's a darned good one, I read it several times a week. For the politics, not the stupid sports stats...

If your employers would like politics junkies like me to keep coming here, enough with the irrelevant sports blather- I hear enough of that crap at work.

Posted by: RobW on May 7, 2006 at 12:51 AM | PERMALINK




 

 

Read Jonathan Rowe remembrance and articles
Email Newsletter icon, E-mail Newsletter icon, Email List icon, E-mail List icon Sign up for Free News & Updates

Advertise in WM



buy from Amazon and
support the Monthly