The coach who never punts

In Little Rock, Arkansas, the Pulaski Academy Bruins play the game of football differently than you’ve ever seen before.

They don’t punt.

They onside kick every time.

And they always go for two.

Kevin Kelley, the architect of the system, studied years and years worth of data and implemented the system with absolute success. He’s won five state titles and has one of the best offenses in the entire country.

There is, sadly, a noisy video at the link, though it is easy to turn off.  Whether you agree with this strategy or not, one of my core views is that we do not have enough experimentation of this kind.  And I’m not just talking about football.

For the pointer I thank Peter Bach-y-Rita.


This is basically what Tim Hartford is saying in his latest book.

Experimentation of this kind is needed with educating kids. It is the reason that I always favor states over feds...counties over states...townships over counties....families/friends over towns

Individuals over families?

Genes over individuals?

"He’s won five state titles and has one of the best offenses in the entire country."

Is the great offense caused by the strategy or what enables the strategy to work? I imagine a high school team that outclasses its opponents will have much more success with this because they convert the plays more often.

That said, there is lots of solid evidence that teams are punting way too often at the NFL level. Possession is undervalued relative to field position.

He adopted the approach after losing to bigger schools too often. After he switched to the no-punt approach, his school's win rate went up significantly.

In the NFL individual Jon security concerns probably dominate. Being the perceived idiot who made an optimal but high risk decision that ultimately failed is bad for a head coaches employment prospects. This may not be true for all organizations and times are surely changing.

On the high school level all you say is true. This strategy increases the already high win probability of the favorite.

For huge underdogs other it might make sense too because increased variance increases their slim chances of winning.

For small underdogs my intuition says that this strategy would be bad. It doesn't seem doesn't make sense for them to go for two if their inferior running game only gets there one third of the time.

As I general rule that flows from pretty fundamental probability concepts, underdogs should make as few risky attempts as possible but should make the sakes as high as possible on the attempts that they do make, while "overdogs" should break their risks into as many low stakes trials as possible.

So, for example, a bank which has the odds in its favor, wants to handle every customer dispute in a tiny, individual arbitration limited to one customer, while a customer who had the odds against him wants to be part of a single class action on behalf of every customer who suffered the same harm.

In generally, taking high risks is an underdog's strategy, which is why most teams only resort to onside kicks, two point conversions and don't punt on the fourth down only when they are losing badly. But, if everybody is systemically overestimating the risks associated with this moves (especially when the risk reduced because your team has much more experience with this unusual tactics than any other team in the league and practices these plays relentlessly), then you are simply arbitraging between apparent risk and actual post-special play practice risk.

there is also a strangeness component to consider: If someone started running say the single wing (hey, job for tebow) they probably would have more success than the formation should yield. similarly you probably have a much better onsides kick team and ability to onsides kick if you do it every time (and thus every time in practice) instead of it being the thing you squeeze in at the end of practice

This is how Navy is able to get bowl wins. Their offense is so foreign to modern football that requires a completely different defensive game plan - which is very difficult to install in just one week.

It's actually the other way around because you get three to four weeks to install a defense before a bowl game. The original architect of Navy's offense Paul Johnson went to Georgia Tech were he had pretty good regular season success but struggled in bowl games because the teams had time to adapt.

actually this is even less interesting than bob dylan.

This story has been sitting at 538 for close to a year.

I read about this guy nearly 10 years ago. At least 5 years ago he was covered in a couple of major sports analytics books.

You both are accurate.

Id add though that IMO, the real laboratory for this stuff in Madden. I know, I know, it's a video game but if you play enough to notice, the simulations are becoming so good it really is a lab for strategy.

Been playing in high level tournaments for years now and in highly offensive games... This is what many good players do. It's psychological warfare while also a statistically advantageous strategy when both teams can score a lot of points.

Yeah, it also seems to be very good practice for clock management.

Obama has ruined sports with all this big data stuff. His 2008 campaign popularised big data and this is what has resulted.

This data stuff was happening in sports before politics. It was being pioneered in baseball in the 90s.

That may be strictly true but Obama really popularised these methods with his 2008 campaign. Its spread to sports is indicitive of a sort of sickness that has befallen this country. Obama and the liberal elites have a deep resentment for the drama and spirit of American sports and wants to see it reduced to a computer algorithm.

I think you win the award for stupidest commenter. Sports have grown increasingly data-driven every year since time immemorial. Look at the development of the shift in the 40s. Look at pitch counts. Look at sabermetrics. Look at Bud Grant. Look at Hack-a-Shaq. Look at Moneyball. Look at the MIT Sloan Sports Analytics Conference.

Really, there are two problems with your comment. 1. The ridiculous a-historical buffoonery of it. 2. The dumb, dumb, dumb idea that data "ruins" sports, a perspective only held by people who aren't actually sports fans. When you try to separate sports from effective practice, from the beauty of strategy development, and from the identification of greatness and outliers, and instead make it about "drama" and "spirit" you're outing yourself as a non-fan who likes to watch sports movies. Go see a fucking play if you want "drama" and "spirit" untethered from reality, and leave sports to people who care about the art of the game.

It's parody, but he got you

JAMRC strikes again. But I love it, OJ needs a takedown.

Does it say more about me or the world that I fell for it?

Definitely negative points for both.

Please, Obama hasn't popularized "all this big data stuff." Winning makes things popular. If analytics didn't work, no one would use them. Winning popularized analytics, because people like to win.

Is there nothing that some people won't blame on Obama? Good lord.

This is almost as crazy as stuff the regular Jim would say.

Since you say you're not just talking about football let's talk baseball. The dumbest play in baseball is the sacrifice bunt. You only get three outs and you're going to give up one of them to move one runner up one base?


Sacrificing with runners on first and second, no outs, can have a positive expected value.

As can swinging away. It's the relative value that counts.

Positive relative to the baseline of swinging away

Right, my point was that bunting there is not obviously wrong.

It ultimately depends on how strong the hitter is, how likely a double play is while swinging away (which is partly the hitter's tendencies and base running speed), how late the game is, and whether you are in a high or low scoring environment.

The sac bunt is usually a dumb play that lowers your team's probability of winning the game. But there are occasional situations where it is a good percentage play.

I'm not familiar with the analysis around two on nobody out. However, one time it can be good is during a tie game at home, where you need exactly one run to win.

The best data suggests that sacrifice bunt slightly increases your chances of scoring one run, and torpedoes your chance of scoring several runs. Sac bunts are also rare enough, relative to every other play in baseball, that in the long run they don't significantly move the needle either way.

After watching last night's game five, I wonder how de-emphasizing sacrifice butning means a lack of practice for most players affects the numbers (both sacrifice bunt attempts ended badly due to poor bunting).

Not just relative value, also variance.

It made a lot more sense fifteen years ago when a lot of pitchers basically couldn't even be counted on to put the ball in play but could be taught to bunt. Now that pitchers for some reason have gotten better at hitting it makes less sense.

Trump, swinging for the fences
Clinton, clinging to the status quo


Too much regulation stifling risk taking, leading to sub-optimal economic outcomes.

Paul ("Dr. Z") Zimmerman discussed a similar analysis for the pros way back in his 1984 book _The New Thinking Man's Guide to Pro Football_ and concluded that punting did not, on balance, yield better field position overall.

When Bill Parcells was coaching the NY Giants, he adopted a limited version of this thinking: when the Giants had a 4th down anywhere between (roughly) the 25 and 35 yard lines, they would often go for the first down rather than kicking a low-odds long (43-53 yard) field goal or trying for a "coffin corner" punt that often went into the end zone for a touchback (net yardage gain or 5-15 yards).

The first couple of times I figured it for a stunt play, but after a while it became part of the regular repertoire and I started referring to that part of the field as the "Parcells Zone."

It's much better to take that sort of approach as a permanent feature once you get to the 50 yard line to the 25; if the coach conceives of his strategy as a 4 down opportunity, it changes the play-calling. A 2nd and 10 with 3 downs to go gives more freedom, as does a 3rd down with another down to use.

At USC Pete Carroll routinely went for it on 4th down anywhere inside the 50-yard line (at the Coliseum, on 4th down the crowd would start cheering "big balls Pete.")

It seemed to work out well, Though i don't have any statistical proof.

"Academy" means it's a private school playing other private schools, with football teams made up of mostly white boys. Pulaski has a 65-player roster, with only a handful of black players.,ar)/football/roster.htm Black players at such schools are recruited and provided with scholarships. Private schools with fewer resources can't compete, resulting in lopsided scores and records. The most lopsided football game in college football was played 100 years ago, Georgia Tech vs. Cumberland. The final score was 222-0. Tech never threw a pass in the game but never let up. The reputation of the head coach at Tech was forever tarnished for running up the score. The coach? John Heisman.

This makes sense since punting well is actually really, really hard. You're placing the ball by kicking it. The best punters are usually ex-soccer players and in high school they are busy playing soccer. Punters are often among the most athletic players on a team. At the college and pro levels, the quality of punting improves dramatically over high school.

Sorry, Rayward, this wasn't meant to be a reply to your comment.

"Punters are often among the most athletic players on a team."

You're from Brazil, Thiago. How athletic are most soccer players? Very, right? Punters are futbol players who play football.

The aren't at all. Ted Craig just made this laughable point up.

rayward, you really need to watch this video. The back story behind 222-0 is fascinating

This is actually an old strategy that has been used regularly by everyone who plays Madden football on PlayStation .

Interestingly, there's an article today on how soccer video games are actually influencing real soccer:

1) Gregg Easterbrook (Tuesday Morning Quarterback) has written about this guy, too.

2) I've read the Romer paper (it's in JPE) and he does not say you should never punt.

3) He also doesn't allow the players to field punts (presumably because the risk of a turnover isn't worth the field possession gain).

4) The onside kicking strategy probably wouldn't work well outside of high school, because better athletes would have less trouble holding on to the ball.

The NCAA changed onside kick rules in 2012 in the name of player safety. However, it also made it significantly more difficult for the kicking team to recover the ball. The "old" onside kick strategy was to kick the ball hard into the ground, resulting in a very high bounce that would give the kicking team time to get in position for a recovery. Now, between the first and potential second bounce, the receiving team has the ability to call a fair catch - meaning the kicking team has to provide the returner with an unimpeded opportunity to catch the ball.

The kicking team is also not allowed to block any players on the receiving team until the kicking team is eligible to recover the ball, and illegal blocks are now reviewable (new for 2015). This rule might be the same in high school, but with fewer and less experienced referees, the sudden change in which infractions to look for, and a lack of instant replay make it much easier to field an onside kick in high school than in college outside the "better athletes."

Also, I've seen data indicating that even in the NFL and college, onside kicks are successful about 60% of the time when they come at unexpected times, compared to about 10% when they are expected.

Also, in HS you usually kick from further up than in college/NFL, and the kickers still can't regularly kick touchbacks. The risk of a standard kickoff is higher and the penalty for not recovering an onside kick is lower.

I still can't believe how many times I've seen HS coaches kick deep from the opponent's 45 yard line (40-yard line normally plus 15 yard penalty) in competitive games that weren't defensive struggles.

Never punting can't be the best equilibrium strategy.

No. But say you run the numbers and decide it makes sense to punt roughly 10-15% of the time. Does it make sense to use up a roster spot(s) on a punter, develop a devoted punt team, and spend valuable practice time practicing punting, all for those 10-15% of fourth downs? Probably not. May make more sense to just occasionally ask your QB to quick kick instead.

4th and 10, up 13 - 9 with 70 seconds to go on your own 10 yard line.

Bottom line is this: Would your opponent rather you have 1st & 10 at your own 20 yard line, trailing by 4, with one minute remaining; or they have 1st & 10 at mid-field, trailing by 4, with one minute remaining? Then do the opposite of what your opponent wants.

Option 3 (turnover on downs): Your opponent has the ball first and 10 and your 10 yard line.

P(Option 3) > 0%.

I understand math is very difficult for some people.

I don't care if it does win games, it is still part of a (((globalist))) conspiracy to take our plays.

Trump never punts.

"There is, sadly, a noisy video at the link, though it is easy to turn off. Whether you agree with this strategy or not, one of my core views is that we do not have enough experimentation of this kind. And I’m not just talking about football."

Strongly disagree; those noisy videos should all go die!

Comments for this post are closed