Going for it on fourth down - lessons from football analytics


On the first drive of Monday night’s Redskins-Eagles game, first-year Eagles head coach Chip Kelly continued a decision-making tendency that became famous during his tenure at the University of Oregon.

He went for it on fourth-and-1. 

This time, he was on the Washington 21-yard line, a spot on the field where most coaches would send out the field goal unit.  Instead, Kelly sent running back LeSean McCoy up the middle of a tired and surprised Redskins front line for four yards to keep the drive alive. 

Just a renegade coach who keeps getting lucky by defying the odds?  Hardly, says a growing chorus of football analytics researchers armed with brute force, artificial intelligence systems. 

In a “Moneyball for the gridiron” revolution, advanced analytics are changing the game at all levels.

At the high school level, where innovation and risk-taking should thrive, there are Kelly protégés who prefer to study the numbers and use statistics to make decisions rather than use conventional wisdom. 

Image of Kevin Kelley, courtesy of Pulaski AcademyKevin Kelley, head coach of Pulaski Academy in Little Rock, Ark., has only one game situation in his playbook when he will punt on fourth down – if his team is winning convincingly and he doesn’t want to run up the score. 

Otherwise, the numbers clearly tell him that punting reduces your chance of winning a game by giving away possession.  In the same way, getting the ball back after a score is just as important. Hence, he will regularly call an onside kick.

“Everyone says football is a game of field position, but it’s not,” Kevin Kelley told ESPN. “It’s a game of scoring points, which only happens when you possess the ball. If you’re not obsessed with field position, then you don’t punt. You onside a lot. You don’t even try to return punts or to block punts, because getting the ball back is far more important than risking a muff or a roughing-the-kicker flag.”

Making bold decisions during a game requires analyzing probabilities and coming to logical conclusions prior to kickoff. 

“In high school, the average opponent’s start after a regular kickoff is the 33-yard line. After a failed onside, it is the 47,” Kevin Kelly said. “So you are risking 14 yards of field position in return for a good chance of a turnover. If there were a blitz action that would risk a 14-yard gain by the offense versus a turnover for your defense, you’d call it constantly. That is the equation for an onside, yet the play is hardly ever called.”

Here’s Kevin Kelley explaining his logic:

Does the risk yield rewards?  Since 2005, when Kevin Kelley started his unorthodox analysis, Pulaski has won two state championships, including an undefeated 2011 season. His program has the second most wins in Arkansas during the last 10 years.

Kelley credits a theory and some software algorithms for his game planning. 

A research paper from David Romer, professor of political economy at Cal-Berkeley, introduced the “lose the punter” concept by asking if organizations always make the right decision to maximize their gain, even if that means taking a calculated risk.

To test the theory, Chuck Bower, an astrophysicist (i.e. rocket scientist) at Indiana University, built a computer software system called Zeus.  Bower and his team took a decade of actual NFL plays, including down and distance, time on the clock, score and field position, and ran millions of simulations of different game decisions.  By tweaking the logic algorithm underlying the software, Bower got Zeus to predict which decisions provide the best chance of winning the game.

Bower explains Zeus in this video.

Will Zeus replace the human element of coaching?  Of course not, but just like sophisticated video editing systems and stats programs, Zeus could be another weapon in the press box.

“Zeus takes the relative output of the simulation and performs an objective analysis of statistical significance and skill sensitivity,” Bower told Science Daily. “It’s a valuable addition to a coaching staff’s tools and one that can provide that elusive edge over the competition.”

The bottom line for coaches is that winning games requires not only talent on the field but innovation and a search for the competitive edge, even when it may be unpopular.

Just ask Kevin Kelley how it felt that first season of change.

“You can just tell people are in the stands thinking, ‘You’re an idiot’ ” he said.  “Just because something’s always been done that way doesn’t mean it should continue to be done that way.”

Dan Peterson writes about sports science and skill development in youth sports, when he’s not watching the Badgers or Packers.  Visit him at Sports Are 80 Percent Mental and follow him on Twitter and Facebook.

Image of Kevin Kelley, courtesy of Pulaski Academy