Here's a little known fact about the 2010 Cowboys. The Cowboys attempted a fourth down conversion 22 times last year, the fifth highest total in the league. They succeeded 13 times for a conversion rate of 59%, ninth best in the league. In total, the Cowboys had a 'Go-for-it-rate' on fourth downs of 35%, the third highest value in the league.
In the 51 seasons the Cowboys have played football, do you know how often they've attempted more than 20 fourth down conversions in a season? Not a single time. The high number of fourth down conversion attempts could be nothing more than random chance, it could well be a result of some very specific down-and-distance situations, or it could be because the Cowboys are listening to one of their newer employees, Senior Analytics Manager Ken Kovash.
Ken Kovash joined the Cowboys in August 2010 with three years experience as the Director of Quality Assurance for the Mozilla Corporation (the Firefox people), an MBA in finance and marketing as well as a BA in economics. After the break, we look at some of the statistical thinking Kovash brings to the table as well as some other statistical topics he might help the Cowboys with.
In a paper published in September 2009, Ken Kovash and Steven Levitt (co-author of Freakonomics and SuperFreakonomics) looked at the run/pass ratio and playcalling predictability in the NFL. Levitt, professor of economics at the University of Chicago, wrote an article last week in which he said that Ken Kovash was hired by the Dallas Cowboys at least in part on the strength of that paper, so we'll look at its contents in a little more detail here.
In their study, "Professionals Do Not Play Minimax," Kovash and Levitt use the "minimax" theory to show that teams in the NFL
- Tend to pass the ball too little and to run it too often
- Are less likely to pass if the previous play was also a pass (and vice versa).
Minimax is a mixed strategy solution to a two-player, zero-sum game. The theory argues that a player, in this case an NFL team, should mix up its play choices in such a way that the chance of success between the choices is the same, thereby maximizing the potential gain: the probability of scoring a touchdown should be the same whether the team decides to pass the ball or run the ball. And just as importantly, the choice of whether to run or pass should not depend on the previous play.
For the NFL, their study used 125,000 plays from the 2001 to 2005 seasons. Using their own measure for the likelihood of scoring based on down, distance, field position etc,. they analyzed the change in a team’s expected points before and after a play. They found that a pass gains .55 yards more than a run on average, is 9 percentage points more likely to result in a first down, and leads to scores with a 3.8 percent probability while runs only have a 2.8 percent scoring probability. The authors went on to show that a team that passed more than the current league average would gain up to a point per game, or about half an additional win each year.
They also showed that a team that passed on its previous play is 10 percentage points less likely to pass on the next play. And a poor result on the previous play further increases the chance that a team will switch from pass to run (or vice versa) by 14.5 percentage points. Statisticians call this a "negative serial correlation." Fans usually call it predictable playcalling and other things that our site decorum does not allow me to write.
Kovash and Levitt estimate that if a defense could better anticipate the play based on the offensive predictability, it would allow 10.8 fewer yards per game, the equivalent of one point a game or half a win a year. Add the gains from calling more pass plays and, in theory, the effects could add up to one extra win per year.
More things the statheads are saying:
Mr. Kovash has his work cut out for him at Valley Ranch. Here are four selected statistical findings that football analysts have been preaching for years, yet have not been widely adopted by NFL franchises. Perhaps Mr. Kovash will gently nudge the Cowboys in the right direction on some of these.
NFL teams should go for more fourth down conversions. The statistical evidence for going for it on fourth down is overwhelming. Yet coaches remain conservative in their approach, perhaps preferring the safe option to the potential public scrutiny of a failed 4th down conversion. Think back to last year's Patriots vs Colts game in which Bill Bellichick decided to go for it on 4th - and the ensuing media firestorm when the attempt failed (Peter King: Bill Belichick made wrong call in Patriots-Colts game - - 4th Down Decision Dumbest Ever? - - Pats Lose on Belichick Stunner). Statistically, the better decision was to go for it, and by a large margin. But after the failed attempt Captain Hindsight had a field day.
The Harvard Sports Analysis Collective argues that perhaps coaches are more worried about making choices that get everyone upset than they do about making a choice that is mathematically correct.
There is no correlation between running the ball early and winning games: One of the most hotly debated topics in football has been debunked by generations of statisticians, yet continues to be a contentious topic among fans. Football Outsiders offer this perspective:
There are two reasons why nearly every beat writer and television analyst still repeats the tired oldschool mantra that "establishing the run" is the secret to winning football games. The first problem is confusing cause and effect. [...] In general, winning teams have a lot of carries because their running backs are running out the clock at the end of wins, not because they are running wild early in games.
The second problem is history. Most of the current crop of NFL analysts came of age or actually played the game during the 1970s. They believe that the run-heavy game of that decade is how football is meant to be, and today's pass-first game is an aberration.
Offenses run too often on first down: According to research by Brian Burke of Advanced NFL Stats (based on all plays from 2000 to 2008, but limited to the 1st and 3rd quarters when the score was within 10 points) 55 percent of first-and-10 runs between the 20-yard lines gained less than four yards. "The majority of runs on first down are actually setbacks," Burke writes.
In a companion article titled "Passing=Winning" Burke offers the following advice to NFL GMs:
If I were advising a general manager, I'd tell him to largely forget about the run. Get a RB who's good at picking up blitzes or catching the ball. Never draft a RB in the first few rounds, and whatever you do, don't waste precious cap space (or payroll budget) on him.
Get a quality QB at all costs. Assess your linemen on how well they pass block, and don't worry as much about their run blocking. Get lots of pass rushers on defense.
Got a LB that's a great run stopper but can't play coverage? Trade him to some sucker team that cares that they only give up 3.8 yards per carry rather than 4.2 yards per carry. That's how you build a perennial playoff contender.
Blitz at your own risk: In 2009, 10 of the top 15 quarterbacks as ranked by passer rating had a higher rating against the blitzed than against a standard defensive front. Only one quarterback had a passer rating more than 5 points below his total season average: Eli Manning had a 83.9 rating against the blitz, and a 93.1 rating overall. Unfortunately, ESPN doesn't publish the blitz splits anymore, so we don't have 2010 data. Regardless, the numbers suggest that dropping back on obvious passing downs may be a better option than blitzing.
The Cowboys coaches, players and staff are now entering the offseason where they will spend a good amount of their time looking for tendencies in their own and in their opponents games and how they can tweak those tendencies for their own benefit. Having a good statistical support in addition to watching game film can only improve the process.
Nobody is suggesting that the Cowboys adopt a statistician's manual to make their decisions during the game. But when it comes to making strategic decisions during games, many coaches remain staunchly conservative in their approach, for whatever reasons. Challenging conventional wisdom and looking at the stats more carefully and perhaps choosing a less predictable option during the game may help move the odds in the Cowboys' favor more often and may ultimately help them win more games.