Second Down and 10: Randomness vs. Balance
By Jason Lisk
Most fans would say you need to run the ball here, and would throw things when their team tries another pass and it falls incomplete. (You’ve got to establish the run!) This is why we shouldn’t listen to fans. Fans assume balance means that you run the ball after a pass. That should not be the goal of an offense. The goal should be unpredictability. If you’ve ever played blackjack, you know that your chances of winning any particular hand is just under 50%, but during the course of an evening, you may have multiple runs where you won or lost several hands in a row. While we know that the results overall should be close to 50%, the next hand’s result is independent of the previous result.
When humans try to randomize, though, they tend to do a poor job of it. It’s why we can use statistical applications like Benford’s Law to identify things like embezzlement or made up results of a coin flip sequence. We also see that NFL offensive coordinators aren’t really good at randomizing either, particularly on second down and 10. Doug Drinen made this observation after the 2006 season in a series of posts on second and 10. According to Drinen, in game situations where clock management shouldn’t be a factor, coaches ran the ball on 2nd and 10 following a pass 55.7% of the time, while they ran it on 2nd and 10 following a run 29.2%. Running after an incomplete pass was also less effective, suggesting that defensive teams and coaches are on to this lack of randomness: 41% of running plays on 2nd and 10 after a pass gained 2 yards or less, while only 10% gained a first down, compared to 33% and 15% respectively, for runs following a failed first down run.
Brian Burke of Advanced NFL Stats added to this last December. He looked at passing percentages on second down at various distance. At 2nd and 11-13 yards to go, teams passed 72% of the time. At 2nd and 7-9 yards, they passed 63% of the time, but on 2nd and 10, they passed only 48% of the time. Does that make sense? Not unless you consider that most teams get to those other distances with runs that either lose yardage or slightly gain yardage, and usually get to 2nd and 10 by throwing an incomplete pass. The problem is, running on 2nd and 10 is not generally successful. Using his EPA model, Burke reports that passes on 2nd and 10 are far more successful than runs on 2nd and 10, when compared to the pass/run payoff at other distances on 2nd down.
So, I decided to look at the 2009 season and see what teams showed the largest discrepancy in playcalling on 2nd and 10, depending on whether they threw an incomplete pass on first down. Here are the ten teams that showed the largest difference in pass/run percentage on 2nd and 10 (percentages are pass percentages following each type of first down play):
It’s an interesting mix of teams. We have 3 playoff teams from last year, and 5 teams who lost at least 11 games (averaged for group = 6.4 wins). Now, I would think that this list should have some bias toward good teams, since I didn’t remove plays in the fourth quarter when other considerations, such as running clock, may factor ahead of point optimization. Your initial reaction might be, of course, you should run it after an incompletion if you have JaMarcus Russell or Trent Edwards or a rookie Mark Sanchez as your quarterback. I am not suggesting that the Jets, with a rookie quarterback and good running game, or the Raiders, with Jamarcus Russell at quarterback, should pass as much as the Colts on 2nd and 10. Every team should have their own individual optimum ratio based on personnel. What I am suggesting, though, is that the Jets probably shouldn’t have as wide a discrepancy in play calling (only 16% passes after a first down incomplete versus 52% after first down no gain) as they did based on how they arrived at 2nd and 10. That’s all theory, though, and in my next installment of the epic 2nd and 10 series, I’ll try to look more closely at those teams to see if, despite the broader data that says running is less efficient after a first down incomplete, these coordinators knew what they were doing with their specific team circumstances. [photo via Getty]