T*steful Predictability: Exploring expectation’s impact on success

Ever since the Brian Schottenheimer hire, complaints have come in about how predictable his playcalling is. I am not equipped to discuss whether it is or not, but I was curious about whether predictability is actually a net negative for NFL offenses.

I get it, this sounds absurd; of course it is bad to be predictable in football. Predictability is why play action doesn’t work in obvious passing situations (shown by Ben Baldwin among others). However, play action works independently of previous rushing history (both volume and efficiency) because current NFL players have been coached to defend the run (which probably matters more at the high school and college levels) for a decade and a half by the time they reach the league. As passing becomes more and more prevalent at these levels, NFL defenders will have less and less of an instinct to stop the run first. At this point, rushing to set up play action may actually become a thing.

Play action works because there is an expectation (rush play) that is replaced by different action (pass play). If you lose the expectation of a rush play (which, I reiterate, is not due to previous rushing volume, efficacy, or Madden rating of the running back), then you also lose the effectiveness of play action.

With respect to Schotty (and OCs in general), the question boils down to this: is it better to be predictably unpredictable, or unpredictably predictable? Below is another convoluted example to explain where I am going with all this:

Vean McShay is calling plays for the fictional Los Angeles Rams (an NFL team in LA is laughable, I know). Vean is a GeNiUs whose playcalling is utterly unpredictable (probably scores more than three points in the Super Bowl, though). This means there is a 50/50 chance that any given play will be a run or a pass. No matter what the personnel or formation, any given play has a 50% chance of being a pass (or run). Also, let’s pretend that football only happens in a neutral game state on 1st down.

What is a defensive coordinator to do? Given that we know passing has a higher chance of generating successful plays, the smart thing to do is to expect the pass. What defensive coordinators do when they “expect the pass” is nuanced beyond my comprehension, but let’s say for argument’s sake that it boils down to moving players in and out of the box. Two or one players can be moved into or out of the box, or no players can be moved out of the box. This is just a proxy, do not get mad at me, film twitter.

And one more assumption: the only two things that dictate success rate for pass/run is the decision to pass or run and the number of men in the box. (This is obviously wrong; just go with it.) Therefore, the opposing defensive coordinator, Cate Perroll, in her infinite wisdom, decides to hedge her bets and pull a defender out of the box to help defend against the pass and hopefully not lose too much support in the run game. Given certain passing/rushing success rates, how effective would this strategy be? And would it be more effective than strategies that create some predictability in order to act against trend later?

Well, now we can have an incredibly rough way to answer these questions. Behold the defensive coordinator simulator. Find the expected success rates for rushing and passing by inputting 2, 1, 0, -1, or -2 defenders in the box and the expected passing rate to see how much predictability matters.

The second tab allows you to see a plot of the average success rates on 100 plays at a certain expected pass rate for each box defender state.  The error bars are the standard deviation from the mean after 100 simulations of 100 plays.

The app is in a rough stage that may be updated later, but it works (for now)!

Screen Shot 2019-07-31 at 9.18.36 PM.png


Please let me know how wrong I am.

Listen to our newest episode of Tasteful Profanity featuring Josh Hermsmeyer of 538 and Establish The Run right here!

Subscribe on iTunes HERE

Subscribe on Stitcher HERE