Predictability Kills

A guest post from Jonathan Bales of DallasCowboysTimes.com.

"I've always believed that it's important to show a new look periodically. Predictability can lead to failure."
- T. Boone Pickens

Predictability seems to be detrimental in many aspects of life, but its presence in football can be a death sentence for an offense. Offensive coordinators are in the business of hoodwinking defenses, and a large part of their ability to accomplish that task is being unpredictable. This is one of the major reasons offenses should throw more on 1st down and from run-oriented formations. The best offensive minds utilize defensive expectations to the fullest extent.

To examine the effects of predictability on an offense, one need only look at Jason Garrett's play-calls for the Dallas Cowboys over the past couple of seasons. Starting in 2009, Garrett began heavily implementing a formation known as "Double Tight Strong" (and variations of it with minor deviations).

Initially, the Cowboys found success from the formation, running for 149 yards on 19 carries (7.8 YPC) through the first five games of the 2009 season. All of those yards came on the same strong side dive play. As the season progressed, though, defenses appeared to catch on, limiting Dallas to 4.4 YPC on 64 more strong side dives from the formation (including just 3.2 YPC against all teams but Oakland).

Simple regression to the mean, right? Not so fast. In 2010, that average plummeted further to 2.2 YPC on 52 strong side dives. Since the start of the 2009 season, the Cowboys have lined up in "Double Tight Strong" 211 times, running the strong side dive 147 times (69.7%) for only 3.4 YPC. The average distance-to-go for a 1st down on these plays has been just above six—not significantly less than the mean distance-to-go for all plays—and 40.7% of them have come on 1st down.

Despite the lack of success on strong side dives from "Double Tight Strong," Garrett continues to dial up the play in 2011. At first glance, it appears predictability in Garrett's play-calling has been at least partially debilitating to the offense.

Of course, predictability is not inherently disadvantageous, as it can indirectly lead to advantages for an offense if the expectations it creates are properly exploited. As it relates to Garrett's strong side dives, their total impact cannot be conclusively stated until we take a more comprehensive look at "Double Tight Strong." Specifically, if the Cowboys can use the expectations created by strong side dives to capitalize on the rest of their plays from the formation, the aggregate value will be more than what initially appears to be the case.

A primary way by which Dallas could take advantage of this particular predictability is with playaction passes down the field. Show a strong side dive, fake a strong side dive, and air it out. Unfortunately, the Cowboys have not only failed to throw the ball downfield out of playaction looks from "Double Tight Strong," but they simply do not attack defenses deep on playaction looks in general. Since 2009, Dallas quarterbacks have thrown 20+ yards on only 8.7% of playaction passes. Further, in 2+ years of passes, Garrett has called a playaction pass only eight times with 1-4 yards-to-go for a 1st down. In comparison, he has called for a playaction look on 11 plays with 20+ yards-to-go. If offensive coordinators can garner an advantage for their team by manipulating defenses based on their preconceptions, Garrett is leaving plenty of potential yards on the table.

So how much predictability is too much? Well, any of it that is not utilized in some other facet of the game is excessive. In more direct terms, you want the "extra" value that comes from capitalizing on defensive expectations (whether it is in the form of big plays, run success rate, or something else) to exceed the "lost" value from the predictability itself. For Garrett and the Cowboys, that means exchanging a lower YPC for more big plays—a swap they are not currently seizing.

Jonathan Bales is the founder of DallasCowboysTimes.com, a site dedicated to film study and statistical analysis of the Cowboys, and has contributed articles to USA Today, the New York Times, and CBS Sports.

  • Spread The Love
  • Digg This Post
  • Tweet This Post
  • Stumble This Post
  • Submit This Post To Delicious
  • Submit This Post To Reddit
  • Submit This Post To Mixx

9 Responses to “Predictability Kills”

  1. Arif says:

    Curious. I would imagine however, that there are a number of offensive looks that have typical plays, and it would pay off to stagger when one deviates from the "established" typical, at a rate, generally of implementing and establishing new looks with new norms.

    Also, is data on how often an offense or a defense lines up with a particular formation widely available? I might even be willing to pay for it, if it includes game-by-game data.

  2. Anonymous says:

    very good article.

  3. Sam's Hideout says:

    Arif: my understanding is that each team has people who compile that information from viewing the coach's tapes--generally this is the very lowest step on the coaching ladder.

  4. Ian S says:

    It's just the same as teams who line up in power run formations on 4th and 1. Gee, I wonder if you fooled the defense. Spread it out, throw deep. It amazes me how coaches seem to call for 1 yard plays when there's a yard to go. There's a whole field out there. Use it!!

  5. Jonathan Bales says:

    @Arif From what I have studied with Dallas, there are certainly some looks which have specific plays attached to them, but none with the same sort of consistency as the strong side dive from DTS. Actually, nothing is even close, and the most alarming issue to me is that Garrett continues to call the play (albeit at a lower rate) in "regular" down and distances.

  6. Anonymous says:

    One of the major problems with this analysis is that tendencies -- something every team analyzes -- can be as much a weapon as they are a detriment. This is nothing new. Teams self-scout and opposing teams scout them. They analyze not just formations (keeping in mind sample size) and they analyze alignments of people like the runningbacks, linemen and so on. The depth of their alignment, where they line up in relation to other players, the splits between players, and so on.

    Sure, if a defense finds a key they will act on it. But a tendency also sets up the counter play. You show a look and throw play-action off of it (something I've seen Dallas do from this look a bunch). It's not as simple as you make it. On the one hand you can try to have zero tendencies but that's silly and overdone. Instead you just self-scout them and break them. It's football.

  7. Jonathan Bales says:

    @Anonymous I think you're dead on with your point about tendencies setting up counter plays, but my point is that Garrett is NOT taking advantage of his tendencies. I checked out my spreadsheets and the Cowboys actually have thrown the ball 10+ yards from all variations of this formation less than 5% of the time over the past 39 games. Even fewer of those have been playaction.

  8. Anonymous says:

    Someone should look at the 1989/1990 Vikings with Herschel Walker. He had to come off of the field for certain formations, I think it may have been pass plays in general. Anyways, most predictable play calling ever! And predictable result.

  9. bigmouth says:

    Absolutely fascinating...

Leave a Reply