Cowboys Film Study: 3rd Down Play-Calling
Page 1 Page 2
Note: This is a two-page entry.
Perhaps our favorite statistical analysis of 2010 was the study we conducted on the Cowboys’ 2nd down play-calling in 2009. We discovered that offensive coordinator Jason Garrett was extremely predictable in his play-calling on 2nd down–so much so that he was 2.95 times more likely to run on 2nd down after a 1st down pass than after a 1st down run, even when the distance-to-go was identical.
In that particular analysis, it is important to note we are not criticizing the team’s run/pass ratio in general. Garrett could dial up a pass on 2nd down 95 percent of the time and we would have no qualms–as long as that percentage remains stable in similar situations whether the previous play was a run or a pass.
Unfortunately, that stability is not apparent. We concluded this was the result of Garrett attempting to “mix it up.” Human beings naturally tend to think the next item in a random sequence will be different from the previous one. This is not the case, however, meaning Garrett’s attempt to “mix it up” with his play-calling has (quite ironically) led to his predictability.
The strength of correlation between Garrett’s 1st and 2nd down play-calls led us to question the relationship between his 2nd and 3rd down play-calls. Before delving into the results, it is important to note that these relationships (that between 1st and 2nd down play-calls and that between 2nd and 3rd down play-calls) are not identical. Plays on 1st down are (almost) all run in the same situation–1st and 10. 2nd down play-calls, however, are more closely linked to the ‘distance-to-go’ due to the varying nature of this distance on 2nd down.
For example, 2nd and 1 plays are likely to be the result of a 1st down pass–a nine yard gain is more likely from a pass than a run. On 3rd and 1, however, the previous play is more of a mystery. The chances of the preceding play having been a run are probably just as likely as it having been a pass.
Nonetheless, we can still draw meaningful conclusions from our film study-derived results. Those findings are below.
The first thing we notice is that the discrepancy between 3rd down passes after a run and those after a pass is nowhere near as great as those on 2nd down (shown below). For example, while the rate of passes on 2nd and 3 to 7 was 2.95 times as high after a 1st down run as opposed to a 1st down pass, the largest discrepancy between 3rd down play-calling occurred on 3rd and 1 to 2, when the Cowboys were 1.75 times as likely to pass after a 2nd down pass as opposed to a run.
Another interesting characteristic of Garrett’s 3rd down play-calling is that the relationship between passes after a 2nd down run and those after a 2nd down pass is positively correlated, i.e. as one increases, so does the other. This occurs in each distance-to-go subset of 3rd down plays and is in direct opposition to the negative correlation displayed in the ‘2nd and 3 to 7’ subset of 2nd down play-calls.
A final intriguing note is that, while the type of play (run or pass) that Garrett dialed up on 2nd down was in opposition to his 1st down call, his 3rd down play-calls were more likely to be the same as those on 2nd down. Put simply, the Cowboys were actually more likely to pass on 3rd down after a 2nd down pass than after a 2nd down run.
All of that is basically a complicated way of saying Garrett was much less predictable in his play-calling on 3rd down than on 2nd down. Still, he wasn’t perfect. Like we said, he was 1.75 times as likely to pass on 3rd and 1 to 2 after a 2nd down pass as opposed to a 2nd down run. The situation is identical, so a perfect play-caller would have an identical pass rate regardless of the call on the previous play. We by no means expect Garrett to be perfect, but we would certainly hope for a more closely linked relationship.
Click Page “2” to read the rest of this analysis.
Page 1 Page 2