Cowboys Film Study: 3rd Down Play-Calling;

[cowboyjoe]

Cowboys Film Study: 3rd Down Play-Calling

2nd down run rate is directly related to 2nd down pass rate, as the pass percentage is simply (100-run percentage).
Note: There were zero runs on 3rd and 8-10 after a 2nd down pass.



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.


2nd down run rate is directly related to 2nd down pass rate, as the pass percentage is simply (100-run percentage).


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.




Cowboys Film Study: 3rd Down Play-Calling


2nd down run rate is directly related to 2nd down pass rate, as the pass percentage is simply (100-run percentage).

Note: There were zero runs on 3rd and 8-10 after a 2nd down pass.



Page 1 Page 2



The most effective way to judge how this predictability affected the Cowboys is to analyze their 3rd down success in particular situations. Did the fact that the Cowboys were 1.75 times as likely to pass in certain 3rd down situations after a particular 2nd down play result in decreased success in these scenarios, for example?


Note: There were zero runs on 3rd and 8-10 after a 2nd down pass.
The graph above displays the Cowboys’ efficiency on 3rd down runs. Notice that the team was much more likely (in the same situations) to be successful on 3rd down runs if the previous play was a pass as opposed to a run, particularly on 3rd and 3 to 7. In those situations, Dallas gained an incredible 2.81 times the yardage after a 2nd down pass (7.86 yards-per-carry) compared to a run (2.80 yards-per-carry).

When contrasting this with the initial 3rd down pass rate chart, it makes sense. The Cowboys were more likely to pass on 3rd down after a 2nd down pass, meaning if a defense anticipated a pass on 3rd down due to the previous play-call (consciously or not), they would be surprised by a run, thus yielding the higher yards-per-carry that we observe.

When analyzing the Cowboys’ 3rd down pass efficiency, we uncover the same sort of results. Because Dallas is more likely to pass on 3rd down after the same play type on 2nd down, we would expect their success on 3rd down passes to be greatest after 2nd down runs. This is the exact relationship which is apparent in the graph below.


Notice that the Cowboys were more successful in every subset of 3rd down passes if they were conducted after a 2nd down run. The most dramatic difference is seen on 3rd and 1 to 2, when Dallas gained 2.11 times as many yards after a 2nd down run (8.71 yards-per-pass) as compared to a 2nd down pass (4.13 yards-per-pass).

The offense’s overall 3rd down efficiency indicates that defenses were prepared to face certain play-calls as a result of the call on the previous play. The Cowboys were much more successful on 3rd down passes after 2nd down runs, and vice versa, due to the lower rate at which these play combinations manifested themselves.

Again, we cannot emphasize enough that we are not criticizing the percentage of runs/passes on 3rd down in general, but rather the percentage that follow 2nd down passes versus the percentage that follow 2nd down runs. In a perfect world, these percentages would be equal. At the very least, we would hope the discrepancy wasn’t as large as the one we observe.

In our study on Garrett’s 2nd down play-calling, we wrote that “the failures of other coordinators around the NFL do not justify the failures of Garrett. His success is independent of that of other offensive coordinators.” We maintain the truth of that claim.

[percyhoward]

Originally Posted by cowboyjoe

The most effective way to judge how this predictability affected the Cowboys is to analyze their 3rd down success in particular situations.

I think a better way to find some connection between the play call and the success of the play would be to compare late-season games with games played earlier in the season, when the "tendencies" hadn't become tendencies yet.

[Jongb35]

Originally Posted by percyhoward

I think a better way to find some connection between the play call and the success of the play would be to compare late-season games with games played earlier in the season, when the "tendencies" hadn't become tendencies yet.

That's a good idea. I will do something like that in my next study.

If you (or anyone else) have any things you want us to take a look at, let me know. We tracked basically every part of each play you could imagine, so just let me know and I'll make sure it gets done. Always looking for cool new ideas.

[craig71]

How do you account for audibles?What about game situations,end of half or down by multiple scores?

Craig

[cowboyjoe]

i thought i put dallascowboystimes.com with the title, guess not, but the article to me which was pretty good was from there;

http://dallascowboystimes.com/2010/0...-play-calling/

[Jongb35]

Originally Posted by cowboyjoe

i thought i put dallascowboystimes.com with the title, guess not, but the article to me which was pretty good was from there;

http://dallascowboystimes.com/2010/0...-play-calling/

No prob Cowboy Joe...don't worry about it. But I definitely do want some forum members' thoughts on potential stat analysis/film study things they'd like us to do.

[cowboyjoe]

Originally Posted by Jongb35

No prob Cowboy Joe...don't worry about it. But I definitely do want some forum members' thoughts on potential stat analysis/film study things they'd like us to do.

they will probably give you some, vela is very good at that type of thing

[percyhoward]

Originally Posted by Jongb35

That's a good idea. I will do something like that in my next study.

If you (or anyone else) have any things you want us to take a look at, let me know. We tracked basically every part of each play you could imagine, so just let me know and I'll make sure it gets done. Always looking for cool new ideas.

We flat out suck in short yardage situations. Yesterday I dug up these numbers on short yardage 2009 (2 yards to go or less, any down):

The percentage is the success rate, in first downs, of each back.

Felix 75% (6 first downs on 8 runs)
Choice 60% (6 of 10)
Barber 57% (24 of 42)

These numbers make it look like we ran Barber far too much in short yardage. It would be interesting to know what formations/personnel/plays we ran on those 60 occasions. Any detailed study of our short yardage offense would be interesting.

[dogberry]

What are the league numbers to compare our short yardage results with?

Thanks for all the work

Kent

[percyhoward]

Originally Posted by dogberry

What are the league numbers to compare our short yardage results with?

Thanks for all the work

Kent

Power Success: Percentage of runs on third or fourth down, two yards or less to go, that achieved a first down or touchdown. Also includes runs on first-and-goal or second-and-goal from the two-yard line or closer.

1. Cin 79%
2. Mia 79%
3. GB
4. Oak
5. Pit
6. NYJ
7. NO
8. Jax
9. Bal
10. NE
11. Car
12. Ind
13. TB
14. Cle
15. Ten
16. SF
17. Atl
18. KC
19. Phi
20. NYG
21. Hou
22. Det
23. Min
24. Was
25. Chi
26. Dal 58%
27. Den
28. Ari
29. Sea
30. StL
31. Buf
32. SD

[Jongb35]

Originally Posted by percyhoward

We flat out suck in short yardage situations. Yesterday I dug up these numbers on short yardage 2009 (2 yards to go or less, any down):

The percentage is the success rate, in first downs, of each back.

Felix 75% (6 first downs on 8 runs)
Choice 60% (6 of 10)
Barber 57% (24 of 42)

These numbers make it look like we ran Barber far too much in short yardage. It would be interesting to know what formations/personnel/plays we ran on those 60 occasions. Any detailed study of our short yardage offense would be interesting.

We had done some research on that in our "Grading the 'Boys" Running Backs segment. I don't want to get in trouble for advertising so I won't post the link, but our film study showed Jones was 5/5 on runs with 1 yard to go (for either a 1st down or TD), Choice was 5/7, and Barber was 14/25.

[percyhoward]

Originally Posted by Jongb35

We had done some research on that in our "Grading the 'Boys" Running Backs segment. I don't want to get in trouble for advertising so I won't post the link, but our film study showed Jones was 5/5 on runs with 1 yard to go (for either a 1st down or TD), Choice was 5/7, and Barber was 14/25.

Your numbers for the 1-yard to go's are consistent with the ones I collected from si.com's situational stats for 1-2 yards to go. The picture of "who did what" is there, but the play calls and certainly the formations can only come from film study.

You'd have about 90 plays to look at that were short yardage. Roughly 60 runs and 30 passes.

[wick]

I'd be curious to see the percentage of times NFL teams call a running play after a first down incompletion. Intuitively, it feels like it happens at an insanely high rate, to the point where I slam my fist in frustration when we do it over and over again.

[cowboyjoe]

Originally Posted by Jongb35

We had done some research on that in our "Grading the 'Boys" Running Backs segment. I don't want to get in trouble for advertising so I won't post the link, but our film study showed Jones was 5/5 on runs with 1 yard to go (for either a 1st down or TD), Choice was 5/7, and Barber was 14/25.

hey buddy, glad your getting some feedback on your questions

when i see some good stuff on your site, i always try to post it, among some other sites i visit

[Jongb35]

Originally Posted by cowboyjoe

hey buddy, glad your getting some feedback on your questions

when i see some good stuff on your site, i always try to post it, among some other sites i visit

thanks joe. i actually did a study exactly like the one the guy who posted above you is searching for...is it okay if i post it?