SITE NEWS: We are moving all of our site and company news into a single blog for Sports-Reference.com. We'll tag all PFR content, so you can quickly and easily find the content you want.
Also, our existing PFR blog rss feed will be redirected to the new site's feed.
Pro-Football-Reference.com » Sports Reference
For more from Chase and Jason, check out their work at Football Perspective and The Big Lead.
ShareThisStill more on second-and-ten
As promised yesterday (also see Monday's post), here is some data on the results of second-and-10 plays. The number in each slot is the percentage of all plays in that category that gained the specified amount of yardage. E.g. 9.0% of all second-and-ten rushing plays after a first-and-10 pass went for negative yardage.
Second-and-10 running plays
AfterRun AfterPass
=====================================
Negative yardage 8.0 9.0
0, 1, or 2 yards 24.9 32.0
3, 4, or 5 yards 29.4 30.2
6 or 7 yards 11.4 10.4
8 or 9 yards 11.4 8.1
10 or more yards 14.9 10.3
Second-and-10 passing plays
AfterRun AfterPass
=====================================
Negative yardage 6.6 6.7
0, 1, or 2 yards 38.6 39.0
3, 4, or 5 yards 9.7 10.0
6 or 7 yards 10.5 9.0
8 or 9 yards 8.0 8.3
10 or more yards 26.7 26.9
On passing plays (the second table), the two columns are essentially the same. So second-and-10 passing plays are equally successful regardless of whether the first down play was a run or a pass. But the first table shows that second-and-10 rushes are less successful following a first-and-10 pass play.
Does this mean anything? I'm not sure. If you line up these same columns in a different way, you get this:
After a first-down run
Run Pass
==================================
Negative yardage 8.0 6.6
0, 1, or 2 yards 24.9 38.6
3, 4, or 5 yards 29.4 9.7
6 or 7 yards 11.4 10.5
8 or 9 yards 11.4 8.0
10 or more yards 14.9 26.7
You could stare at this chart for awile and debate whether the pass plays or the run plays were more effective in this situation. The pass plays get you the first down more often, but they also leave you in 3rd-and-long more often. The fact that it's debateable is, in my opinion, evidence that teams are generally using roughly the right run/pass mixture in this situation.
Now look at this chart.
After a first-down pass
Run Pass
==================================
Negative yardage 9.0 6.7
0, 1, or 2 yards 32.0 39.0
3, 4, or 5 yards 30.2 10.0
6 or 7 yards 10.4 9.0
8 or 9 yards 8.1 8.3
10 or more yards 10.3 26.9
Clearly passing is more effective than running on second-and-10 following a first-and-10 pass, which indicates that teams could benefit from passing more. I'm not convinced that there are sufficient long term benefits to throwing all these changeups to justify the decreased effectiveness on the current drive. I am not suggesting that teams should stop running on second-and-10; I'm merely suggesting that they could possibly pick up more first downs if they ran more like 30% of the time rather than 55% of the time.
This entry was posted on Wednesday, January 31st, 2007 at 5:09 am and is filed under General. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.
"I think there’s a real long-term benefit to mixing up your playcalling. If a defense knows you’re going to be passing every play, they can send rushers hard off the edge to get pressure on your quarterback, and drop their linebackers and safeties deep to cover passes better."
This quote was from Vince late yesterday. I do not disagree with the general premise. And this is why teams don't want to be trailing the Colts with Freeney, if they don't have to fear the run.
However, mixing up play calling to me would involve passing more than running except short yardage, sometimes passing when the team expects a run, sometimes running when the team expects a pass, and sometimes, continuing to pass or run consecutively when the team may expect the normal response to be "mix it up."
Mixing up play calling is not pass, then run. When teams go incomplete pass/run, it results in a gain of 2 or less yards 41% of the time. That suggests teams are playing run.
As an extreme example since Vince opined that good coaches mix it up, here is the Pass/Run sequence on 1st and 2nd down only, in the 1st half of the New England at Minnesota Monday Night game in week 8. I've marked runs as O, since R and P are too close (and so I could spell POOP), and Q is for Quarterback run.
POPPPPPPPPPOOPPPOPPPOPPPQ
Is there a column for turnovers because I think that is an important part of this. Looking at the data for 2nd and 10 after a pass I see a 30% chance of having a 3rd and 5-7. I dont have the data but i would think that passes with 6 yards to go for the first down are much more likely to go for 1st downs then 10 yards to go. I would rather a shot at a run for 4 yards and then a pass for 6 than 2 attempts for 10 yard passes. Especially since on 3rd and 10 it is very easy to defend. With the run option you are putting yourself in a better 3rd down position while also having the chance of getting a first down on the play. The coaches are not just looking at 1st and 2nd downs as you have been doing but are factoring 3rd downs in there as well.
Or, since the basic goal of an offense is to move the ball, maybe we could see the percentage of successful first downs after each sequence:
Pass-Pass
Pass-Run
Run-Pass
Run-Run
Including when the 1st was gained on 2nd down? I mean, we can talk about how good it is to say out of 3rd-and-long, but shouldn't we actually look at how often they convert anyway?
Factoring in the resulting 3rd down situations (or the lack of need for them (because you already got the first)) is precisely what this post is all about.
You really think the run column in that last table offers a better set of possibilities than the pass column? I admit I didn't do any serious analysis to justify that, but that was because I thought it was too obvious to need any analysis. If you traded a bunch of actual first downs (in the pass column) for 3rd-and-sixes, the columns would look very similar.
As always, I could be wrong.
Doug, i am not saying that the run list is a better set of possibilities. I think that if the outcomes are close, which they may be, of getting a first down on that drive (not neccessarily on 2nd down) then it may explain why coaches are running there. A run is definitely safer and the theory is that it tires out the defense more. I would be interested to see the turnover rates and also the data that Jason asked for.
The more I think about it (intermittently, as I am supposed to be working) I think that looking at one series independently is probably a mistake.
Is it true or a myth that the running game gets better as the game goes on? If it is true, then is it linked to how many plays the offense has run earlier in the game?
If there is a link between # of plays previously run, and success in the running game, then there is a clear advantage to NOT having a quick-strike offense, but rather one that consistently moves the ball in small (4-8 yard) increments.
The point being, that there may be reasons to not seek the 1st down on every play, when evaluated over the course of a game, rather than a single series. Thoughts?
OK, MO, I see what you're saying. Unfortunately, turnover data isn't going to be possible (but it has been duly noted that, if it were possible, the data could potentially change my mind). Looking at the third-down results is going to be tricky, but I might be able to do it.
While I do think that the "after-run" mixture is closer to optimal, that's a bit of a side issue. Everything you said here:
would indicate that coaches should be running more on 2nd-and-10 after a run.
I still just don't understand the dichotomy. If a run is good after a pass, then a run should be good after a run, right?
I understand that people want to look at this series in the context of the whole game, but teams average about five 2nd-and-10s per game. Running 30% of the time instead of 56% of the time would add up to about one fewer run per game. I just don't see how that can have any meaningful long term effect. But I do see a short-term advantage to running 30% of the time instead of 56% of the time.
Again, I'm not saying pass every time on 2nd-and-10. I'm saying pass more often than you run on 2nd-and-10.
I think the issue is that you would have to do this analysis, and continue to broaden it.
* Isn't it statistically better to pass on 1st down (more average yardage)?
* Isn't it statistically better to pass on 3rd down?
* How about 3rd down?
In what situation does running give a higher average yards per play? If never, then by this reasoning (if followed) isn't passing always better?
Didn't some teams in the late 80's and early 90's subscribe to this theory?
NOTE: I have no idea if any of the above is actually true.
Ooops - I mean how about 4th down.
"* Isn’t it statistically better to pass on 1st down (more average yardage)?
* Isn’t it statistically better to pass on 3rd down?"
That depends on how many yards to go for a first down, the time remaining, the score, the defense you're facing, and a bunch of other factors. If it's 1st-and-10 in the first quarter with no score, I'd say passing is probably the way to go. On the other hand, if it's 2nd-and-1 on the opposing teams 1 yard line, any offensive coordinator that calls a pass should almost certainly be fired. There is no decision that is "statistically better" in all situations.
"In what situation does running give a higher average yards per play? If never, then by this reasoning (if followed) isn’t passing always better?"
I don't think running ever gives a higher average yards per play (except maybe goal line plays from inside the 2 yard line), but there are reasons why passing always isn't better:
1) Running plays are more likely to gain a first down in short yardage situations. A pass is more likely to result in a big gain, so it has a higher average yards per play, but it's also more likely to result in no gain at all (incompletion). In a short yardage situation, the highest priority is getting the first down. Any additional yards on top of that are gravy.
2) At the end of games, running takes time off of the clock, which helps to preserve a lead.
3) Running plays are less likely to result in a turnover. Getting more yards per play doesn't help as much if you have a lot more turnovers.
"Didn’t some teams in the late 80’s and early 90’s subscribe to this theory?"
You're thinking of the Run-and-Shoot offenses. These were run by the Oilers, Lions, and Falcons in the early 90's. They didn't always pass, although it was a fairly pass heavy offense. What they did was to determine their play call based on how the defense lined up. So if a team stacked the box, they would throw. If a team only left 5 or 6 men in the box, and used the others to cover the 4 wide receivers that the Run-and-Shoot used, they would run. And they had a good deal of success with running, as evidenced by the performance of Barry Sanders on the Lions.
I guess I should have addressed 3rd down just then, so here goes: if it's 3rd-and-1 on the opposing teams 1 yard line, any offensive coordinator that calls a pass should probably be fired.
Alex, thanks for the response. I guess my sarcasm wasn't obvious, but you answered in spades.
The point is, that analyzing the average number of yards gained on a play is just not complete enough, and if you use that to analyze 2nd and 10, you have to use that as the metric to analyze any play - why is 2nd and 10 special?
Jacob, the record will indicate that I never once mentioned yards per play.
In fact, on the "after first down run" table, it's pretty clear that passing has a greater average gain. But yet I said,
Average yards is a straw man. No one is arguing based on that.
That's consistent with yesterday's article if defenses are catching on that offenses are increasing their second-down run frequencies after an unsuccessful first-down pass.
Jacob wrote:
[quote]Isn’t it statistically better to pass on 1st down (more average yardage)?[/quote]
If a team passed on 100% of its first downs, it would probably average very few yards per first down because it would be so predictable.
The proper strategy is almost certainly not to run 100% of the time on first down, nor to pass 100% of the time on first down, but to mix the two up so as to keep the defense honest.
(The appropriate mix depends on how well a run works against a defense expecting a run, how well a run works against a defense expecting a pass, how well a pass works against a defense expecting a run, and how well a pass works against a defense expecting a pass. And that is obviously way oversimplified since there are more than two types of defensive play calls -- and more than two types of offensive play calls as well [should screens be considered runs or passes for this purpose, or be a separate category?].)
Also, the results cannot be measured simply in yards per play. For a sufficiently positive expected number of yards, a smaller standard deviation is preferable for the offense. So rushing for an expected four yards can be equal in value to passing for an expected six yards if the rushing results have a smaller standard deviation. (Incidentally, this is why I believe at least some of the criticism of Passer Rating -- namely that, by considering completion percentage in addition to yards per attempt, it unduly rewards dink-and-dunk passers and penalizes longer passers -- is misplaced.)
Running and passing breaks down as follows:
Running and passing are about equally successful when you use the definition of yards needed to be a successful play by down and distance as refined from "The hidden game of football" by sites like http://www.footballoutsiders.com and http://www.twominutewarning.com. By this measure running and passing come in at around a 48% success rate.
Passing leads to bigger plays more often and thus the higher yards per attempt but passing has a much higher turnover rate via interceptions plus the fumbles that occur on sacks. The average RB only turns the ball over at rate of 1.5% wheras a QB's avg turnover rate combining fumbles and sacks is cloer to about 5%
So factoring the turnover difference there is much less of a premium for passing than sometimes appears.
Doug:
Clearly passing is more effective than running on second-and-10 following a first-and-10 pass, which indicates that teams could benefit from passing more.
Wait, why? I don't think it's that clear. I think you'd need to factor in the resulting 3rd down conversion rate to see what the chance is that you're going to convert the 3rd as well. Passing on 2nd and 10 puts you in a 3rd and 7+ situation more than running does.
Sidenote:
NFL.com reported something I had long suspected, that teams run a draw upwards of 60% of the time on any first and more than ten situation (following a penalty). I had noticed that anecdotally, and found it interesting both because of the general coaching philosophy behind it and because some defenses seem routinely unprepared for one of the most predictable calls in any game.
I think you’d need to factor in the resulting 3rd down conversion rate to see what the chance is that you’re going to convert the 3rd as well.
Not only that, you'd also want to check how it affects the rest of the drive, particularly if a 3rd and medium situation puts you in a better position for a big play. Then if you were concerned about the effects of the current score and time remaining, how does it affect your chances of winning the game? And of course how the increased likelihood of a turnover plays into it. I love this analysis, and I believe Doug's conclusion is probably correct, but there's a lot of complexities to this question not fully addressed.
Obviously one needs to factor in the 3rd down conversion rates. I was doing that. I was just doing it mentally with intuition and no data. As I said earlier, I could be wrong.
But whatever. I've gotten way off track, and I've done a horrible job of communicating what I think is interesting about all this.
If you look at this:
Run Pass
==================================
Negative yardage 9.0 6.7
0, 1, or 2 yards 32.0 39.0
3, 4, or 5 yards 30.2 10.0
6 or 7 yards 10.4 9.0
8 or 9 yards 8.1 8.3
10 or more yards 10.3 26.9
and are not sure if the runs are the passes are more effective, then you can't possibly look at this:
Run Pass
==================================
Negative yardage 8.0 6.6
0, 1, or 2 yards 24.9 38.6
3, 4, or 5 yards 29.4 9.7
6 or 7 yards 11.4 10.5
8 or 9 yards 11.4 8.0
10 or more yards 14.9 26.7
and NOT think the runs are more effective.
Maybe 56% is the right running percentage. Maybe 30% is the right running percentage. Maybe it's 85% of 12%. I happen to think 30% is about right, but I don't really care. My point is that 56 and 30 can't BOTH be right.
The irrationality I keep talking about is not necessarily the 56% run percentage. It's the fact that there is such a huge difference between the running percentages in the two situations.