"Establish the run" might be the three most over-used words in football analysis. Bad football analysis, that is. You would give yourself one heck of hangover playing one of those fraternity-type drinking games going bottoms-up every time you heard that phrase on a Sunday.
Establishing the run could mean a lot of different things. To most people I think it means that offenses will demonstrate the willingness to run frequently, hoping that defenses will bias towards stopping run later in the game. I think this interpretation has been debunked fairly thoroughly, so I'm going to look at it from a different angle.
Establishing the run could mean running often early in a game in hopes that a defense will weaken, and runs will be longer later in the game. I think this is more plausible theory. It makes sense on the surface. Running plays tend to batter the defense while passing plays allow the defense to go on the attack. By the fourth quarter, it's believable that defenders would be battered and fatigued. Defenders would get off of blocks slower and tackles would be sloppier. Watching Baltimore's two consecutive long TD runs late in the fourth quarter at Dallas in week 16 made me wonder: Does running frequently lead to longer gains?
To answer the question, I compared the average gains from running plays based on when they took place in a game. By "when," I mean which run play it was, not what time it took place in the game. In other words, if running frequently fatigues a defense, then the gain of a team's 30th run should tend to be longer than its first run.
We'd expect that the more runs an offense calls, the longer the subsequent runs should tend to be. The graph below depicts the average gain of each rushing play based on its order in which it was run. It plots the average gain of each team's first run of a game, 2nd run, 3rd run, ...etc.
(Data is from all regular season games from 2000-2007. All runs except kneel-downs were included.)
There's no increase in average gain as the number of runs increase. A team's very first run of a game is just as long as, if not longer than, the 20th, 30th or even 40th.
This result is despite the expectation that teams that are good at running would naturally run more often. Teams that are ahead toward the end of the game Those teams would therefore be the ones that we'd expect would accumulate more attempts. If so, we'd see the average gains increase as the run attempts mount. But we don't.
So this is evidence that runs are just runs, no matter how many have come before them. An offense can expect the same average gain on the first snap of the game as on the 80th, after 30 previous run plays.
The run may not set up the run, but does it set up the pass? Does running frequently allow longer gains on passes later in the game? I'll look at that question next.
Edit: Follow-up here that addresses many of the comments below.
Establishing the Run?
By
Brian Burke
published on 12/29/2008
in
basic,
game theory,
research,
run-pass balance,
strategy
Subscribe to:
Post Comments (Atom)
Some factors that skew the average down on late-game runs:
1) Runs at the end of the game happen mostly by teams that are ahead and running out the clock ("4 minute drill"), not by teams playing from behind. In these 4 minute drills it's hard to pick up big runs because the defenses are expecting the run.
2) When the team with the ball is behind by three points with just a few seconds left and in field goal range, they'll just call meaningless run plays and hope to set up the game winning field goal, and maybe (but rarely) get closer to the end zone.
I think if you neglect the effect of these plays, you'll definitely see a higher run average late in the game.
...........
The effect that "establishing the run" has on the defenses is I think indisputable from a common sense point of view. Especially with such a bruising runningback-by-committee team - even defenses will admit to being worn down. Whoever causes contact inflicts the most damage - this is why O-linemen like run blocking over pass protecting, and yes, why defenses wear down quicker than runningbacks.
When blind stats oppose such obvious football knowledge, we could question them - is it correlation or causation?. Stats can deceive if not analyzed correctly.
I think I should mention a few other things I did. I looked at the same graph while limiting the data to when the score was within 8 points. This eliminates the vast majority of the "meaningless" runs. The graph was almost identical and just as flat.
Also, if a team does not need a first down to run out the clock, they would kneel, right? So even in those late-game clock-eating situations, teams are still trying to get a first down--just like the Ravens vs. Cowboys.
If these runs are so meaningless, and yardage is so unimportant, then why "establish" the run in the first place? Why bother when the yardage late in the game is as unimportant as suggested?
I doubt the need-a-FG-to-tie/win ball-control runs are frequent enough to bias the data, but I can go back a limit the data to score differences beyond 3 points--or better, to situations outside FG range. I'll report back later today.
But if establishing the run is definitely, obvious, indisputable common sense, then I doubt anything would be convincing.
IMO, the two long runs by Baltimore vs Dallas happened because everyone "knew" a running play was coming based on the game situation. Dallas played with 11 men in the box, lost gap containment, and the runner was gone. Twice in row. The same thing happened to the Giants in week 17 vs the Vikings. Third and 1, 9 men in the box, Peterson goes 65-70 yards for TD.
Offensive lineman like running plays because they get to attack instead of being passive.
Common sense tells us that a successful running attack sets up future plays (run or pass) because a defense must adjust to stop the run. And a 3 to 4 yard run can be very effective because it moves the chains.
Teams that are behind in the 4th quarter will also call runs to slow down the pass rush and will call runs to make make first downs (3rd or 4th and short) to keep the ball.
I think that your chart does not prove, or disprove, your point.
I don't doubt any of that, but except in the rarest of cases are offenses not trying to maximize yardage when running the ball. A 3 or 4-yd run can get a first down as easily in the 1st quarter as the 4th. And no team would ever stop at 4 yds when they could get 5, 6, or more...
Despite these situations I believe we should expect to see even the slightest increase in yds gained the more runs there are. And if you say it doesn't matter or the effect is negated, then what's the point of wasting downs early in the game with non-optimal play calls?
I agree it's not proof, and I was careful not to claim anything beyond "evidence," which I think it is.
Limiting the data to plays outside FG range (the 34) shows the same flat graph. The average gain is higher across the board because there are no red-zone/goal-line runs factored in, but the result is the same--no increase for later runs. I also removed all runs from the last 3 minutes of the game with the same result.
Add to your article on game theory and play calling the last paragraph on the first citation and I think you have why it is important to "establish the run" covered.
JMM-I fully understand the importance of the mixed strategy of running and passing. But the point is being missed here. I'm not saying that running is not important at all. I'm saying that runs don't get any longer the more often you call run plays.
But further, for the run-pass mixed strategy to be effective, a team should not disproportionately call for runs early in the game. It is enough merely to be willing to run the ball at the situation-optimum mix at any point in the game, and not front-load runs to prove it.
Brian, as you authored the game theory article, I assumed you understood the importance of the mixed strategy. I think that article is an opportunity to build upon.
I suggested this as a "better" definition to the one you suggest in the article. With some notable possible exceptions (Jerome Bettis, Jim Brown) I would not expect any impact from the defense wearing down as I think you have shown.
And it was the last article on the first page of the first citation that captures the value of disproportionate running, or at least a change in the optimum ratio as the game clock ticks.
BTW, I enjoy your work, keep going and have a great '09.
You know what's funny, every time I watched Texas Tech play this year I heard some announcer say how tired the DLine got rushing the passer all day long
I have a feeling that what really wears a D out is doing something that works
JMM-Sorry, I misread your comment and was a little defensive.
I think we can agree that the Giants this year are a power running team who look to establish the run. I looked at their yards per run and it is similar to your chart..they do not improve as the game progresses, but they did average about a yard higher than the yards shown on your chart.
If a successful running team does not increase yards per run as the game wears on, then your premise ..."Establishing the run could mean running often early in a game in hopes that a defense will weaken, and runs will be longer later in the game." is most likely not correct. I think your first definition "that defenses will bias towards stopping run later in the game" is more correct, BUT the early running must be successful enough to be both a real threat (possible big plays) latter in the game AND successful enough that the defensive team needs to adjust to stop the run. It is easier to pass effectively with 8 defenders in the box.
There is difference between TRYING to establish the run by calling running plays and ESTABLISHING the run with successful running plays.
By the way, after looking at the play be play for the Giants to calculate the yds per run, it becomes apparent that it is difficult to know exactly when a team that is ahead starts to play the clock, which may skew stats one way or another.
Passing wears out D-lineman more than running. For DL stopping the run mostly consists of digging-in and not getting shoved into a LB. Stopping the pass means doing a 15-yard wind sprint while trying to get past a 300-LB man coming after you with bad intent, while you try and catch a smaller, lighter and presumably fleeter QB. Pass rush wears out the DL, not stopping the run.
Passing to set up the run will usually result in a soft, tired DL that can then be pushed around by a coordinated OL.
It's not strong evidence, but I have to say that I am surprised how little covariance is evident. It deserves some more detailed analysis.
About a year ago you had a series of posts about risk/reward and pass/run ratios, which proved quite ineresting.
I'll repeat a thought I made in the comments then: that average yards per carry may be a less meaningful indicator of rushing success than median yards, or the mode.
Intuitively it's always seemed to me that trying to rush for a lot of yards on any given carry is foolishness -- for a bunch of yards, it's almost always a better shot to pass (absent specific tactical considerations).
In fact, I've always kind of felt that there's a "Curse of the Great Break Aaway Running Back" -- those guys always seem to play on teams that are losers (unless they also have a top QB to throw for yardage).
Gayle Sayers, OJ Simpson, Earl Campbell, Eric Dickwerson and Barry Sanders all played on teams with career losing records, as did Walter Payton for the first 10 years of his career until Ditka and Ryan showed up with the 46 defense.
OJ averaged 4.7 yards per carry over *ten years* while his teams went 43-95-2, about an average of 4-10.
To me the "curse" makes sense if, say, Barry Sanders's team thought: "We have Barry Sanders, so we are going to optimize our team to have Barry run for yardage, instead of pass for it" so they sacrifice their passing game. Which is how the Sanders Lions always looked to me.
But OTOH, running is better than passing for converting *short-yardage* first downs, controlling the clock, having a reliable alternative to passing for shorter yardage plays (to make the D worry about two things).
For all these purposes (especially the first two) *reliablity* at getting a few yards *every time* is much more valuable than getting a higher average yardage with higher variability of result. In short: On 3rd-and-two, getting two yards every time is a lot better than averaging four yards per carry, twice as much, with a distribution of 0, 1, 3, 6, and 10 yards, so you miss the first down and turn the ball over 40% of the time.
If true, the median or mode for yards per carry would be more indicative of rushing success than the mean.
In empirical support of this, I remember looking at some of Lombardi's Packer teams and seeing they had near the *lowest* yards per carry in the NFL, while crushing through the league with its top running game.
Lowest yards per carry, yet the dominating best -- apparently an example of how a team that averages 3.5 per carry each and every time is an unstoppable scoring machine, while one that averages 5.0 per carry with runs distrinuted from -3 to +15 will be punting a lot.
All that said, it's all just a feeling with me, I've never seen data on the subject, nor even a data base that had the data.
But now you have play-by-play data from 2000-2007?
Hey, if you ran the numbers on this and could answer my questions here ... I'd become an even more dedicated reader than I am. All off-season too!
Jim,
To save Brian some work here, I've been working on this and you're right to some extent. Basically, the yards gained that are less than the "to go" are more valuable, so those very long, breakaway runs should be discounted. On the other side, the payoffs when the gain does not result in a first down are generally convex, which is a fancy way of saying consistency is not beneficial. If it's first and ten, a RB that gets 6 yards half the time and 0 yards half the time is more valuable than a running back that always gets 3 yards. Of course, a consistent running game is much better on third and short.
Eventually I'm going to write all my research on this topic and at least link to it here.
-Steve
Median stats would be very interesting. Can't wait to see the stuff on the value of consistency. Steve, please let us know when you've written it up.
When I hear/think/say "establish the run," I think it means "show that you are capable of running successfully." That would be a necessary part of mixing the run and pass plays. Perhaps I'd go further and say I don't think "SHOW that you can run successfully," but simply "run successfully." I realize I'm talking about connotative interpretation of words here, but to me "establishing the run" isn't about frequency of run plays, but success of run plays.
However, this is giving commentators too great a benefit of the doubt--often they explicitly follow up and show that they mean it in the ways that have been well debunked.
For many years I've had the feeling that teams "burn" the clock early in the game to get to a point where the game comes down to the last segment of the game. I'm not yet sure how long that segment is. In basketball you often hear that one need only watch the last two minutes of the game. I've felt that the same is true of football but not as obvious.
If this were the case then all teams would run the football early in the game, establish the run if you will, getting to a point where the clock would be short and the score would be close. Now a team would only need to march down the field and score to win.
Ed
Here in New York, people love to talk about how Derrick Ward basically owes his 1000 yard season to Brandon Jacobs for "wearing down the defense." The theory is that Jacobs pounds the defense with his bruising style in the first half and then Ward's runs are that much more effective later in the game because the defense cannot keep up with him.
This theory got even more legs a few weeks back when Ward was the featured RB in a game that Jacobs was out. Ward had a tough day and was not very successful at all.
What do you think about this idea? Your research did not distinguish between the type of running back. It is probably fair to say that Brandon Jacobs inflicts more punishment than say Reggie Bush. So do you think it is possible that a "bruising back" can set up better runs for the "slash and dash" type of back?
Brian - If you throw the weight of the RB into your initial regression as an interactive effect on the attempts, it might give some interesting results.
-Steve
Another question is, "Does passing show the same consistency as attempts increase ?"
ANother thought - are there in fact more breakout runs in the 4th quarter? It would be good to see the distribution of 45+yard runs by quarter as well as 65+yd runs by quarter. Clearly if there are as many in the first quarter as there are in the 4th quarter then the argument of wearing down the defence goes out the window. OF course, if the frequency is greater in quarter 4 there may still be another reason for it, viz, the defence is jamming the box to stop the run.
Ed
I think a better way to analyze this would be what the person above me said, analyzing the number of long runs as opposed to the average length of a running play. Teams that are ahead at the end of a game tend to run for less yards on average because the defense knows they will run and sells out against the run. So it's natural to think that running plays would actually gain less yards towards the end of the game, as the data suggests.
I think a better way of seeing if running more wears down a defense would be to look at the number of 20+ yard runs and number of yards after contact. I really don't think it would though, doesn't running the ball wear out the offense too?
Check out 2006 Football Prospectus article, "Do Bigger Backs Break Down the Defense?" (p134)
The short answer is that they do not.
In fact it appears that to the extent that anyone 'wears down the defense' it's the smaller backs.