NFL's Short-Yardage Strategy: Hiding the Sneak?

NFL's Short-Yardage Strategy: Hiding the Sneak?
NFL's Short-Yardage Strategy: Hiding the Sneak?
Photo: USA Today Sports Images

by Scott Kacsmar

Before the 2013 season started I did a study of the most unstoppable play in the NFL: the quarterback sneak. When something works over 85 percent of the time regardless of the quarterback's height, speed and other abilities, then how does this not become a more widespread play call when a team needs a yard? Wouldn't the Colts be better off letting Andrew Luck pick a spot to squeeze in a few inches behind his center than to hand off to Trent Richardson four yards in the backfield? Using the quarterback as a short-yardage weapon more often is something every team should look into. After crunching the numbers for 2013, I can say the results were on par with past years, except for one major difference.

Quarterbacks carried the ball on third-and-short at the lowest rate (7.6 percent) in five years and the lowest rate on fourth-and-short (14.9 percent) since 2000. There was a 35 percent decrease on third-and-1 runs by the quarterback in 2013 compared to 2009-12. So instead of seeing the usual 90-100 of those plays, there were just 61 this season.

Despite a league built with the biggest, fastest, most athletic group of quarterbacks in NFL history; an unrivaled attention to safety and protection of those players; massive offensive linemen; and the liberty to make the quarterback an effective runner, the most unstoppable play in the game mysteriously went back into hiding. Just when you think you've planted a flag for "Analytics" into the NFL landscape, a tank runs it right over.

Did the conversion rates justify the suboptimal decision making from this past season? Let's examine.

Note: All data is for the regular season only and excludes spikes, kneel downs, intentional safeties and botched kicks (so this play by Miami does not count as a failed fourth-and-1 run). Passing stats include sacks. The only plays studied are on third or fourth down with 1-2 yards to go, because these are the most important short-yardage plays that essentially decide if the offensive drive will continue. Most data goes back to 2009.

Third Down

Our first table looks at the summary of plays on third-and-1 and third-and-2 since 2009:

3rd-and-1 Passes 3rd-and-2 Passes
Year Plays 1st Downs Conv. Rate Year Plays 1st Downs Conv. Rate
2009 190 112 58.95% 2009 325 176 54.15%
2010 211 122 57.82% 2010 355 187 52.68%
2011 200 114 57.00% 2011 352 169 48.01%
2012 241 126 52.28% 2012 356 186 52.25%
2013 244 121 49.59% 2013 357 174 48.74%
Total 1086 595 54.79% Total 1745 892 51.12%
3rd-and-1 Quarterback Runs 3rd-and-2 Quarterback Runs
Year Plays 1st Downs Conv. Rate Year Plays 1st Downs Conv. Rate
2009 101 88 87.13% 2009 19 13 68.42%
2010 89 67 75.28% 2010 25 17 68.00%
2011 89 75 84.27% 2011 26 19 73.08%
2012 96 79 82.29% 2012 30 19 63.33%
2013 61 51 83.61% 2013 37 22 59.46%
Total 436 360 82.57% Total 137 90 65.69%
3rd-and-1 Runs (non-QB) 3rd-and-2 Runs (non-QB)
Year Plays 1st Downs Conv. Rate Year Plays 1st Downs Conv. Rate
2009 516 350 67.83% 2009 173 100 57.80%
2010 446 299 67.04% 2010 171 100 58.48%
2011 429 294 68.53% 2011 148 83 56.08%
2012 447 290 64.88% 2012 150 81 54.00%
2013 459 310 67.54% 2013 128 85 66.41%
Total 2297 1543 67.17% Total 770 449 58.31%

Generally, the difference between conversion rates for a quarterback run and regular run is about the same as the run and those dangerous (but more rewarding) passes. With the quarterback runs down to 98 third-down plays in 2013, those opportunities had to go somewhere else. The passing game, or the least efficient way to convert a short-yardage situation, was the main beneficiary. There were more pass plays on third-and-1 (244) and third-and-2 (357) than there have been in five years. The results were not impressive with both converting under 50 percent -- below average compared to past seasons.

The NFL's obsession with throwing the ball on third-and-2 has always been hard to figure out. The numbers still favor the run for the best chance of conversion, but teams throw it more than two-thirds of the time. Those numbers would be even higher if factoring in the scrambles on pass plays. That's the biggest difference in how quarterbacks run the ball on third-and-1 versus third-and-2.

Since 2009, 72.7 percent of the quarterback runs on third-and-1 are a quarterback sneak. In that same time, 51.1 percent of the quarterback runs on third-and-2 are a scramble after a designed pass play. The sneak has only been used seven times (five conversions) on third-and-2. The extra yard absolutely makes a difference for the sneak, but it's not enough to ignore the regular running game the way teams continue to (increasingly) do every year.

The following table summarizes the types of quarterback runs on third-and-1 since 2009. The "Quarterback dive" is basically the play Cam Newton has mastered in Carolina. Some may call it "QB Power" but you know it instantly when you see him take that direct snap and plow ahead with little resistance. The read-option/zone read is included with the option.

NFL Quarterbacks: 3rd-and-1 Runs Since 2009
Type of run Plays 1st Downs Conv. Rate
Quarterback sneak 317 279 88.01%
Scramble 29 27 93.10%
Quarterback dive 22 20 90.91%
Option/Zone-read 21 14 66.67%
Bootleg 12 11 91.67%
Botched snap 12 0 0.00%
Botched handoff 10 0 0.00%
Sweep 5 4 80.00%
Quarterback draw 4 3 75.00%
Broken play 4 2 50.00%
Total 436 360 82.57%

How can you argue against the quarterback sneak when it's been working 88 percent of the time? That means defenses only stop it 12 percent of the time. If you focus on the 119 non-sneak runs on this table, 18.5 percent of those were stopped by an offense simply botching the snap or handoff. The zone-read plays may increase in time -- even Chad Henne had three keepers on third-and-2 (two failed) -- but sometimes the hesitation during the read can be disastrous. The shortest distance between two points is a straight line, which has as much to do with the sneak's conversion rate as anything else.

In 2013, the sneak was 32-of-35 (91.4 percent) on third-and-1. Really, the shocking number is 35, given the previous four seasons produced 60-83 sneaks. In spite of this dominant effectiveness, usage has actually decreased each year (83, 75, 64, 60 and 35). Meanwhile, quarterback runs on third-and-2 have increased each year, though 37 (2013) is still a relatively small number of plays.

Regular running plays had a good 2013 on third-and-2, converting 66.4 percent of the time. The league did not break 58.5 percent in any of the previous four seasons. Of course, the 128 third-and-2 runs in 2013 were also a five-year low.

In 2013, 29 of the league's 32 teams had more passes than runs on third-and-2. The only head coach who really took advantage of the numbers and went run-heavy on third-and-2 was Chip Kelly. The Eagles had 14 runs to four passes and their conversion rate of 72.2 percent trailed only Denver (75.0 percent). If more teams are going to pick a down-and-distance to hand it off more, this should be the one.

Fourth Down

Banging the drum for NFL teams to do the right thing more often on fourth down is an old tune people are slowly discovering. Let's keep the focus on what teams are calling when they do decide to go for it on fourth-and-short. The quarterback runs are still very effective here, but 2013 was a different story with the worst conversion rate (68.8 percent) and fewest conversions in 14 years:

4th-and-1 Quarterback Runs
Year Plays 1st Downs Conv. Rate
2000 30 25 83.33%
2001 44 32 72.73%
2002 42 32 76.19%
2003 44 37 84.09%
2004 35 25 71.43%
2005 51 36 70.59%
2006 36 31 86.11%
2007 50 42 84.00%
2008 39 32 82.05%
2009 55 46 83.64%
2010 44 36 81.82%
2011 30 25 83.33%
2012 34 27 79.41%
2013 32 22 68.75%
Totals 566 448 79.15%

While last season ended a pretty successful seven-year stretch, we are talking about 32 plays. Something very unusual could distort those numbers and we did see something very unusual immediately in Week 1.

I boasted in the first short-yardage article about Tom Brady and the Patriots -- pause right there if you want to flirt with cardiac arrest -- being the gods of the quarterback sneak. Brady had 56 straight conversions in the regular season on these short-yardage runs and was 88-of-91 in his career. One of the main reasons I cited for the incredible rate was he did not have the botched snaps or handoffs that plague other teams on these plays. Sure enough, in the season opener in Buffalo, the Patriots went for it on fourth-and-goal at the Buffalo 1 in the third quarter and Brady botched the center-snap exchange. It did not appear he was going to run the ball himself, but these plays go down as a carry for zero yards and a fumble by the quarterback. Brady's streak was broken and three weeks later in Atlanta he again had a botched snap with his center.

The Patriots had as many botched snaps on fourth-and-1 in 2013 as all teams had in 2009-12 combined. Throw in one more by the Texans and those three plays alone turn an otherwise 75.9 conversion rate into a 68.8 rate. So this year was just a small-sample fluke, but the sneak was held to a 17-of-24 mark. The number of fourth-and-2 sneaks also has been on the decline each year.

Here's the summary of all quarterback runs on fourth-and-1 since 2009:

NFL Quarterbacks: 4th-and-1 Runs Since 2009
Type of run Plays 1st Downs Conv. Rate
Quarterback sneak 159 128 80.50%
Bootleg 12 11 91.67%
Scramble 8 8 100.00%
Quarterback dive 6 5 83.33%
Botched snap 5 0 0.00%
Sweep 2 2 100.00%
Option/Zone-read 1 1 100.00%
Quarterback draw 1 1 100.00%
Botched handoff 1 0 0.00%
Total 195 156 80.00%

For those counting at home, the bootleg is 11-of-12 on both third-and-1 and fourth-and-1 plays. Peyton Manning executed a perfect naked bootleg in Dallas this season, but as he pointed out, that's a play you can only really show every few years. It's not as easy to set up as the sneak, which interestingly has been worse in four of the last five years on fourth-and-1 compared to third-and-1. The 2013 season saw the highest conversion rates on fourth-and-1 passes and fourth-and-1 runs in five years.

[ad placeholder 3]

When two yards are needed, running's ignored even more on fourth down than it is on third down. Passing has been even harder with teams converting just 45.4 percent of their fourth-and-2 pass plays since 2000.

It's heresy to suggest putting the ball in the quarterback's hands is an inadvisable strategy, but when it's a short-yardage play, asking them to run might make more sense than dropping back to throw. When you look at these quarterback conversion rates since 2000 (minimum 20 attempts), it's impossible not to see overall quarterback talent has little-to-no impact on short-yardage rushing success (active players in bold):

NFL Quarterbacks: Short-Yardage Runs Since 2000 (Minimum 20 Carries)
Quarterback Runs 1st Downs Conv. Rate Quarterback Runs 1st Downs Conv. Rate
Tom Brady 95 90 94.74% Jake Plummer 31 26 83.87%
Chad Pennington 33 31 93.94% Matt Hasselbeck 38 31 81.58%
David Garrard 65 60 92.31% Ben Roethlisberger 48 39 81.25%
Kyle Boller 21 19 90.48% Cam Newton 51 41 80.39%
Matt Ryan 26 23 88.46% Steve McNair 49 39 79.59%
Jay Fiedler 34 30 88.24% Michael Vick 61 48 78.69%
Josh Freeman 34 30 88.24% Matt Cassel 27 21 77.78%
Carson Palmer 39 34 87.18% Joe Flacco 34 26 76.47%
Drew Brees 46 40 86.96% Daunte Culpepper 73 55 75.34%
Alex Smith 23 20 86.96% Matt Schaub 32 24 75.00%
Philip Rivers 30 26 86.67% Jay Cutler 20 15 75.00%
Aaron Brooks 44 38 86.36% Kordell Stewart 35 25 71.43%
Aaron Rodgers 29 25 86.21% J.P. Losman 21 15 71.43%
Jeff Garcia 42 36 85.71% Jason Campbell 24 17 70.83%
Brian Griese 27 23 85.19% Mark Brunell 20 14 70.00%
Ryan Fitzpatrick 33 28 84.85% Quincy Carter 20 14 70.00%
David Carr 33 28 84.85% Jon Kitna 23 16 69.57%
Donovan McNabb 58 49 84.48% Mark Sanchez 23 16 69.57%
Brad Johnson 25 21 84.00% Jake Delhomme 23 14 60.87%
Drew Bledsoe 31 26 83.87% AVERAGE 36.4 30.1 82.69%

Did Chad Pennington embody the image of a big, bruising runner? Did Kyle Boller seem like the competent field general capable of knowing when to run? Is Philip Rivers fleet of foot? Even Jake Delhomme (60.9 percent) down at the bottom is still ahead of Marshawn Lynch (54.9 percent) for his career. Logically, no one would prefer Delhomme over "Beast Mode" to convert a short-yardage play, but with the way the positions are placed on the playing field, it's hard for the quarterbacks not to be the best option.

Here is the five-year summary of each short-yardage play type:

Short-Yardage Summary (2009-13)
Type Plays 1st Downs Conv. Rate
Quarterback sneak 484 413 85.33%
Quarterback run (all) 782 617 78.90%
Run (non-QB) 3598 2321 64.51%
Pass 3272 1705 52.11%
Total 8136 5056 62.14%

Conclusion: Fear Is the Mind-Killer

With each season rewriting the record books for passing, it will be interesting to see if passing continues to increase in short-yardage situations. While there are obvious benefits, namely bigger gains, to throwing the ball, the best chance for a short-yardage conversion is a run. The closer to the line of scrimmage the runner is, the better his chances. For some drives, the first down may not be that big of a deal. A deceptive play-action pass to generate a huge gain may be worth the risk more than just getting a yard to convert at your own 25. There are always different strategies at work in the NFL, but if the first down is all a team needs, then not considering using the weapon that is the quarterback is absurd.

The Pittsburgh Steelers have shunned football analytics to this point. Offensive coordinator Todd Haley has only called two quarterback sneaks since 2012 (both successful) out of fear of Ben Roethlisberger being injured. Is there no fear of Roethlisberger being injured when he drops back to pass 500-plus times in a season and at least a third of the time he's facing heavy pressure in and out of the pocket? Some injuries are just unavoidable, freak occurrences, much like the concussion Le'Veon Bell suffered a play after the Steelers failed to capitalize by sneaking this ball into the end zone:

Red (reality) or blue (ignorance is bliss)?

No stats should be necessary to figure out which player is more likely to reach the goal line before the defense based on that look. When it's the fourth quarter and you are down by eight points, pressed by time, worrying about injury should be the last thing on a coach's mind.

The sneak should actually be one of the safer plays in the game since it takes place quickly in a small portion of the field without players able to gain much momentum and deliver a violent, high-impact hit.

The quarterback sneak is the closest thing in scrimmage play to an extra point. They are mostly boring to watch, uneventful, over quickly and almost unfairly hard for the opponent to stop. The league might be able to get rid of the real extra point, but the main thing holding back the most unstoppable play in the game from growing is fearful coaching.


25 comments, Last at 06 Mar 2014, 3:28am

1 Selection Bias

I think a huge problem with studying short yardage calls is selection bias.

Teams only run QBS when they think they are going to work. I think this would apply to a lesser extent to running back runs as well.

I think it doesn't apply to throwing the ball. Obviously a team will throw if they think it is going to work. However, I think teams that teams who believe nothing will work will probably throw the ball more than they will run it. Also, if both the offense and the defense does not believe that the offense can run in short yardage, then they'll both default to playing pass.

A lot of QB sneaks these days are option plays. The QB communicates the play only to the center and makes his decision to sneak or run the called play. So unless the QB is equal to or worse than flipping a coin to decide, it's going to positively bias the results.

Just some rambling food for thought.

3 Re: Selection Bias

In reply to by tuluse

All fair points, but I think the larger point of the article (teams should sneak more) still stands. You're right that teams are probably only using the sneak in the most advantageous situations right now, which contributes to the very high success rate. However, there's probably a slightly less advantageous set of situations where the success rate might be lower but still good (say 70%) where teams are not choosing to sneak. If teams started to use the sneak in those situations, it would drag down the overall success rate on sneaks. However, they would still come out ahead in terms of conversions.

4 Re: Selection Bias

In reply to by tuluse

Selection bias is something I would argue when it comes to converting 4th and short in general (such as Brian Burke's conclusion that going for 4th and 1 is the optimal strategy anywhere on the field), but after watching QB sneaks, my thought is that it's simply a very good strategy regardless of the situation.

One thing I do hate is fans' second-guessing every single 3rd-and-short/4th-and-short strategy when it fails. I saw this a lot this year with Seattle, who for some reason was the worst in the NFL at converting these.

Give it to Lynch up the middle? Way too predictable.
Let him try the outside? But he's a power back.
Let anyone else run it for have Wilson pass it? Why be too cute and go away from your team's strengths?

I think these fans need to play as Denver in the Breaking Madden Super Bowl simulation, and see what brilliant strategy they have for converting a short-yardage situation when the offensive line can't get any push.

19 Re: Selection Bias

In reply to by tuluse

I think you're spot on here.

Teams usually will run a sneak only if (a) they have a QB and line who are capable and (b) the pre-snap alignment is favorable.

If those two are true, then a high success rate is what you would expect. Running sneaks more often might require running them when there is a defender in each A gap, or a nose tackle to out-guess. Those will fail quite often.

I have seen Tom Brady run a "sneak" between the guard and tackle. But that's tricky to execute, since it takes longer to get going forward and risks the QB getting his legs cut out from under him as he moves sideways.

20 Re: Selection Bias

In reply to by tuluse

I think it might be simpler than this. Just looking at the data and the fluctuations, I'd say that 5 years of data is not enough to ascertain any trend at all in the rates of QB runs in short-yardage situations.

2 Re: NFL's Short-Yardage Strategy: Hiding the Sneak?

Cool stuff.

Sorry if this sounds greedy, but I'd like to see DVOA data for these plays in addition to conversion data. That would factor in the big play upside of passing on short yardage. I'm pretty confident the numbers will still point to the QB sneak and running in general as the better play, but it would be a more complete picture.

EDIT: I should add that part of the reason sneaks and runs are successful is likely that teams pass in those situations as much as they do. So it's little bit like stock investing -- the best return is usually a low fee index mutual fund but that would probably cease to be the case if it's all most people ever invested in.

6 Re: NFL's Short-Yardage Strategy: Hiding the Sneak?

I found it interesting that the active QB with the longest career apparently didn't reach the 20-carry threshold. Understandable that coaches wouldn't sneak Peyton post-fusion, but Indy must've thought they had a better idea for 3rd/4th-and-1.

10 Re: NFL's Short-Yardage Strategy: Hiding the Sneak?

Draws involve giving the illusion of a pass. The quarterback will take a half-step back and maybe even pump fake before trying to run. Tackles will drop back to pass block and then try to steer their men to the outside.

In QB Dive, the offensive linemen all blast forward, and the QB (usually Cam Newton) picks a hole and attacks.

21 Re: NFL's Short-Yardage Strategy: Hiding the Sneak?

As a Jets fan, I fondly remember the Sanchez Shotgun Sneak. Send your (ineffectual) receivers (whom he can't throw a pass accurately to anyway) spread out wide with a play (which he can't run effectively) called, look up, if the middle linebacker isn't there, audible into a run through the A gap. Worked every time.

Of course, if the middle linebacker was there, tradition demanded Sanchez throw an intercepion.

11 Re: NFL's Short-Yardage Strategy: Hiding the Sneak?

Big sneak fan and every time it works I wonder why teams don't do it more...My Eagles conceded a couple vs the Saints in the playoffs and I was grinding my teeth. The other factor with a sneak is: if you do it enough, teams will start to crash down hard inside to stop it. That's going to lead to some opportunities for outside runs (like a fake sneak, pitch out type play) , or passes with more open field.

That overhead picture is killer. Shortest route wins. It's logical too. You give the ball to a deep running back to let him gain momentum and give your linemen time to control their blocks - none of those things are particularly important for a 1 yard gain. ie. Linemen don't need to control their blocks, they just need to hold them long enough for 1 yard to be gained.

22 Re: NFL's Short-Yardage Strategy: Hiding the Sneak?

The read should be very simple as well. In short yardage situation, is there a down lineman in both A-gaps? If not, sneak through the unoccupied gap.

In fact, this should have such a high success rate that I would consider any defensive formation on 3rd and 4th and one without a down lineman in both A-gaps to be fundamentaly unsound.

12 Re: NFL's Short-Yardage Strategy: Hiding the Sneak?

The point made by tuluse (#1) is relevant here. The running attempts and the passing attempts are not being made arbitrarily. Offenses that feel like they're not getting any push against the defense are more likely to go to the pass than those that are getting a strong push each down.

Now if defenses were distributed continuously over the entire range of possible strengths against the pass and against the run, then the natural advice here would be that coaches should be rushing more until the success rates were equal for both approaches. After all, if the success rate is 75% vs the run and 65% vs the pass, the run should be used, right? (Dismissing for the moment how it's easier to coach the defense when you know the offense is using a pure strategy instead of a mixed strategy.)

Of course, things might not be that simple. Maybe a given offense can convert 85% on a 4th-and-1 against 16 defenses but only at 45% against the other 15 defenses. If the pass conversion rate is a constant 65% against all defenses, then you'll see a well-coached offense with a 85% run conversion rate and a 65% pass conversion rate. But it doesn't mean that the coach should be running more often.

While this kind of data clustering might seem unlikely, my point here is that the math alone cannot make the argument that running more often is advisable, even if the run conversion rate is higher than the pass conversion rate.

It is, however, suggestive.

13 Re: NFL's Short-Yardage Strategy: Hiding the Sneak?

This article doesn't really take into account that all 3rd/4th and 1's are not equal. Since 4th and 1 can mean anything from 4th and 54 inches to 4th and 1 inch, it doesn't stand to reason that QB sneaks should be called more if they can only reliably net 15 or 20 inches for most teams.

When it's 4th and less than a foot, I agree QB sneak is the optimal play.

It drove me crazy the Carolina didn't use the QB draw with Cam vs. the Niners in playoff when they were at the goal line, that's the single most efficient play I've seen.

16 Re: NFL's Short-Yardage Strategy: Hiding the Sneak?

Maybe it's just me, but I'd really like to see if the physics of this hold true for fullbacks versus traditional running backs. In other words, do fullbacks--despite being slower and less efficient at running in general--do better simply by being closer to the line of scrimmage. Right now you're really just comparing the closest versus the farthest and it brings up some questions like tuluse mentioned.

Essentially what I'm hoping to find is that fullbacks could be real handy here, without the risk factor of injuring your quarterback and at a very cheap price (is there a cheaper non-special teamer in today's NFL?).

17 Re: NFL's Short-Yardage Strategy: Hiding the Sneak?

This year, that's what the Saints did. Brees ran the sneak at times, and others gave it to the FB Jed Collins. IMO, QB sneaks are great at the goal line, where you can extend the ball out for the TD, break the plane, and then pull it back to avoid a fumble. They are also great where it is 3r/4th and "inches", where the distance needed is 1/2 yd or less. Because the line can usually get a small surge, the QB with the ball can more or less "disappear" into the pile, and voila--the 1st down is picked up. If I have a full yd or more, I want a non-QB, non-receiver run or a simple pass. If I have a "mobile QB", I would add in QB draws, zone reads, and other non-sneak options for my QB.
Certainly matchups play a part in these equations, and having an audible so that my QB isn't running a sneak when the defense has decided that 5 DL, each lined up over the OL in front of them, is their call on this crucial play.

Follow-up to this article: could we get a slightly more detailed breakdown of the failures? How many were the result of great defense vs bad offense? Was it a DL, LB, or DB that made that great defensive play? Was it the O-line that failed, or was it the RB who slipped making his cut? Or was it just a bad play call for that situation?
[Note to Sean Payton--TE end-arounds are ALWAYS a bad play call in ANY situation!]

23 Botched Snaps

The sentence: "If you focus on the 119 non-sneak runs on this table, 18.5 percent of those were stopped by an offense simply botching the snap or handoff." implies that the 12 botched snaps counted were all on non-sneak plays. Actually, running a sneak probably makes it more likely to botch a snap since it is usually a silent signal between the QB and center that snaps the ball and changes the play.

24 Shotgun is hard to sneak from

I suspect the increased use of the shotgun has something to do with it.

A traditional behind the center snap can be either pass or sneak. A shotgun snap is not likely a sneak. A team that runs predominantly out of shotgun can no longer 'sneak' the sneak.

Rivers used to run the sneak frequently under NORV. But last year I don't recall even one sneak, and many 'very short' 4th and 1's that ended up punts. The offense also significantly increased the number of shotgun snaps.