The Best (and Worst) Fourth-and-1 Play Calls

(Ed. Note: This piece was finished and submitted several hours before Thursday Night Football, and we have the emails and timestamps to prove it.)

NFL Week 5 - The NFL fourth-and-1 conversion rate is 70%.

If you take one fact away from Football Outsiders this week, make it that one: when a team goes for it on fourth-and-1, in non-goal-line situations, they make it seven times out of 10. The actual number is 69.45%—648 conversions on 933 attempts, from 2018 through Week 3—but there's a lot of imprecision in the data, so let's stick with the colloquial, easy-to-remember rounding for the big takeaway.

Walkthrough is talking about fourth downs today because everyone is always talking about fourth downs in the modern NFL. But we're not talking about the wisdom of fourth-and-1 conversions: they are generally wise, or at least justifiable, in any situation where a coach would dare to attempt one. A 70% chance of retaining possession of the ball speaks for itself in most circumstances. Instead, we're looking at the best and worst plays themselves.

You know the drill. It's fourth-and-1 in a nationally televised game. The conversion attempt fails. Your Twitter timeline instantly and unanimously identifies the reason why: it was obviously the wrong play call! They should not have passed. Or run outside. Or run from the shotgun. Or risked their quarterback's health with that sneak. Silly coach.

Of course, what probably happened is that one of those three times out of 10 where the offense fails to convert just cropped up, for whatever reason (excellent defense, offensive execution error, slippery field, whatever). But it stands to reason that some fourth-and-1 calls naturally have a higher probability of success than others. And that's what we're about to investigate.

A bit of methodology before we continue: all of the figures below are based on fourth-and-1 plays from beyond the offense's 10-yard line. We're separating goal-line plays out because goal-line offense and defense is different from conventional offense and defense. We'll sneak-peek the goal-line data at the end, then revisit it on a deeper level in a future Walkthrough.

I also round to tenths of decimals throughout the rest of this study so you can see small differences, even if they are not necessarily statistically significant. You will understand why once all the 67.3's and 65.7's start bouncing around the screen.

Finally, all data comes from Sports Info Solutions, from the 2018 season through Week 3, giving us a sizable sample to work with.

Fourth-and-1 Rushing vs. Passing

It makes more sense to run than to pass on fourth-and-1. You know it. I know it. And playcallers know it: teams executed a designed running play on 709 of the 933 plays (76%) in our data set, calling just 224 passing plays. And yes, scrambles were teased out of the runs and added to the passes; it was quite a pain in the butt.

The success rate on all designed running plays, including quarterback sneaks, is 71.6%. On all designed passing plays, including scrambles, it's 62.5%. If those rates are surprising at all, it may be how high the passing success rate is.

So the data shows that teams should never, ever pass on fourth-and-1, right? If these were static probabilities on some dice chart, that would be true. But of course these are dynamic probabilities, with that 71.6% success rate dependent on the threat that the offense might pass. Passing on 24% of fourth-and-1 opportunities still seems too high as a tendency-breaker, but there are lots of variables to consider, one of which will become clearer as we sift through the data.

The play-calling rates above appear to reflect rational coaching decisions: teams run the ball in three fourth-and-short situations out of four, but with someone such as Patrick Mahomes at quarterback, it's worth trying something like this once in a while.

I’m in shambles pic.twitter.com/iYqlu4vM3g

— alex (@highlghtheaven) October 3, 2022

Oh c'mon Clyde.

Fourth-and-1 Rushing Plays

So let's move on to a question that came up on Thursday night in Week 4, when Cincinnati Bengals running back Joe Mixon was stuffed on an outside run on fourth-and-short. Is running outside in such situations a bad idea? Let's break down running plays by direction:

Surprise, surprise (/sarcasm): the quarterback sneak has a significantly higher success rate than any other running play, and therefore of any play. Again, coaches know this: more than a quarter of all fourth-and-1 conversion attempts are sneaks.

Of course, a sneak is more likely to be called on "fourth-and-inches" than when the offense needs 4 feet or so. Shorter distances mean a higher conversion rate. Unfortunately there is no way to tease "inches" situations out of the data. Also, as the video linked earlier demonstrates, a sneak does carry a non-zero risk of quarterback injury. Still, the success rate for sneaks is so high that it really should be the default call unless there's a good reason to call something else. It's fourth-and-almost-2 is a good reason. Our offensive coordinator designs such cool plays is not a good reason.

Once the offense chooses to hand off, the direction of the run is not all that significant. When I first collated the data, the outside run rate was way up at 71.9%! That was because 13 scrambles, 12 of them conversions, were lumped into the outside runs. Take those out and the differences between the play types look like nothing more than margin-of-error fluctuations.

Outside runs do come with a higher per-carry average: 5.4 yards per rush on fourth-and-short, compared to 3.2 for off-tackle runs and 2.9 for inside runs. (Sneaks rarely gain more than 1 yard.) That extra yardage may be worth pursuing in some circumstances. Take the Jaguars' James Robinson counter play on fourth-and-1 when leading the Chargers 16-10 in Week 3. A well-designed play results in a 50-yard touchdown in a situation where the Jaguars have a little breathing room from a win probability standpoint. Also, the Jaguars were clearly not in an "inches" situation:

This just in: @Robinson_jamess is good at football.#JAXvsLAC on CBS pic.twitter.com/OGbfWq7Vlg

— Jacksonville Jaguars (@Jaguars) September 25, 2022

No one watching that play said, "Yuck, running outside in short yardage situations. Bad decision, Doug."

Fourth-and-1 Passing Plays

As mentioned earlier, the conversion rate on fourth-and-1 designed passing plays is 62.45%: 140 conversions on 224 attempts, with a total of 13 conversions on 15 scrambles in the data. The fact that the scramble success rate is so high, albeit in a limited sample, may make passing in short-yardage situations more enticing for teams with Mahomes, Lamar Jackson, or Josh Allen at quarterback.

Let's break the passing numbers down a bit. The following categories are not mutually exclusive: a play-action pass can also be a rollout pass, a shot downfield, etc.

The play-action passes come with an asterisk due to the limits of the databases I used: I was able to sort play-action completions on fourth-and-1, but not conversions. There are certainly a few completions for no gain or a loss in the data. What's certain is that play-action passing is a common short-yardage tactic, as it should be in a situation when everyone is expecting a running play.

It turns out that all of our "wrinkles" have a lower success rate than the overall passing success rate! That's because the best passing play on fourth-and-1 turns out to be a short pass (5 air yards or less) from the pocket. Such plays convert 59 times out of 82, or 71.9% of the time.

Wait: does that mean that short-pocket passing is more effective than rushing? Not precisely, because it's impossible to determine air yards when there's a sack. Just five sacks on what were supposed to be short passes would take the rate down to 70.1%. There are other factors such as penalties to consider: false starts and holds, invisible in our data set, are more common on passes than runs. It's more accurate to say that the success rate of a short pass is roughly equal to that of any handoff, and that both are less reliable than the good ol' quarterback sneak.

One tactic which doesn't work is emptying the backfield. Fortunately, coaches don't do that very often on fourth-and-short.

And then there's the ever-popular Kliff Kingsbury YOLO bomb. A "shot downfield" is generously defined here as a pass of 10-plus air yards. They're rare, but not as rare as empty backfields. And when they work, they work: the 21 first downs listed above netted 451 yards, or 21.4 yards per completion. It's a tactic that's used far less than once per team per season. At that rate, it may be worth the occasional gamble for a coach who either knows he won't get fired or knows he will get fired.

Fourth-and-1 Shotgun Plays

And then there's the ever-unpopular shotgun run on fourth-and-1, as popularized by Justin Fields and the Chicago Bears in Week 2. That was a goal-line play, and again, I will be exploring such plays in depth in a future Walkthrough. I also studied shotgun short-yardage runs briefly a few weeks ago, but I included fourth-and-2 situations in the data. Conversion rates decline swiftly as yards-to-go increase. Here is what we get when we focus exclusively on fourth-and-1 situations:

There were six scrambles, all for first downs, moved from the rushing data to the passing data. There are also eight sacks in the passing data.

Excuse me for a moment while a stare at all the data we have collected so far and cackle like the Joker.

Me, analyzing four years' worth of 4th-and-1 play calling decisions for an upcoming Walkthrough. pic.twitter.com/dNR5G0iCuM

— Mike Tanier (@MikeTanier) October 4, 2022

Conclusions

The overall fourth-and-1 conversion rate is 70%. The non-sneak rushing rate is a tiny bit below 70%. The passing rate, assuming a short pass with few bells and whistles, is roughly 70%. The shotgun rushing rate is essentially 70%. The shotgun passing rate is lower, but the shotgun formation is also used to help disguise rushing plays, making them more effective. There are 10 Baltimore Ravens designed quarterback runs from the shotgun in the data, nine of them for first downs. And if we tease fourth quarters out of the shotgun passing data, taking away meaningless end-of-game drives, the conversion rate creeps up to 37-of-58, or 63.8%, which is right in the ballpark of the overall passing rate.

All of our quibbling and Tweeting and pontificating over what a team should have called only comes down to a percentage point or two of probability! And those percentages cannot be statistically significant when compared to the differences between the strengths of offenses and defenses, injuries, field conditions, and so on. We're just spitballing! Nearly every play call has probabilistic merit! Give it to Derrick Henry! Or use Henry as a decoy for a short pass! Do it from the shotgun! Or under center! Do ANYTHING besides a triple reverse, as long as you execute properly, and the probabilities stay relatively the same!

I entered this study honestly expecting to find more passing probabilities hovering around 50%, outside run conversion rates 5 to 10 percentage points below inside rates, perhaps something meaningful in the shotgun data. Instead, the data insists that sneaks are better than runs are better than passes, but that the pass rates are high enough to be worthwhile tendency-breakers, and that the most significant differences in conversion rates come on plays (such as an empty backfield) that come with such obvious disadvantages that they rarely used. The data shows that coaches, on the aggregate, are behaving rationally. It shows that blunt tools like clumping "inside runs" together cannot capture whatever tactical differences might make one play or another more advantageous; unfortunately, teasing out "I-Formation 22-formation power runs" only results in uselessly tiny sample sizes.

Look, if Zac Taylor calls a Devin Asiasi tight end jet sweep from an empty backfield on fourth-and-1 and loses 6 yards, I am still gonna rip him. But normal stuff? A Mahomes pass? A Henry sweep? Lamar or Josh Allen in shotgun? We should probably admit that there's a logical reason to call such plays, based on game-planning decisions which are a little more subtle than what we can surmise from the couch.

Just make sure to call a sneak if it's fourth-and-inches.

Fourth-and-1 Near the Goal Line

OK, you were promised a sneak peek at some goal-line data. This data set extends from 2018 through all of Week 4 except the Monday night game. We have changed parameters, so changing the pool of games slightly doesn't matter.

We're also examining all fourth-and-1 plays inside the 10-yard line, not just fourth-and-goal from the 1. I want to look at situations where the defense does not need to keep safeties deep or worry about the whole field. Conversions include touchdowns and first downs for the rare occasions a team picks up one on fourth-and-short from the 5-yard line or wherever.

As you might suspect, the short-yardage conversion rate is lower near the goal line, where the defense can commit completely to stopping a 1-yard play. Here are the numbers, with five quarterback scrambles (all for first downs or touchdowns) reclassified as passing plays:

Wait a minute … does this mean that there is only a tiny difference in conversion-rate probability between rushing and passing? That the proper choice of play probably comes down to personnel and schematic subtleties, not simple "coach be dumb" second-guessing? That coaches probably understand these odds, instinctively if not down to the decimal place, and make their decisions accordingly?

Hehe. Hehehe. He he...

Me, analyzing four years' worth of 4th-and-1 play calling decisions for an upcoming Walkthrough. pic.twitter.com/dNR5G0iCuM

— Mike Tanier (@MikeTanier) October 4, 2022