Rushing Props and Game Script in the NFL

A running back does not decide how many times he carries the ball. The depth chart and the scoreboard do. Research a rushing line in that order, then respect how often a single long run decides it.

By SlatelinePublished
A distribution of rushing outcomes with a tall cluster of short gains and a long thin tail to the right

Receivers at least have some say in their volume. They can win on a route and draw the next target. A running back mostly cannot. His carries are handed to him by a depth chart set during the week and by a scoreboard that changes during the game. That makes rushing props one of the most structural markets in football, and it means the most important research question about a rushing line is usually not about the runner at all.

Split the line the same way the receiving piece splits a receiving line. Expected rushing yards are carries multiplied by yards per carry. Carries come from two sources, the team's total rushing plays and the back's share of them. Yards per carry is the noisy efficiency term, and its distribution has a shape that catches out anyone who only reads averages.

Carry share is a role, and roles come in pieces

Carry share is a back's rushing attempts divided by his team's. It is one of the quicker stabilizing numbers in football because it is assigned rather than earned play by play. It is also rarely a single role. Many backfields split the work by situation: one back takes early downs, another comes in on obvious passing downs and in hurry up situations, and a third may appear only near the goal line. A back's share of total carries blends those situations, and the blend changes whenever the game spends more or less time in each one.

So look at where the carries come from, not only how many there are. A back whose carries are concentrated on first and second down in neutral situations has a role that grows when his team leads. A back who is mainly the passing down option gains snaps when his team trails, but those snaps produce targets more than carries. Two backs with the same season carry share can react in opposite directions to the same forecast.

Quarterback runs sit inside the team total as well. Designed quarterback runs take carries that would otherwise go to a back. Scrambles are pass plays that turn into runs, so they come from a different budget. When you estimate the team's rushing plays for a back's prop, separate the quarterback's contribution out first, or a mobile quarterback will quietly inflate every back's projected carries.

The spread is a forecast of carries

A point spread is a handicap set so that the stronger side must win by more than a given margin, as the Wikipedia entry on spread betting describes. For prop research its value is different. It is a compressed public forecast of which team is likely to be ahead, and by how much, for how long. Leading teams run to shorten the game. Trailing teams throw to lengthen it. A back's carries are therefore tied to the spread more tightly than almost any other player prop is tied to anything.

The effect is concentrated late in games. Early on, most offenses call plays close to their normal mix. The script takes over in the second half, when the gap on the scoreboard makes one type of play much more valuable. A heavy favorite's lead back can pick up a meaningful block of fourth quarter carries, often against a defense that knows the run is coming, which adds attempts while holding down yards per carry. A heavy underdog's lead back can lose those carries entirely.

One accounting detail distorts team rushing numbers in exactly the games where script matters most. The quarterback kneel is used to run out the clock at the end of a half or to preserve a lead, and it is typically recorded as a rushing attempt for a loss of one or two yards. A team that wins often kneels often, which adds rushing attempts that no back will ever receive. When you estimate a team's expected rushes from its history, strip the kneels out first.

Example: One made up back, three spreads

For illustration, suppose an invented lead back takes about 60 percent of his team's designed runs, and his team usually runs about 26 designed rushing plays in close games. As a small favorite, a reasonable forecast is around 16 carries. As a large favorite, extra fourth quarter work might push the team toward 30 designed runs and him toward 18. As a large underdog, the team might run only 20 times and he might get 12, while a passing down back takes more snaps. At a similar yards per carry, those three forecasts sit roughly 25 yards apart from top to bottom, and the player has not changed at all.

Goal line work adds carries and subtracts average

Carries inside the five yard line cannot gain many yards, because there are not many yards left to gain. A back who owns the goal line job collects short carries by construction, and those carries pull his yards per carry down even when he runs well. The red zone usage piece covers what that role means for touchdowns. For yardage it cuts the other way: goal line carries add attempts and add almost nothing to the yardage total.

This is why raw yards per carry is a poor way to compare backs in different roles. The early down back who also takes goal line work will look less efficient than a change of pace back who only runs from open field. Before comparing efficiency, separate the carries by field position, or at least by whether they came inside the ten. The same logic applies to short yardage carries on third and fourth down, which are designed to gain one or two yards and succeed when they do.

The long run tail decides more lines than it should

Most carries gain a few yards. A small fraction break into the second level and gain a lot. That makes a back's single game rushing total a sum of many small gains and occasionally one very large one, and the large one can account for a big part of the day. The distribution is skewed to the right, so the average sits above the typical game, which is exactly the situation reading a distribution warns about. A projection built on an average can sit above a line while the most common outcome falls short of it.

The tail also makes recent yards per carry nearly useless as evidence. One sixty yard run can lift a back's average for weeks. Treat any yards per carry that swings sharply on a single play as noise, and shrink it hard toward what backs in the same role typically produce. Carries do not have this problem, which is the practical reason volume leads this research.

Tracking data offers a partial fix. The Next Gen Stats data dictionary defines rush yards over expected as a rusher's yards gained compared with the expected rush yards. Because the expectation is built from player tracking rather than from the box score, the idea is to separate what the blocking and the situation gave a back from what he added himself. That is a real improvement over raw yards per carry. It is still built on the same rare long runs, so read it as a direction over a large sample, never as a verdict on a few games, and remember that the expectation is itself a model with its own errors.

The receiving half of a back's day moves the other way

Many backs carry a rushing line and a receiving line at once, and script pushes those in opposite directions. The trailing game that costs a back carries often gives him targets as a checkdown outlet, and the leading game that adds carries takes those targets away. A rushing plus receiving yards prop therefore reacts to script less than either piece does alone, while a rushing only prop reacts more. Know which one you are researching before reading the spread.

  1. Estimate the team's designed runs from recent close games, with kneels and scrambles removed.
  2. Adjust that total for the spread, concentrating the change in the second half.
  3. Apply the back's share by situation: early down, passing down, and goal line separately.
  4. Apply a yards per carry anchored on his role, discounted for goal line and short yardage work, and shrunk hard toward typical values.
  5. Compare the median outcome, not only the average, with the posted line.

Where this breaks down

The spread forecasts the likely script, not the actual one. Games go against the forecast all the time, and when they do the carries follow the real scoreboard, not the pregame expectation. A single fumble, an early injury, or a defensive touchdown can flip the script and take a rushing line with it. A backfield split can also change midweek with no public explanation, and the first sign of it is often the first quarter of the game itself, well after any decision has been made. The research above makes the estimate of carries better. It cannot remove the variance of one football game, and the long run tail means even a well estimated rushing line can go either way on a single play.

References

Follow NFL on Slateline

The public NFL page lists each market with its graded record where one is published and, when a slate is open, the props the free plan opens. No account is needed to look.

Open the NFL page

Not ready to sign up? Get the weekly note.

The Slate

One weekly note: what published, what the graded record shows, what changed. No picks, and nothing is sent until you confirm your address.

Keep researching

Rushing Props and Game Script in the NFL · Slateline