Running Backs and Balloons (Re-Post)

Thinking about RB volume dynamically

This post first appeared as an introduction to Part 2 of my Rookie Draft Walkthrough. As news of possible Free Agent RB additions circulates twitter I decided to re-post this as a *FREE* standalone work with some additional notes, and for easier future reference.

If you haven’t already read this, I hope it helps you think about the RB position more imaginatively and dynamically.

Square Pegs and Square Boxes

Imagine you have a box that is 3 feet wide and 10 feet long. You also have three more boxes that are 5 feet wide and 5 feet long. One is filled with helium gas, one is filled with water, and one is filled with a plastic cube that fits perfectly inside the box.

What would happen if you transferred the contents of each 5x5 box into the 3x10 box?

Well the solid cube would not fit. (Ever heard the one about the square peg and the round hole? )The water would fit in the box and alter its shape while taking up the exact same volume. The helium gas would flow into the new box and expand its volume to fit the size of the new box.

I think this metaphor generally describes how the fantasy industry projects the quarterback, wide receiver, and running back position. Quarterbacks either start or they don’t. They project for roughly 15-25 points per game or zero points per game. In this way the QB is the plastic cube.

For receivers, the “targets are earned” war has mostly been won. While the target competition and pass volume of an offense causes a wide receiver’s projection to ‘change shape,’ they more or less own their target shares regardless of situational change.

When projecting a team such as Philadelphia or San Francisco, the collective strength of the teams’ pass catchers pressurizes the target-earning capacity of each player toward their lower limit, while tertiary, starting members of the offense rarely see the ball. (see: Watkins, Quez) The opposite is true of a team such as the Panthers or Patriots; each player is to some extent artificially boosted beyond their inherent target earning ability due to the dearth of options available, but the primary resulting effect is a diffuse target tree spread across all skill position players. (Think of the 2021 Saints, or post-Cooper Kupp injury 2022 Rams as examples)

While target competition (or lack thereof) alters the pressure enacted upon a receiver’s projection, it shifts their range rather than altering it dramatically. In the world of our metaphor, the receiver position acts as our liquid.

At running back, the industry seems to treat the position - in my opinion too often - as a gas. Every time a team signs a backup-level free agent or drafts a back in the middle rounds you will find people claiming it as an indictment of the incumbent, or racing to proclaim the (often) mediocre talent just added will “take more work than you think.” Inherent in this line of thinking is that if allowed to fill up the entire workload, a running back, like a gas, can fill the space un-inhibited.

This greatly aggravates me because I think it is far too simplistic a way to view running backs, each of which is - as Ben Gretch would say - “its own delicate little flower.”

Thinking About Running Backs Dynamically

Some running backs truly do behave like gasses. Josh Jacobs last year is a great example of a player who - on a team devoid of depth running back talent - was able to subsume all other available touches and perform well on those touches. It was an incredible feat, but a rare one.

But most of the time that’s not the case. You’ve likely read content by now on the idea of the “Running Back Dead Zone.” A more nuanced take on the dead zone shows that it is not a magical death-trap through which otherwise talented running backs fall if drafted between Rounds 3-6. Instead, the dead zone speaks to (or spoke to) a tendency of the fantasy community to over-draft ordinary talents or backs with incomplete skillsets in the early-mid rounds of drafts based on the upward inertia of projectable volume.

Running Backs and Comparative Advantage

I prefer to think about Running Backs in terms of comparative advantage.

Comparative advantage is an economics term related to free trade which I will explain briefly. It posits that trade between countries should be based not on which party can produce the most of a good for the lowest cost, but which country can do so for the most efficient cost relative to the country it is trading with. No matter how efficient a given country is, its resources will eventually be exhausted. Therefore, each should aim to focus on producing the goods it can produce at a comparative advantage to other counties and trade for other goods.

A great example of comparative advantage at work in running back rooms is the Cleveland Browns. Nick Chubb is one of the best runners in the league. He will have a comparative advantage vs. almost any other back in that area. As a receiver, it’s fair to suggest he is at least as good as D’Ernest Johnson. However, when Kareem Hunt has missed games, Johnson has still received the majority of passing down snaps.

Why is that?

Likely, the Browns view Chubb’s effectiveness as finite, or at least feel it has a point of diminishing returns. They would rather utilize Chubb in situations that he provides a significant advantage, and use other backs in situations in which Chubb and the alternative are more interchangeable.

The Bengals are perhaps the best example. Joe Mixon was a better pass catching prospect than Samaje Perine and has been the more efficient receiving back in the NFL. Nonetheless, Perine played most passing downs. The Bengals are one of several teams who have signalled a conviction against using one back in all situations, and are thus allocating touches based on a perceived comparative advantage rather than an actual advantage in hopes of generating the most effective backfield on net.

The backfield I’ve been most puzzled about the reactions to post-draft is the Chicago Bears. I wrote about Khalil Herbert and D’Onta Foreman extensively after the latter’s signing. In that article, I discussed the elite level results that Herbert and Foreman have offered on limited touches in recent years, and applauded the Bears for finding a cost-effective path to generate 350-400 carries of high-value play by focusing on backs with high-end but limited skillsets. I also suggested they may look for a pass-down option in the draft as this is neither Herbert nor Foreman’s specialty.

When I saw the Roschon Johnson pick, that was my first reaction. Johnson is hailed as the best pass protecting back in this class, and a better receiver than either of their current options. He’s also a capable runner who specializes in tackle breaking and has demonstrated success in zone running schemes - traits shared with Herbert and Foreman.

Foreman is on a very cheap deal, and has no guaranteed money beyond this year. It would not shock me if he’s rendered a redundant member of the backfield by mid-season, if Johnson proves to be just as effective on the ground. However, Khalil Herbert ranked 1st in Rush Yards over Expectation per attempt, and 4th in Rushes over expectation percentage in this scheme last year. He’s averaged over three yards after contact per attempt in both his NFL seasons.

His pass blocking is a major deficiency and he offers little as a receiver. Further, we have never seen him receive a full workload on the ground so his point of diminishing returns is currently unknown. But we do know this: On between 100-150 carries in this scheme, Khalil Herbert has provided elite level rushing effectiveness at every opportunity. Given his profile is elite both in terms of net efficiency, and play-to-play consistency (ROE%), there is little reason to perceive him as being a change-of-pace specialist only.

There seems to be a rush to simply hand Roschon Johnson - a 4th round pick - the entirety of the Bears backfield, and to me it makes no logical sense. At the very least, Herbert on a limited share of rushes provides elite value. Johnson has a major comparative advantage on passing downs. Beyond that, the remaining early down rushing is likely to be allocated based on how far Herbert (the player with the most demonstrated success in this scheme) can push his workload up before hitting diminishing returns, with the spillover going to the better of Foreman and Roschon Johnson. By the same token, if you have two highly capable runners and are already dedicating pass-down snaps to Johnson, it would make little sense to push your Round 4 rookie to his breaking point, and instead opting to utilize Foreman and Herbert in the areas they are best.

A similar phenomenon is likely to happen in Philadelphia after the arrival of D’Andre Swift. The Eagles - between Penny, Swift and Gainwell - have done a similar job of building their RB room to the Bears. They’ve embraced comparative advantage to construct one of the most cost efficient assemblies of talent in the league.

Due to Penny’s injury history and limitations as a pass catcher, he cannot be relied on as a catch-all solution to a running back room. But he is still one of the most explosive and efficient runners in the league. Swift is an immensely talented and explosive running back with pass-catching ability, but as we’ve written about, his best results have come when his usage is tailored to his strengths rather than when he’s been used as an all-purpose back. He also has a length injury history. Gainwell has been an effective satellite and change of pace back, but his limited size likely renders him a complimentary piece on any team.

The Eagles do not have the kind of do-it-all back teams spend 1st round picks or 10+ million dollar second contracts on. But for the price of two day three draft picks and a cumulative $4M, they have sufficient running back talent to generate positive value on up to 500 total touches from their running back room. Penny and Swift in particular are great compliments; each has such different strengths that both are hardly limited by the other when both healthy. Gainwell offers the residual talent to be able to compliment either back in either area of the game should one get hurt to avoid diminishing returns from their backfield as a whole.

Running Backs and Diminishing Returns

This all brings me back to Tony Pollard (as so many things do). Pollard was an unequivocal winner of the draft, with Dallas adding just Deuce Vaughn in Round 7. However, there remain concerns about whether this signals Dallas’ desire to bring back Ezekiel Elliott on a reduced salary.

This is not going to be a “Zeke sucks” rant. I just think we need to have a conversation about how large a swath of the range of outcomes for Dallas’s RB room decisions have almost no effect on Tony Pollard vs. how the market perceives them.

Last year Dallas had 524 total running back touches. Tony Pollard took a career high 232, which was enough to finish as the RB8. But his season was really split in three parts. From week 1-6 he operated as the 1B of a tandem backfield alongside Ezekiel Elliott, cresting 50% of snaps just once. Elliott was injured mid-game in week 7, after which Pollard operated as the clear lead back for two games. In one of those, he was granted a true-bell cow level 87% of snaps, 22 carries and 6 targets, In the other, he handled just 15 touches on 53% of snaps (though he scored over 30 fantasy points on them).

Upon Elliott’s return, Pollard took on the 1A role: playing a majority of the snaps in every remaining game aside from a meaningless Week 18 contest, a blow-out win over the Colts in Week 13 in which starters sat the 4th quarter, and the divisional round game he suffered his injury in. From Weeks 11-16 he handled over 17 touches per game, and scored over 20 fantasy points per game. As a 6-0, 209-pound running back, that is close to his maximum over a long sample of games before he would hit a point of diminishing returns. That point would likely hit eventually whether the backfield includes Elliott, any of the mid-round rookie backs they passed on, or their current depth chart.

The point is; a running back addition to the Cowboys (or any other team) does not necessarily impact the touches allocated to Pollard, or any other lead back. We know Pollard has been fantastically efficient on up to 232 touches, to the point where reducing that load would be nonsensical. We can guess that for a back of his size, limited volume throughout his career, and skillset, that the range of 280-320 is a likely firm ceiling (it may be even lower). This leaves over 200 touches for non-Pollard backs.

If the Cowboys had drafted Bijan Robinson, we would have had to massively shift our Pollard projections. Now we have a situation where the point of diminishing returns for each back extends past each other’s and either, or both, will need to be limited below their ceiling workload. If they drafted Roschon Johnson, we should have no such concerns. The same applies to Ezekiel Elliott.

Any substantial changes in Pollard’s projection after an Elliott signing need to explain how many more touches for Elliott you are allocating after he’s fully displaced Malik Davis and Ronald Jones’ currently projected workload, or just how many touches you projected Pollard for in the first place.

In short, Pollard can contend to be top-five RB this year. And it is not contingent upon him receiving hypothetical touches he likely won’t receive whether or not Ezekiel Elliott is re-signed.

Running Backs = Balloons

Tying this all together, I posit that Running Backs are not gasses, liquids or solids in the opening metaphor. Running Backs are balloons. Depending on the air blown into the balloon, they can be wildly different sizes. However, as you blow more air into the balloon the colour beings to lose its richness, and eventually the ballon becomes more and more brittle and unable to resist any pressure applied to it. If you continue past that point it eventually pops entirely.

When placing a ballon from the 5X5 box to the 3X10 box, your ability to fill the whole of the new space depends on the shape of the balloon, and your ability to fit the ballon into the new box at all depends upon whether you have blown too much air into the ballon for it to fit.

This is how I view running backs. Where the ceiling or floor is on any back is ultimately subjective. But we have a general idea.

We talked about Zach Charbonnet in Part 1 but that backfield is worth reflecting on again here. Ken Walker was almost certainly never going to play every passing down for a whole season. He might do it for a game or two. But coaches do not plan to let RBs play in areas that a replacement level option has a comparative advantage in for prolonged stretches of time. By the same token, Walker is one of the most explosive backs in the NFL. For as long as that remains to be true, there is little chance a coach would voluntarily hand the backfield over to someone else at the expense of using Walker’s strengths.

When assessing what running backs to add to our dynasty teams, we have to take the zoomed out view of the skillset they provide to a team, and what kind of role exists for each back in their base case, and what kind of roles they can fill on a contingent basis.

We also have to keep in mind the replacement level. Let’s take Tank Bigsby vs. Roschon Johnson for example. Due to the ambiguity of the backfield in Chicago and the limitations of Herbert and Foreman, Johnson is the surer bet to get on the field, and has a higher chance of being the “lead back” without injury.

But in what scenario can his role rise meaningfully above replacement level production? As mentioned, Herbert has thrived on his touches, as has Foreman. Johnson - a backup behind Bijan Robinson in college - has never played a workhorse role before. While the path to 8-10 points per game from Johnson is fairly plausible, it is conceivable that he would require two injuries to receive a true bell cow role. Beyond that, the Bears offense is a difficult one for running backs because of an uncertain scoring projection, low-throw rate to the back, and Fields competing for goal-line carries.

Bigsby may well be a straight backup in Jacksonville. Or he could not. But to some extent that’s beside the point. We do have a history with Bigsby of handling the majority of a team’s rushing production, as well as playing in all situations. As a third round rookie with inconsistent college efficiency, we should expect Etienne to play clearly ahead of Bigsby. Etienne was highly effective on his carries last year, and the driving force of this backfield split is likely to be based on Etienne’s point of diminishing returns rather than what Bigbsy brings to the table. Nonetheless, as a plausible all-purpose back on a top offense without any entrenched third option, BIgsby’s contingent value should Etienne suffer an injury or struggle is immense.

Between Johnson and Bigsby, I view Tank as the more flexible balloon. Despite starting smaller, his holistic profile and offensive environment suggest greater room to grow. When you combine this with the replacement level (in most leagues), it makes me more inclined to prioritize Bigsby’s contingent value over Johnson’s standalone value.

When Does the Balloon Pop?

Today, free agent Leonard Fournette visited with the Patriots — which for many folks is a substantial mark against Rhamondre Stevenson. I think this is a perfect case to apply our balloons theory of running backs, as well as the liquid theory of receivers.

As presently constructed, the Patriots running back room can afford all the volume Stevenson can handle (and likely much more). This is a team which had a 23.7% team-wide target share to running backs in 2022, and a team whose primary alterations to their pass catching group was substituting Jakobi Meyers and Jonnu Smith for Juju Smith-Schuster and Mike Gesicki; both of whom are lesser target earners on a per-route basis than the players they are replacing.

Among their projected starting three-WR set of DeVante Parker, Juju Smith-Schuster, Tyquan Thornton — and their top-two TEs: Hunter Henry and Mike Gesicki — the average targets per route run between 2021-22 is 0.169. Smith-Schuster is highest among the group at 0.181.

This means that in a vacuum, we would expect the average New England pass catcher to earn a target just one of every six routes they run in a vacuum. As mentioned near the beginning of this article, the Patriots are a classic example of a team in which no player is sufficiently qualified to consolidate volume in their passing game. The typical consequences of an environment such as this is decreased efficiency, decreased passing volume, and flat target distributions.

Naturally, this also means more usage for running backs; both because they rank higher in the play-maker hierarchy of the team, and because primary and secondary reads on passing plays are simply not getting open frequently, leading to more check downs. We saw this last year with an extremely high 23.7% team-level target share to the running back position; with 88 of those targets going to Rhamondre Stevenson.

Stevenson acknowledged feeling over-worked toward the end of the season; an understandable consequence of fielding such a massive pass-game role in addition to being the primary runner in his first season as an NFL lead back. From a real-life football standpoint, it makes perfect sense why the Patriots would want to relieve pressure on Stevenson by adding a veteran insurance policy in Leonard Fournette.

While Fournette has not been an efficient runner in several years, he offers a similar stylistic skillset to Stevenson as a big-bodied runner who can handle a workload and provide pass-catching ability on every down.

The initial reaction of much of the community — if Fournette or any other veteran back signs in New England — will be to immediately start taking touches away from Stevenson. This would be mis-guided.

First, we should think holistically about the Patriots as an offense; a team led by a pocket-passing QB and a bottom-three receiver group in the league. The team-wide reliance on the running back position is much higher here than most. The issue becomes; how do we back-up that necessary reliance with only one proven player at the position? Whether you are approaching this question from a football lens or from a fantasy-projections one, you wind up in the same dilemma.

Adding Fournette would relieve this issue.

Prior to stealing touches from Stevenson, first allocate the touches afforded by default to satellite back Ty Montgomery, and unproven grinder Kevin Harris to the more proven Fournette. Next, consider whether the increase in capable bodies at the position would increase the total number of running back touches available the Patriots; I expect that it would.

In my current projections, I have 137 carries and 74 targets to New England backs not named Stevenson. This is not because Pierre Strong, Ty Montgomery or Kevin Harris are “stealing touches” but because there are only so many touches it is reasonable to bestow upon Stevenson until his metaphorical balloon pops; (I am currently projecting 306).

Stevenson is simply much better at running the football than Fournette at this point of their careers. The former Sooner ranked 2nd in Yards after Contact per Attempt, and top 15 in PFF grade Breakaway Run rate, and RYOE/attempt.

There is no actual advantage or logic in allocating carries from Stevenson to Fournette within an offense and analysis which relies on any narratives leading someone to this conclusion ought to be afforded little weight. In the passing game, Fournette compares much more favourably, although there ought to be plenty of targets available to both backs in the context of this offense.

The most likely result of this potential backfield is that Stevenson operates as the clear-cut lead runner and Fournette is used as a capable supplement to ensure Stevenson does not need to push past his point of diminishing returns.

The total touches available to the running back position in the context of this offense allows for a much smoother entry of another back than would be the case in offenses with superior receivers, a quarterback who monopolizes a significant share of the team’s run-game, or a quarterback worthy of a high pass-first philosophy. But it is likely that if the Patriots were such an offense, they would not feel the need to supplement a back as capable as Stevenson.

I don’t mean to suggest that every NFL transaction is logical, or every backfield is predictable. But we ought to ground our analysis in an approach that is tethered to the context of each offense, the skills of each back, and is dynamic in how we respond to “competition” for touches.

Takeaways

This was a long-winding discussion but I hope it helps you in thinking about how to project running back rooms this year and make choices in your drafts. If you want the “TLDR” version here goes:

• The allocation of running back touches depend less on the amount of bodies in the backfield and much more on the comparative advantages and points of diminishing returns of the backfield’s most talented players, as well as the context of the offense as a whole.

• Most (but not all) split backfields would remain split even if one of the members disappeared because most backs either have limited skillsets or have a ceiling on the amount of touches they can provide positive value on. The ‘1B’ likely did not ‘steal the work.’

• We commonly over-estimate both the potential touch ceiling of most “lead” backs, as well as the impact of depth signings on the end result of a backfield. When a depth option is added to a backfield, they first displace touches allocated to inferior depth options before encroaching on the starter’s role.

• In order to project any backfield, you have to look individually at a player’s strengths and weaknesses to determine their range of outcomes rather than simply assigning volume due to an absence or presence of competition.

• When considering any timeshare or backup running back in drafts, you need to weight not only their base projection, but their contingent value relative to replacement level production in your league

Following this process will certainly not ensure you project every backfield with perfect accuracy. However, it does allow us to find value in backfields in which there is more overlap between the contingent values of each back that cannot show up on a projection. These backfields represent the smaller balloons, resting against one another with their full colour intact, prepared to expand if given the space. It also allows us to better sort between which backs on seemingly wide open depth charts have the skillsets and volume ceilings to meet the high end of their project and expand to all open space, and which are more likely to remain limited. Lastly, it allows us with firmer footing upon which to respond in line with or against the market when adjusting to a backfield which adds a new member.

I hope you enjoyed this piece — whether this is your first or second time reading it. I will be back soon with some projections analysis.