I don’t have a clear plan for what this article will turn out to be, but now that we have a decent sample of carries for most running backs, I wanted to take an overview of how players around the league are performing on the ground. So I guess this’ll be part data-dump, part thinking-out-loud as I come across interesting things in that data, and part (almost) mid-season update to the ongoing player evaluations that have become central to my dynasty rankings process.

To start, I want to outline some changes to my thought process in assessing the rushing performance of individual players. Through recent reflection and self-scouting (a never-ending but only intermittently fruitful endeavor), I’ve realized that I sometimes get tunnel vision on specific metrics or concepts or aspects of running back play, particularly with things that I’ve personally developed. Part of that is probably the result of egotism, but I also think it’s a natural product of thinking about and finally stumbling upon a solution for a given evaluative problem, and it’s often been justified (at least in my mind) by demonstrable utility of the metrics or concepts in question. Take Box-Adjusted Efficiency Rating. I still believe that BAE Rating is, at the least, an excellent contextual companion to, and, in many cases, a straight-up superior (especially in its evaluative usefulness) descriptor of rushing performance than raw statistics like yards per carry and even other contextualized measures like Rushing Yards Over Expected, but in the 19 months since I developed the metric (and its counterpart, Relative Success Rate), my preoccupation with the new insights it allowed for caused me to sometimes miss insights that it was blind to but that other statistics may have illuminated. It gave me the tools to properly calibrate expectations for the rushing performances of players like Isaiah Spiller, Jahmyr Gibbs, and Dameon Pierce, but my enthusiasm for those tools caused me to miss the forest for the trees with guys like DeWayne McBride and Isiah Pacheco. While BAE Rating and RSR revamped my evaluative process in a way that has been hugely beneficial on aggregate, implementing them on the fly has also come with growing pains as I continually learn more about their limitations, their blind spots, and the corresponding best practices that should accompany their use.

Because of this realization, I want to do a better job going forward of using BAE Rating and RSR as part of a complete breakfast rather than as a big bowl of Cheerios consumed on its own. Such a lifestyle change is obvious to anyone who isn’t a cereal-eating man-child, but everybody starts somewhere and at least I wasn’t turning the most important meal of the day into a one-component show made up of only, say, Fruity Pebbles or coffee (as would the Speed Score zealots and film-only raw-doggers, respectively).

Another change to my thought process is less of a change and more of a solution to a long-standing problem found (or settled on). The premise underlying much of how I go about evaluating players and analyzing on-field data is that -- especially with college players but also with pros -- everything requires context. That assumption is fundamental to the way that BAE Rating and RSR modify raw statistics, but the analysis of their results -- as well as of the results produced by RYOE metrics and raw statistics themselves -- still benefit from further contextualization. BAE Rating tells us more than raw yards per carry does, but only if we can de-shroud the chasm of context that separates them and thereby observe their relationship (on an individual player level) armed with the things necessary for understanding it. Part of that context is held in raw stats themselves, but it’s also in level of competition, backfield teammate talent and performance, and, surely among other things, offensive line play. The last of those is my focus here.

This isn’t profound, but I decided to aggregate several different measures of offensive line performance and situational context in order to get the most holistic view possible of the relative degree of difficulty of a given running back’s rushing attempts. The stats I used are the following:

Each of these metrics tells you something interesting on its own but is unable to adequately account for all of the context surrounding a running back’s carries (and I realize that some of them -- like yards before contact -- straddle the line between being an “offensive line stat” and a “running back stat”, something that strikes me as a relatively minor complication that gets mostly folded over in the grand scheme of things), and while that’s also true of their use in combination with one another, I think the latter method more completely fills in the gaps and covers the bases (even though it also probably comes with some overlap and double-counting) to give us as good a picture as we could hope to have (or at least as good a picture as I can currently devise) of how well or poorly a given player is set up to succeed on his rushing attempts.

In order to smash all these metrics together into something workable, I compiled numbers for all the teams and running backs in the league in each of them, averaged the percentile rank of all the metrics for each player, and used that final number as what I’m (very creatively) calling Offensive Line and Carry Context Score. OL + CC scores are expressed on a 0-100 scale, with 0 representing the (relatively) worst possible circumstances in which a player could carry the ball, 100 representing the best, and 50 representing situational factors surrounding a player’s rushing attempts that are -- on average -- exactly average. Here’s how those scores shake out for the 69 NFL backs who’ve carried the ball at least 20 times this season (excluding numbers from Thursday’s game):

These aggregated scores are interesting on their own, but I put them together to serve as context for rushing efficiency, so let’s do that.

In much the same way as with the offensive line stats, I’ve aggregated various rushing efficiency metrics to get a multi-lens view of how players are performing on the ground. Those metrics fall into four general categories and are as follows:

You could also divide these metrics into those dealing with per-carry averages (yards per carry, BAE Rating, and RYOE per attempt), those dealing with per-carry rates of success (success rate, Relative Success Rate, and positive RYOE rate), and those dealing with contact (MTF per attempt and YAC per attempt), which is exactly what I’ve done. More specifically, I’ve used players’ percentile ranks in each of these metrics to produce scores (again on a 0-100 scale) that a) represent their relative performance in those three categories, and b) combine to represent their relative overall performance as a rusher so far this season. Here are the overall scores for the 69 backs with 20 or more attempts as of week seven:

I first want to point out that a rushing score that is lower than a player’s OL + CC score does not necessarily reflect that a runner is underperforming his situation. D’Andre Swift’s rushing score of 70.2 is lower than his OL + CC score of 89.1, but he has marks above the 70th percentiles in each of BAE Rating, RSR and positive RYOE rate to go with raw marks above the 85th percentiles in each of yards per carry and success rate. It would be silly then to conclude that he is not doing what he should be doing on a per-carry basis, as even with a high bar set by a situation that is (arguably) the best in the league for a running back, Swift is producing excellent raw efficiency numbers while outperforming most team-relative and player tracking-based expectations within that cushy situation. To get a start on reaching my data dumping quota, here are Swift’s numbers in all the metrics I’ve decided to look at as part of this exercise, as well as his corresponding scores (on a 0-100 scale) in the three categories we talked about earlier:

And for the rest of this piece, let’s hop around to some of the more interesting runners and backfields in the league, starting with Jahmyr Gibbs:

Basically, Gibbs has not been great as a runner so far this season. His raw efficiency is decent, but you’ll remember that his OL + CC score was the eighth-highest among qualifying backs: the Lions’ offensive line’s marks in Adjusted Line Yards, run-blocking rating, and PFF grade are all in the top quartile of NFL teams this season, and Gibbs has enjoyed very light defensive fronts (just 6.41 defenders on average, the lowest among qualifying runners this season and a number higher than those for all but eight backs in the last two years) on his carries. Just as in college, though, he has so far failed to perform up to snuff given those advantageous ball-carrying circumstances, especially in terms of down-to-down consistency. Over three seasons at Georgia Tech and Alabama, Gibbs posted a positive mark in Relative Success Rate only during his freshman year, and finished with a career mark of -3.6% (which lands in the 19th percentile).

The silver lining to Gibbs so far being the same kind of runner that he was in college is two-fold. For one, the thesis behind taking him near the top of rookie drafts was never (or never should have been) based on his skill as a runner. Subscribers to this website will remember that my enthusiasm for Gibbs in that context was based on a tenuous assumption that he could score fantasy points like an RB1 by following the Reggie Bush blueprint: dynamic and high-volume receiving usage that offsets a double lack of rushing volume and efficiency. While the first half of his rookie season has been frustrating in that regard, I think that avenue to justifying his high rookie draft value is still very much in play.

The second part of the silver lining is that we’ve seen Gibbs-ish players run efficiently in the NFL before. Pretty much every running back ahead of Gibbs on the OL + CC score list belongs to a similar archetype as he does as a smaller runner who is best deployed out in space, and other than Kenneth Gainwell, each of them has been a net positive on the ground in 2023. I don’t have any illusions that Gibbs’ usage will suddenly mimic the ways in which De’Von Achane and Raheem Mostert are being unleashed in Miami, but I don’t see why Gibbs can’t be James Cook as a runner (which I suppose is either encouraging or frustrating to consider, depending on where your expectations started with this guy). For whatever it’s worth, Cook also entered the league as a sub-25th-percentile RSR guy.

Let’s talk about Kenneth Walker and Zach Charbonnet:

The above numbers belong to Walker. I haven’t studied the film necessary to build more solid opinions of his physical ability and decision-making skills, but I have seen Walker play, and I think he’s probably the most extreme example of a running back who feels so much better when you watch him than what his efficiency metrics would indicate. There’s probably not another non-Bijan Robinson runner in the league who can pull off the types of dead legs and jump cuts that Walker is capable of, let alone one who can pull them off and then explode away from defenders immediately afterward like Walker can. That unique skill-set is reflected in the Doak Walker Award winner’s high-end numbers versus contact, as he currently sits in the 71st percentile among qualifying backs in YAC per attempt and ranks sixth in the league in MTF per attempt (he trails only Raheem Mostert among backs averaging at least 10 carries per game).

His per-carry results are not as impressive. His averages are fine, but they’re kind of just fine. A 97.0% BAE Rating obviously represents output slightly below a back’s situational baseline, and while 4.13 yards per carry and 0.10 RYOE per attempt are both decent numbers that land in the 60th percentiles, Walker feels like a guy who could outperform per-carry expectations more than players like Joe Mixon, Brian Robinson, or Joshua Kelley would (the three guys who are right ahead of Walker on this season’s RYOE per attempt leaderboard).

As was also true during his rookie year, Walker’s biggest area of weakness as a runner is his down-to-down consistency, with RSR and positive RYOE rate marks that each fall below the 30th percentile. You could argue that this feature of his profile is a necessary evil resulting from the bounce-happy running style that allows him to take advantage of his speed, but (again, with the acknowledgment that I haven’t studied the film myself and would therefore be open to changing my take on either the following points or on the legitimacy of this characterization of Walker’s game in the first place) I’d argue that running style a) isn’t resulting in the kind of big-play production and overall efficiency that it would need to in order to justify itself, and b) could yield better results by incorporating greater discipline and restraint. I think Walker’s elite athleticism allowed him to get away with his jittery and avoidant tendencies to a much greater degree in college -- he finished his career with a 90th-percentile RSR to go with a 90th-percentile BAE Rating -- but while the aesthetic experience of watching Walker play has stayed the same in the NFL, the on-field results have taken a bit of a dip.

In part because of that, I still believe Walker is slightly over-valued in dynasty leagues. My assumption before the season started was that Zach Charbonnet would eat into Walker’s hypothetical ceiling by offering the Seahawks a more competent option on passing downs and a nearly-as-competent option on running downs, and on top of nothing having transpired that would change that assumption, Walker’s failure so far to improve upon what he offered on a per-carry basis as a rookie (despite running behind an offensive line that has taken steps forward in both PFF run-blocking grade and run-block win rate) reinforces it.

Charbonnet playing well also helps:

To be clear, the fact that Charbonnet’s numbers have been better than Walker’s so far this year (though he doesn’t qualify for RYOE numbers) does not mean that I think Charbonnet is better than Walker or destined to supplant him as Seattle’s RB1. I don’t think I’d rule out his being better than Walker as a possibility, but that was possible before the season started and I mostly view his quality performance on relatively low volume as a small-but-growing point in favor of believing he could a) be a good player in the NFL, and b) put a cap on Walker’s opportunity share.

Let’s look at Kyren Williams:

I’ve been pretty vocal in designating Williams as a sell in dynasty throughout the hot stretch of fantasy scoring he’s put together this season, largely because I believe him to be a JAG-level talent, a position grounded in his collegiate profile and substantiated by the poor efficiency he’d posted through five games. In his sixth game, however, Williams legitimately smashed, turning 20 carries into 158 yards and bringing his bottom-barrel RYOE numbers back into respectable territory along the way. As he also boasts solid raw marks and excellent team-relative numbers (though no doubt helped by above-average offensive line and situational factors across the board and a collection of depth running backs in Los Angeles that have offered very little on their own carries, respectively), what are we supposed to think about Williams now?

I don’t think it’s take-lock to say that my opinion on Williams’ talent as a ball-carrier did not shift significantly in light of his week six explosion. The very fact that a sub-10th-percentile RYOE runner could have a single blow-up game that boosts his season-long numbers so extremely is itself a good reminder that a) these are still relatively small samples we’re working with, and b) efficiency measures are best viewed as descriptors of player performance (or, more exactly, player output) and not as predictive forces or constantly-shifting referendums on player talent. It’s of course necessary to draw conclusions about player talent from their on-field results, but acting as if a player is his YPC or RYOE per attempt or BAE Rating mark is short-sighted and reductive.

That said, I’ll need more from Williams to convince me to modify my stance that he’s a replacement-level NFL runner, but it’s clear that the Rams value his other contributions and I anticipate this stretch of productivity earning him a line of credit that he’ll be able to leverage toward a few seasons’ worth of jack-of-all-trades role-playing in the same vein as guys like Devin Singletary and Samaje Perine (in addition to getting most of his role back when he’s able to return to the lineup this season).

Let’s talk about Najee Harris and Jaylen Warren:

As we’ve come to expect, the above numbers from Najee Harris are simply not very good. While a porous Steelers’ offensive line (they rank below average in run-blocking rating and PFF grade and well below average in adjusted line yards, run-block win rate, and hit-at-line percentage) contributes to raw efficiency marks that are mediocre at best, Najee’s team-relative and versus expectation numbers speak to the fact that, for a second straight season, he’s not making the most of the situation he’s in. The best thing you can say about his performance is that he’s making a lot of defenders miss, but with a YAC per attempt average below the 50th-percentile, his through-contact metrics track with his reputation as a big runner who dances a lot but doesn’t really go anywhere.

All of that makes perfect sense, or at least it would if those numbers were actually Harris’ and not Warren’s. Here are Harris’ marks through six games and 77 carries:

As you can see, the raw marks for Harris and Warren are pretty close, with Najee having a slight advantage in RYOE metrics and a solid advantage when adjusting for the kinds of defensive fronts they’re each seeing on their attempts (Harris has faced 7.08 defenders in the box on his average carry, a 60th-percentile mark, while Warren has faced 6.92, a 43rd-percentile mark). Neither of them are playing great right now, but a) the circumstances surrounding each of their carries have been quite adverse and rank near the bottom of the league, and b) the idea that Warren is simply better than Harris isn’t reflected in their current on-field output.

What these things mean for their respective futures and dynasty values is anybody’s guess, but I’m still of the belief that we’ve never seen Najee in a situation that would allow for a legitimate evaluation of his abilities. Pittsburgh has been awful on offense ever since he was drafted, and the 2022 campaign that saw him fall well behind Warren’s per-carry efficiency -- still the only season of his career in which he’s been less effective than the depth runners behind him -- also saw him suffer a Lisfranc injury in the preseason and play much of the year with a steel plate in his shoe (he averaged 3.22 yards per carry with it and 4.00 yards per carry without it that season). I’m still keeping the light on for Harris in dynasty because I think there’s a greater chance that he’s good at football than the market indicates and than general sentiment would suggest, but it’s also true that he’s still stuck with the Steelers for another year and may have had league-wide opinion of him irreparably (and I’d argue unfairly) damaged by the poor on-field results he’s produced thus far in his career.

On the flip side, I’m relatively unenthused with Warren as a dynasty asset. Credit to him for carving out an NFL role after going undrafted last offseason, but he’s also signed with Pittsburgh through next year and simply isn’t playing well enough right now to warrant his status as a top-36 running back valued above players like Kendre Miller, Tyler Allgeier, and Zack Moss who either have higher-upside talent profiles or whose short-term situations are much more conducive to contingent-scenario fantasy production, or both.

Let’s look at Tony Pollard:

While he’s clearly outdoing the Cowboys’ breather backs on the ground, Pollard has taken a huge step back in overall efficiency in his new role as RB1. I speculated this offseason that such a decrease could take place under the weight of his newly heavy volume, but I didn’t anticipate sub-four yards per carry.

On top of the difficulty that comes from simply handling more work, I think part of Pollard’s dip in per-play output is a shift in the ways he’s been used as a runner. As a former wide receiver with a relatively thin running back frame at just under six feet and 210 pounds, Pollard was wisely deployed on a lot of zone and outside carries against mostly light boxes during his first four years in the league, and this season has come with substantial changes in each of those factors:

With more inside gap stuff into heavier defensive fronts, Pollard’s through-contact numbers have suffered greatly, as his previous career average of 0.23 missed tackles forced per attempt -- a 75th-percentile mark -- has taken a nosedive into the 21st percentile. There’s a chance that he’s just ill-suited for the kinds of things he’s being asked to do as a runner right now, a possibility that seems likely considering the Cowboys’ offensive line isn’t playing badly: they rank top-ten in the league in each of PFF run-blocking grade, run-block win rate, and hit-at-line percentage. Pollard is currently on the franchise tag and therefore isn’t tied to Dallas beyond this season, and I would anticipate a drastic change in situation for him entering 2024, whether that means returning to his current backfield alongside the addition of a more traditional inside runner or jumping ship to greener pastures on a team that won’t treat him like a traditional inside runner. For now, we’ll simply have to hope that his awful touchdown luck improves (other than Joe Mixon at 0-for-11, Pollard’s 11.8% touchdown rate on 17 carries within the 10-yard line is the worst among backs with double-digit attempts in that area of the field).

Lastly, let’s take a look at another guy who has stepped into increased opportunity this season. Here are Jerome Ford’s to-date efficiency numbers:

While Nick Chubb’s presence in the sample of carries from other Browns’ backs keeps Ford’s team-relative marks subdued a bit, it’s very clear that we’re looking at a boom/bust runner of the highest order here. According to PFF, Ford has paired his basement-dwelling consistency numbers with extreme output in the open field, as he’s gained a larger percentage of his total yardage on breakaway runs than any non-Breece Hall and non-De’Von Achane runner in the league (among those with at least 20 attempts), and considering he entered the league with a 118.0% BAE Rating to go with a -5.2% RSR, such a feature of Ford’s game has been present since he was in college.

While it’s not necessarily good to make huge portions of your overall impact on infrequent big plays, it is impressive that Ford has experienced any success given the adverse circumstances in which he’s been asked to run the ball. The Cleveland offensive line ranks below the 40th percentile in each of PFF grade, run-block win rate, run-blocking rating, hit-at-line percentage, and adjusted line yards, but despite all that and the fact that Ford has run into defensive fronts more packed than those faced by all but seven qualifying backs this season, his combination of elusiveness and burst have allowed him to gain impressive amounts of yardage both before and after contact. I think his limitations as a down-to-down producer will keep him from settling into a long-term starting job, but it’s clear that Ford can play in this league. He’s the kind of guy people were hoping Israel Abanikanda could be this offseason, and it would be a lot of fun to see him in a Baltimore-type environment with wide open running lanes.