RB talent doesn't matter, but it's the most important thing
Plus mobile QB offenses; free post on macro concepts
I already had an intro idea I really wanted to dig into in today’s Stealing Signals, but I got a great question about some analysis from yesterday’s Part 1 that sparked a longer thought that is becoming a standalone piece today.
Hey Ben! I'm wondering about your analysis of the Baltimore backfield. When JK Dobbins returned from injury, your thesis (which I agreed with) was "this is positive, he looks healthy, his role should grow." Gus Edwards (a guy I get the sense the Ravens love) returns from injury in the same offense with the same backfield competition, gets more than twice as many carries as Dobbins did in his first game back, and your thesis is "I don't like this role, he's a TRAP, I sell this." What's the major difference here? Why can't/won't his role grow in the same way we envisioned Dobbins' growing? Thanks for your help!
This is a really good question, and I should have explained it better yesterday. So here’s what I wrote back, to make sure it didn’t get lost in the comments for anyone else who might have had a similar thought. And then beyond that, I have a whole bunch of other thoughts about how to think about offenses with mobile QBs that this conversation sparked.
TRAP refers specifically to percentage of touches that are HVT. Dobbins has been used as a receiver more, run a lot more routes, so there was some upside for a receiving role that Gus has never shown. On top of that, Dobbins’ prospect pedigree suggested some skill-based efficiency upside, i.e. he might just be a difference-making runner for a stretch, like Saquon or JT or Breece looked and Ken Walker looks.
While RBs without the draft capital, athleticism, and elite college production that goes into this type of profile can still be very productive — which creates the “RBs don’t matter” logic — it’s just simply not as often we see RBs of Gus’s caliber create huge plays consistently, compared to just getting what is there in a good situation for rushing success. The ones that succeed in the face of tougher rushing conditions or elevate good rushing conditions to truly absurd efficiency numbers do tend to be the elite talents — Derrick Henry and Nick Chubb are two more that come to mind, and CMC/Swift/Etienne are all more.
Gus has shown he’s a good NFL RB but there’s that little gap between good and great and part of the upside of Dobbins was the potential he frankly hit on some explosive plays like we had been seeing from Breece, or see from Chubb, or both Seattle RBs, et al. Gus doesn’t really profile as being able to bridge that good to great gap, which is a tough thing to quantify, but in Baltimore’s offense where the receiving is never going to create an Ekeler or Fournette type workload that’s so great for fantasy, I want a guy to have a path to that elite rushing upside.
So it’s a combination of those two things that had me optimistic Dobbins could have a little better weekly range of outcomes despite a situation that isn’t terrible for RBs (Lamar helps all of them in terms of rushing efficiency) but also definitely isn’t ideal (limited targets, so you want the ones that are there, plus the potential for rotation which relates to other RBs but also to a degree relates to Lamar himself, because some weeks the RBs as a whole take a back seat in the rushing attack as Lamar runs 10+ times). So for Dobbins, feeling like 2-3 catches per week and more efficiency upside created a stronger scoring range.
And then I had so much more to add. The title for this post was something I actually wanted to write up in August; I was only half kidding when I tweeted it to my buddy Josh Hermsmeyer.
So much of what I was thinking in relation to this idea is in the answer to that Dobbins and Edwards question above, but I want to emphasize multiple other elements.
The first is this idea that “it’s the most important thing” relates to the market. Player value bakes in expected workload to a very high degree at this stage of fantasy football’s life cycle. Nick Chubb is a player I struggled with tremendously this offseason, because felt his situation hadn’t changed and might have actually improved, but his ADP would push him down to the late third round in high-stakes drafts where just a couple years ago you couldn’t get him to fall out of the second. I said and wrote in a few different places that at those prices, he becomes kind of interesting, because for as much as the arguments that RB is a replaceable position are pretty valid, guys like Chubb are exceptions. There’s no denying he is an elite talent, and I learned this the hard way through years of fading Derrick Henry and questioning my process. I’m definitely stubborn, but long-time readers will recall that early last year I was more accepting of my fate on the Henry miss than I had been in past seasons, where I was much more defiant that he couldn’t keep doing exactly what he kept doing.
In today’s landscape, players like Joe Mixon who are known inefficient players — we have a lot of data this guy does not fit into the bucket of players I wrote about above, and the unfortunate reality with RB is when you aren’t in that bucket, you’re basically just replacement level. There are backs who do some things well, and Mixon I think is a little more of a natural pass-catcher than the team’s usage of him has ever reflected, but those nuances — while very interesting to me as an NFL fan — are less important for fantasy. As far as Mixon is concerned for fantasy, the expectation is he will go as far as his workload takes him, and not much further. My Stealing Bananas podcast cohost Shawn Siegele does a great job describing this using RotoViz’s Expected Points (EP) and Fantasy Points Over Expected (FPOE) metrics. As he often says, the elite RB seasons almost require 3-4 FPOE per game from the RBs — they have to have these really strong EP numbers that reach up near 20 expected points per game (which is close to the cap you’ll see), and then they also have to add to that workload.
Guys like Chubb don’t get the work that reaches up to 20 expected points per game, but his efficiency is consistently well above the baseline, and the extent to where that can reach is, if not limitless, a little hard to quantify. He’s just so good. So is/was Derrick Henry. When these guys turn low-value rush attempts into 40-yard touchdowns like it’s nothing, that’s just a ton of FPOE on the efficiency scale, because that carry has a very low Expected Points figure. This is semantics, but it’s important to understand how these statistics are compiled to understand how much to regress them or trust them. It’s something that I think a lot of the analysis out there doesn’t get detailed enough to understand.
To get back to Mixon, his massive green zone role is driven by his own inefficiency, as its that old conundrum that because he gets stopped, the Bengals run another play in close, and it’s what I referenced when he had that game where he had at least two green zone touches on three separate drives; in all of those cases, if Mixon could have converted his first try — and sure, not all of that is on him because of offensive line play and those factors, but it still becomes a major difference between him statistically and someone like Chubb or Austin Ekeler — then Mixon wouldn’t have gotten more tries and more EP and the way the market analyzed him would be different. I reference Chubb and Ekeler because their “EP profiles” are hurt by them scoring from distance compared to a hypothetical scenario where they got tackled inside the 5-yard line and then converted on the next play. In that scenario, their EP would spike with that extra really high-value touch, and their EP would look more stable and trustable and their efficiency (FPOE) wouldn’t look so in need of regression. And that’s all created by them being better than expected. It’s a faulty way we apply these stats.
But as I digress a little, I want to circle back to this idea that the market understands High-Value Touches these days, and RB value in the fantasy football landscape pretty comfortably understands Expected Points and understands efficiency and understands the concept of regression, which I’ve argued is now so overplayed as a form of analysis that it is in some cases directly antithetical to the actionable approach. The current fantasy football landscape of 2022 creates room to buy efficiency, and elite talents, because the prices don’t reflect the potential of these really good players to just continue being really good, and basically breaking these opportunity/efficiency models. I think I first argued this all throughout the 2020 offseason as it related to A.J. Brown and his massive rookie-season efficiency. I made comp groups that discussed how his specific type of efficiency hit on multiple elements — his rookie season put him in a group of like 15ish WRs in the air yards era that an aDOT as deep as his was but also generated a ton of YAC (yards after the catch), which is notable because downfield targets tend to create less YAC. The only other cutoff for that comp group was a target floor of like 75 targets, to make sure these were higher-volume players, and the comp group was almost entirely comprised of elite talents like Julio Jones, Calvin Johnson, and Tyreek Hill, and it also featured a lot of those players across multiple seasons, arguing that this particular skillset was a uniquely fantastic blend of what makes WRs great — an ability to earn volume, down the field, and still create YAC. I believed that blend made it less likely Brown’s efficiency would regress, at least unless his volume spiked dramatically in a way that hurt his efficiency but wouldn’t be bad for fantasy. I argued that offseason very specifically that he had a much safer floor than people realized, while others were arguing his ceiling was not very strong (my take on his ceiling is it was virtually uncapped in a way we couldn’t comprehend, but I’ll save that for the actual intro of this week’s Stealing Signals because that’s all about team-level chaos within a season and how that alters our season-long expectations as defined by offseason projections).
As I love to joke, all roads lead back to A.J. Brown, but that additional tangent I just went on I hope is easily to relate back to the broader RB point here about fading this concept of efficiency regression and understanding player talent. It’s extremely difficult to figure out for RBs — we know this! It’s the whole idea behind “RBs don’t matter,” which is not something I’m arguing isn’t rooted in some legitimate data/logic — but that difficulty makes it all the less likely to be parsed by the market. As fantasy football research continues to get more advanced, and frankly sharper, and ADP evolves in ways that makes the game more difficult to be played, this ability to parse RB talent becomes a challenge more worthwhile to undertake. To be clear, if you’re just in a casual league where people aren’t even paying attention to regression, and trades just follow fantasy points, this might not apply. But based on the comments I get, I think most of you are in leagues where at least a leaguemate or two is pretty aware.
And everything I said in the quoted section above about RB talent and how it manifests for fantasy matters. Workload still matters, and a guy like Ekeler is an example of someone who didn’t have an exceptional prospect profile. I’m tempted to argue Ekeler is so elite because of his specific workload profile — how perfectly designed it is to be a HVT god — but that undersells just how good he is. He’s efficient! And one thing that’s worth noting here is part of why RBs don’t matter gained steam is that RB life cycles are short, so guys are efficient until they aren’t — Ezekiel Elliott and maybe even Derrick Henry would like a word here — and you have to understand how reliant they are on that efficiency for fantasy production, and what type of a range of outcomes that manifests as, to understand what an efficiency dip might mean for their specific profile. Ekeler has been efficient, and he has continued to be efficient this year. One thing we do know about the late-round RBs that hit is they tended to be very productive in college, and some of the ones who come from smaller schools can really surprise. That’s a piece of RotoViz research from way back in the day, and it was used to argue favorably about I believe Zach Zenner of all people, but it was built on the production of one Danny Woodhead who was the overall RB3 in that 2015 season I’m always talking about as the RB-pocalypse where Devonta Freeman was the RB1. Anyway, Ekeler is another small school guy perhaps overlooked by NFL scouts but we know he’s an elite athlete — an absolute workout warrior — and he fits into some of the buckets that might make him an exception to the rule.
Then there are the productive later-round RBs that probably don’t have the elite ceilings, because they can’t elevate their situations to that degree. Cost is always relevant, so in scenarios like dynasty formats, it makes sense to be targeting the sleeper profiles there. And with that, it’s usually a surprising production profile that sticks out that captures that player’s hidden ability — that was the case for Myles Gaskin and James Robinson (who also fits the small-school, perhaps overlooked bucket that Ekeler is in) and a guy like Eno Benjamin. This is also why Shawn always makes the point to compare Benjamin’s profile to Clyde Edwards-Helaire’s. It’s a bit of two-way analysis — it argues CEH is not actually the type of RB we would expect to live up to his Round 1 billing and be a guy, from an athletic or profile perspective, that could consistently outperform his circumstances, and that’s a negative note for CEH at the expectation levels his draft capital carries; it also argues Benjamin, at far lower expectation levels, is a candidate to be a surprising Round 7 pick who winds up being a future fantasy producer. Shawn’s been making this point since the offseason these guys were drafted, in 2020, and it has proven to be extremely accurate.
To wrap this point up, the argument here is not to only target RB talent. It’s that RB talent is undervalued by the market, and there exists opportunities to benefit from that that still require us to understand the RB’s opportunity and usage profile, and how the various probabilities might play out. And this gets to the whole other point this question sparked in my mind about offenses run by mobile QBs.
Baltimore’s offense is the prime example of one where on a significant chunk of the plays, the ball never leaves the quarterback’s hands. Lamar Jackson is an incredible rusher, and he can also take sacks at a bit higher clip than average — not as bad as some mobile QBs — because that’s a trait that comes with the territory when your QB is able to move in the pocket and extend plays. Sometimes, they don’t get out of the pocket. It’s more or less that simple.
But between the rush volume and the potential for sacks, we’re talking about a 10%-20% tax on the number of plays these offenses run that won’t even get to a skill position player. This is super evident to me in researching Stealing Signals over the years. The average team runs between 60 and 65 plays per game; a mobile QB who rushes just six times, which mobile QBs do pretty frequently, is taking 10% of the total play volume on those snaps. If the QB rushes, say, 8 or 9 times and takes 3 or 4 sacks, that’s closer to 20% of the total play volume.
Then there’s the element where rushing keeps the clock moving and ultimately shortens games and reduces play volume. Even when mobile QBs are in offenses that call passes at a reasonable rate, those pass calls that turn into scrambles or sacks work the same way as rushing attempts in terms of how it impacts a running clock. So when you get a mobile QB in an offense that wants to be run-heavy, you get the Bears and the Falcons this year. Their style is already limiting total play volume, then those QBs are taking a significant tax off the top of an already limited number of plays via those rush attempts and sacks. The Falcons only ran 45 plays this week — that’s been in their range of outcomes, as it’s been in the Bears’, although Chicago had a really good game that led to a season-high 70 plays for them.
So we had both ends of the spectrum in Week 7. In Atlanta’s game, Marcus Mariota rushed six times and took three sacks, accounting for 20% of their 45 plays. That’s how you wind up with such low pass volume overall, which is partly the high number of called rush plays but also just the fact that Atlanta is capable of running an absurdly low number of plays where the ball even leaves the quarterback’s hands.
Chicago was on the high end, right? But Justin Fields still ran 14 times and took four sacks! Now that’s an extreme outcome, but 18 of those 70 plays got wiped out for the skill position players (at least in terms of how the volume shows up in the box score, but obviously there are elements here like Fields rushing so much because his skill players aren’t getting open). There was still production in Chicago’s offense this week, but honestly not a ton — the RBs were both good, but the passing game continued to be light on overall volume with just 21 pass attempts.
The point is the whole range of outcomes is different in an offense with a mobile QB. That was frankly part of my concern with the Ravens this year, as it related to someone like Mark Andrews, and we saw in Week 7 how that can manifest in a game where the Ravens actually ran 63 plays — one off their season high — but were extremely run heavy, and we get back to the various ranges of how PROEs can look when you have a mobile QB, and those ranges extend very far to the negative because of designed QB runs. Anyway, Lamar ran 10 times and took three sacks, which again was a 20% tax off the top of the total play volume.
Now Andrews is a great example of how this isn’t the end all, be all. A 10% or 20% tax off the top is significant, but Andrews is an elite talent who earns volume at an elite level and has multiple years of displayed elite efficiency. But it’s kind of funny to look at his Week 7 and see basically Kyle Pitts’ whole season, which is to say that the way their run rate led to just 16 pass attempts made the margins a helluva lot thinner even for Andrews. This is why I get so annoyed when people say Pitts sucks. There’s almost nothing that suggests Pitts has done anything wrong, and his season is explainable nearly 100% by the factors I’m describing here about how mobile QB offenses keep the clock running, can have a lower range of total play volume, and when that offense also chooses to call run plays at an extremely high rate, the low end of that play volume spectrum — plus the mobile QB tax on the play volume where 10% or 20% of the plays won’t leave the QB’s hand — creates an absolutely disastrous situation for the skill position players to do anything. It just sucks for fantasy football, there’s no other way to put that.
Luckily for all of us, Baltimore hasn’t been that lame this year. And to circle back to the Dobbins point, there’s also the element I talk about a lot with respect to running backs in mobile QB offenses where those QBs that extend plays in the pocket don’t wind up throwing to the RB underneath at the same rate. This is reflected in the data. So we don’t get that HVT profile we like to see, but the RBs do get positively influenced in terms of rushing efficiency by the way defenses have to play a guy like Lamar, who can obviously pull the ball on any inside handoff and created an explosive play around the edge (or in some other offenses like with the Giants, Daniel Jones can pull the ball on a power play where everything is crashing to one side and roll out for a naked bootleg to the opposite side for a big, uncontested play, which forces the backside of those defenses in the future to have to respect that and not be able to crash the backside of similar run designs and force the RB to pick his running lane that much quicker; the point is, there are multiple ways mobile QBs make it harder for defenses to commit numbers to stopping the running back that rushing game).
Compare all these notes to the range of outcomes for skill position opportunity in an offense run by someone like Joe Flacco. Flacco was barely trying, getting the ball out quickly and often underneath. More passes means more clock stoppages on incompletions and also on completed passes near the sidelines that go out of bounds. Flacco refuses to get hit, as well, and will throw the ball as quick as possible to avoid it, so you don’t get a lot of sacks, and you wind up with outcomes where he’s throwing 50+ passes in the right game script, and the team is running 70, 80 plays. The sheer number of targets is massive, the number of available RB targets is massive, etc. The efficiency of the whole offense might not be there, but the opportunity is great, and there’s just a massive gap between how many targets and overall touches are available for the skill position players in a game quarterbacked by Joe Flacco versus some of these mobile QBs. No other way to put it.
So again, to try to circle back to Dobbins and Edwards to land this plane, in the Ravens’ offense where some weeks Lamar is going to be the leading rusher, and there’s probably also a bit of a RB split no matter what, the hope for Dobbins was he could be healthy and productive enough to kind of force the other RBs into more of a backup role, where Dobbins was playing 60% of the snaps or more (this is the significance of Edwards only playing 36% Sunday, which I wrote about yesterday). The other hope was with Dobbins having run more routes historically — and Edwards having basically no receiving profile whatsoever throughout his career — that for Dobbins there’s this potential to average 2-3 catches per game, which isn’t massive, but when you add in 20-30 receiving yards you’re talking about this 5-point floor that is really nice before getting into his rushing opportunity and efficiency. That likely won’t be there for Edwards (this again gets back to the snaps; Edwards has never been a guy to play on many passing downs, but I’m not saying that’s impossible, to be clear — it’s just what I’m thinking and seeing). But if Edwards doesn’t have that potential floor, he’s now almost entirely dependent on the rushing work.
And then there’s this discussion of RB talent. And as I wrote above, Edwards has been very good, but the hope with Dobbins — a guy who was universally thought of in that same 2020 class as CEH and JT and Swift as being ahead of CEH and being right there with JT and Swift — is the potential for the number of big plays we’ve seen out of the really elite RBs like the Chubbs and the Swifts and even the guys like Etienne over the past few weeks, and it’s a big reason I’ve been so glowing in my Etienne praise as that’s built (along with, of course, Etienne’s prospect profile from college, which suggested the potential for another level of talent than the average RB has, and especially than what Robinson looked like he was physically capable of giving this year, which sometimes the guys we think are “good” get blocked by competent backs that aren’t losing the job, so this is another element to consider and why I kept emphasizing that there was potential here for Robinson to play himself out of the job and create the opportunity for Etienne to capitalize; that was a key part of that situation).
Alright, that’s all for this absurdly long secondary introduction to this week’s Stealing Signals. Hopefully it helps explain some of the process I’ve been going through lately, and how my specific analyses keep developing. I’m sending this one out without a paywall as it’s a good example of how I try to write about what I’m thinking that I described in this piece on Tacit Knowledge. I try to take the time to explain the “why” behind my analyses when possible, and this was a fun morning to really get into some concepts that might have led to confusion in a few micro spots as I’ve written about different teams this year.
Stealing Signals Part 2 will hit your inbox later tonight.