It's true that RB is not a premium position, and IMO not a good use of a top 10 pick, but a lot of these studies are weaker than they sound.
NFL offense: Running for three yards is like going backwards.
Then in 2014 Dr. Ed Feng looked at 10 years of data that showed that rushing efficiency contributed to only ~4.4% of the variance in wins, vs the passing game contributing ~62%
One of the classic bad arguments in favor of running the ball is to point to the data about how teams usually win games where they run the ball a lot. This gets the causality backwards - in reality winning leads to running, running doesn't lead to winning. Teams with a lead run the ball a lot to grind out the clock. It's not that having lots of rushing attempts is incredibly helpful for winning a game, rather, having a big lead is incredibly helpful for winning a game, and teams with a big lead tend to run the ball a lot.
Ed Feng's study as the same problem (though a milder version) except in it makes running look less important than it is. As some of the comments to his post point out, grind-out-the-clock running plays are generally inefficient - they get low yards per carry. So having a late lead leads to a bunch of inefficient running plays. Rushing efficiency helps a team win, but also teams that win also have a bunch of low-efficiency runs late in the game, which (if you just look at the overall relationship between rushing efficiency & winning, like Ed Feng did) will partially cancel out or hide that benefit. Passing efficiency is a lot more helpful than rushing efficiency, but this study exaggerates the size of the difference.
https://thepowerrank.com/2014/01/10/which-nfl-teams-make-and-win-in-the-playoffs/
How passing and rushing affect winning in the NFL
Around 2017-2018 the data really cranked up on RBs themselves.
Josh Hermsmeyer looked at 10 years of rushing data, and found that two factors accounted for 96%(!) of rushing gains...
Those two factors? Field position, and # of defenders in the box...
Josh Hermsmeyer's study only looked at the average relationship between field position, # of defenders in the box, and yards per carry. He took every rushing play and categorized it into 30 bundles, based on field position (in 10-yard increments) and men in the box (6-, 7, or 8+). For example, one of those 30 bundles was: on average, how many YPC did teams get when they ran the ball from their own 30-39 yard-line against 7 men in the box? There wasn't any chance for other variables to matter, because he had thrown them all away by averaging those plays together. If a good offensive line, or certain play designs, or whatever else leads to higher YPC, it wouldn't show up here, because running plays behind good offensive lines are just averaged together with running plays behind bad offensive lines before he tries to fit his model.
So the 96% R^2 isn't that meaningful. When you look at how YPC varies based on field position and # of box defenders, almost all of that variation is explained by field position and # of box defenders. Why is it just 96% instead of 100%? It looks like he fit a relatively simple model using field position and box count, so he didn't have enough degrees of freedom to perfectly fit how YPC has related to field position and box count. But YPC mostly does relate to field position & box count in a pretty simple way, so his simple model captures most of that.
Add that to the study by Ben Baldwin showing that RBs drafted in the top-20 are no better than the league average on a yards per carry basis.
There's also a nugget in here about RBs having the highest 1st round bust rate of all offensive positions
In 2008, Malcolm Gladwell made a big deal out of a study which had found that quarterbacks' passing efficiency had little relationship to when they were drafted. Team are terrible at evaluating quarterbacks, Gladwell argued, and he speculated about what biases they might have that made them so terrible. The details of the study ("Catching a Draft" by Berri & Simmons) are kinda complicated, but you can see similar results with simpler methods, e.g. in 2022, first round quarterbacks threw for 7.10 yards per attempt while all other quarterbacks averaged 6.94 YPA.
So do quarterbacks not matter? (Or barely matter, at 0.16 yards per attempt of value?) Is drafting a quarterback like randomly picking names from a hat? Of course not. There's a huge selection effect here - most late round quarterbacks don't see the field. Tom Brady, Dak Prescott, Jalen Hurts, etc., really bring up the average of non-first-round quarterbacks. Todd Husak, Cardale Jones, Jacob Eason, etc. don't bring down the average of late round quarterbacks because they aren't playing. It's possible to get a Dak Prescott caliber quarterback in the fourth round (Dallas did), but it takes an awful lot of 4th round picks to find one. It takes fewer top 10 picks to find a quarterback that good. The distribution
And it's similar with RBs, where Aaron Jones & Chris Carson are bringing up the average for day 3 RBs while TJ Logan and Matthew Dayes aren't doing much to bring the average down. It's *possible* to find an Aaron Jones in round 5, but a round 1 or 2 pick is much more likely to bring in a RB of that caliber. There's a huge selection effect, where the YPC of later RB picks is mostly the YPC of the guys who were good enough to earn a bunch of carries.
I'd still hope to see first round RBs have *some* efficiency advantage over other RBs, if I was thinking of drafting one early, even if this selection effect means that we shouldn't expect that big a difference. Ben Baldwin's data shows no difference. But there are a few issues with the study that could make it misleading. One is that it's looking at a pretty small sample of first round RBs - just 17 of them - which is not necessarily a big enough sample to reliably detect differences (especially if we're looking for something like the 0.16 yards per attempt difference that we saw with QBs, rather than a huge difference). Another is that YPC is not that great a rushing efficiency stat, e.g. getting short yardage or goal-line carries tends to pull down a RB's yards per carry, and maybe the first rounders got more of those. Also, when Ben Baldwin compared 1st round RBs to league average rushing efficiency, I'm not sure if he was just looking at the average of other RBs or if that average included all players - QBs and WRs tend to get higher YPC (I clicked through to see if I could check this, but it looks like a PFR site redesign broke the links to the data he was using).
Since we know with these studies (and others) that running backs have very little influence on the success of a running game, and that running game success has little impact on winning games, the argument moves to the passing game.
Does a RB matter if he can be a pass catcher?
Well, in this study, Eric Eager (again) shows that passes thrown to RBs are the less efficient, less valuable, and less stable year over year, than targets to any other position.
Ben Baldwin confirmed that study in this one, where his data showed that targets to RBs have about 1/4 of the EPA per play of targets to TE, and even less vs targets to WR, and a lower success rate (positive EPA) than targets to TE or WRs.
Where the pass gets throw depends on what happens during a play. A lot of passes to RBs are checkdowns, which are an attempt to salvage some positive value after the main thing that the play was trying to didn't work. It's not the RB's fault that the main play design didn't work.
Similarly, down-the-field throws to a WR have a lot of value on average, but those throws only happen if the protection held up long enough for the QB to make the throw, and the receiver got open enough for the QB to decide to throw it; if the playcall leads to a sack or a scramble or a throwaway or a dumpoff to a RB or TE then it won't affect the average result of throws down the field to a WR even though it should definitely count towards the average result of playcalls that are designed to get the ball down the field to a WR.
So the average value of passes to RBs being low doesn't tell us much, one way or the other, about how much value there is in having a RB who is useful in the passing game.
