Which Draft-Hopeful Running Backs Produced Like Studs in College?

Football nerds have been discussing the lack of impact that running backs make at the NFL level since the Obama administration. Probably even before that.

Being a football spreadsheet warrior myself, I, too, have had loads of conversations about why running backs are overvalued at the professional level. Over the years, I've written about it, tweeted about it, and argued about it.

And I've also retired from talking about it. So let's move on.

Running backs may not be a big deal in real football, but they matter a hell of a lot in fantasy football. They're one of the key positions in the game about a game, and because true bell-cow backs are so scarce these days, we typically see a host of them fly off fantasy draft boards early and often every August.

That includes rookie backs.

Since the transition from college to the pros isn't as difficult at the running back position, first-year rushers have a chance to make a huge fantasy football impact. A lot of times, you can just take a look at where a player was selected in the NFL Draft, and if a team spent up to get him, then you can feel good about his Year 1 fantasy potential. The fact is, draft capital matters a whole lot at the position for fantasy purposes.

There's more to it, though. And that's where my prospect model comes into play.

The Model

Data can help us increase our probability of choosing the right rookies in our fantasy drafts, taking our analysis beyond just "draft capital". That's what my model is all about: I'm looking at things like a running back's college production, his BMI, his weight-adjusted 40-yard dash time, whether or not he had strong competition in his college backfield, and a handful of other inputs. A score is compiled for a prospect with all this info, and that score is compared to how fantasy running backs performed in the NFL during their first three years as pros, ensuring the model is focusing on the right metrics.

To put this in a more concise way, the model is attempting to predict how a prospect will perform in fantasy football through his first three years in the league. And, to be clear, the predictiveness of the model is far better than draft capital alone.

On the production side of things -- which is what we'll focus on today -- the model looks at how a college running back performed during his best season in three major categories: total touchdown share, reception share, and total yards per team play. Since the model's using historical player information to help find good prospects, that means the majority of top running backs in the NFL -- the top running backs in fantasy football -- were also productive when they were in college within these categories.

What does that mean for the potential success among backs in this year's class then?

The Process

The goal here is to look at a subset of successful NFL running backs, see how they performed within the three main statistical categories in college, and use that information to spot gems in this year's group.

But, uh, how exactly do we define who's in that subset? What's a "successful running back"?

Quite simply, I've arbitrarily defined a successful running back as one who's posted multiple top-20 fantasy seasons since 2011. That gives us 53 running backs. After filtering out the older players and ones who weren't invited to the combine, the sample dropped to 46. Those 46 successful NFL backs are our "studs."

Here's how those 46 players performed in the three major statistical categories referenced above.

This likely doesn't mean a whole lot to you, so let's give these numbers some context. If college production was irrelevant, then we'd expect these figures to be no different than a random running back from this year's class, right? Well, unsurprisingly, players from this year's class don't have the same type of production marks as our stud sample. Because, remember, good players were generally good in college, too.

Now, in the past, I've used NFL Combine invites to help generate the sample of players for this year's class. With the combine not happening this year, said sample is shifting a bit. Instead of looking strictly at invites, I'll be using rookie rankings from our friends over at DynastyLeagueFootball.com. They've got 24 running backs listed in their rankings, and those 24 players will make up our "2021 running back class" sample.

Here's how they compare in the three main production metrics, on average, to the stud sample.

The NFL stud running back group -- the players who've had multiple top-20 seasons over the last decade -- were more productive in college. And you shouldn't be surprised by that.

But which specific players from this year's class, based on the college production metrics that the model looks at, look like studs?

The Results

Total Touchdown Share Concerns

Our 46-player sample of stud running backs showed that they averaged a total touchdown share (the percentage of team touchdowns scored by a player) of 35.4% during their best collegiate season. Among the 24 running back sample from this year's class, 19 failed to hit that mark, with some falling well below it.

Let's start at the top, where you'll see highly-touted prospects Javonte Williams and Travis Etienne. They're on this list because they didn't hit a 35.4% touchdown share in any season of their college careers, but let's not pretend they're far off. It's important to keep in mind that these cutoffs aren't meant to be looked at with a black and white lens: we really should only be concerned when someone's far away from the stud sample's average. With Williams and Etienne, this is no big deal. After all, 26 of the 46 players in the stud sample also failed to hit a best-season touchdown share of 35%.

Trey Sermon is the consensus number-five back with the Dynasty League Football rankers, but he's got flags to his profile. Among the 24 running backs in our 2021 group, Sermon has the second-lowest touchdown share after splitting his Ohio State backfield, and his reception share and yards per team play totals aren't anything special, either. He has the right size, and maybe draft capital will end up being on his side, but for now, he's someone who looks to be overvalued, at least according to the model.

Another player with a below-average touchdown share is Kenny Gainwell. Over the last couple of years, we've seen Tony Pollard, Darrell Henderson, and Antonio Gibson come out of Memphis with interesting skillsets, and that's no different for Gainwell. Like the Memphis backs who came before him, Gainwell's a great pass-catcher who can bring that dynamic to an offense at the next level. His low touchdown share isn't ideal, but he at least makes up for it with positive scores elsewhere in his profile, which consists of just a redshirt freshman campaign.

Reception Share Concerns

If a running back does a lot of work as a receiver in college, it can tell us that he's got a chance to be a good pass-catcher in the NFL. That much is obvious. But it also tells us a little bit about intent.

When a running back is a playmaker, a coach is going to want to get the ball in his hands as much as possible. Sometimes that's in the form of kick and punt returns, but it can come in the form of pass-catching as well.

That's likely why my model cares about reception share, or the percentage of completions that are handled by one player in an offense. When that number is high, it allows us to feel good about a player's ability to catch passes, but it also tells us that a team is trying to do more than just run the player between the tackles.

The stud sample that we're working with had a best-season reception share average of about 12%. That was roughly two percentage points lower when looking at our 2021 grouping. While 37% of the stud sample saw their best-season reception share south of 10%, that number was 58% within our 2021 sample.

The receiving numbers for this year's class aren't all that bad. For some perspective, the lowest best-season reception share in the studs sample came from Derrick Henry -- who had plenty of other attributes to love as a prospect -- at 3.9%. Every back in this year's sample cleared that.

Peep the touchdown share table from above, and you'll see Rhamondre Stevenson's name at the bottom of the list. He had the lowest best-season touchdown share in the 2021 class sample, and his reception share isn't spectacular, either. To be fair to Stevenson's data profile, he transferred to Oklahoma as a Junior, and then was suspended for a good chunk of the 2020 season for a failed drug test at the end of the previous campaign. That's going to impact his overall numbers a bit.

Looking strictly at games he played in, his touchdown share was closer to 22% instead of 12%, while his reception share was an impressive 14.6%. That's strong for a guy who's been listed at 240-plus pounds. We'll see where he ends up getting drafted, but he's one of those cases where these general production trends may not be as applicable.

Total Yards Per Team Play Concerns

The production metric that's weighted heaviest within my prospect model is total yards per team play. It's sort of an all-encompassing statistic that captures backfield share, efficiency, and receiving efforts. That's likely why it seems to matter as much as it does.

The stud sample that we're working with saw 25 of the 46 running backs hit a best-season yards per team play of at least 2.0 yards. That's good for a 54.3% rate. Just 4 of the 24 players (16.7%) in the 2021 class subset hit that mark.

I mentioned Javonte Williams earlier, but I failed to mention his backfield teammate Michael Carter. Williams is the one getting love in the draft world right now, but Carter actually tallied more rushing and receiving yards this past year. This doesn't mean he's a better prospect or that my model likes him more -- there are other factors involved, like size and draft capital -- but Carter's an interesting player to watch as the process unfolds.

Demetric Felton's another player worthy of a shoutout. His yards per team play is nothing to write home about at 1.59 -- that's a number hit by 36 of the 46 players in our stud sample. But Felton played both running back and wide receiver in college and, as a result, he's got the highest max-season reception share in the class. He'll be a fun PPR asset for fantasy managers if he can find a pass-friendly offense in the NFL.

The Studs

Thus far, we've been focused on players who haven't had the most complete production profiles. Let's shift our focus to what you probably came here for: the studs.

When taking the stud sample's average in the three production categories and using that as a filter on this year's class, only one player -- one studly player -- remains.

Najee Harris is a stud. On the best team in the country this past year, Harris accounted for about 38% of his team's touchdowns, well over 13% of their completed passes, and he ended the year with a yards per team play rate of 2.1. When looking at the running back sample that just includes players from this year's class, Harris is fourth in best-season touchdown share, sixth in best-season reception share, and third in yards per team play. If he ends up having strong draft capital -- which is likely -- then he'll be the model's RB1 comfortably.

Funny enough, if we loosen those filters just a tad, the result shows the consensus top-two running backs in this year's class.

Doing this exercise sometimes yields interesting results, but sometimes it just spits out exactly what we expect, like we're seeing with Harris and Etienne.

That's not a bad thing. It means that what people are seeing matches what's happening on the production end.

And that should make us feel pretty good about both Najee Harris and Travis Etienne.

For more information on the prospect model and this year's draft class, listen to The Late-Round Podcast this week, where JJ will break things down in more detail.