Saquon Barkley May Be a Generational Talent, But He'll Still Be Overdrafted
I'm a firm believer that football analysis shouldn't be some binary, black-and-white thing. Numbers matter, but so does film. Josh Allen's potential to bust is high because of a low college completion percentage, but Josh Allen can still work out at the NFL level.
I'm not really a fan of this world of off-the-wall takes and fire emojis. Especially when it comes to a game as complex as football.
So let's not oversimplify things: it's not that running backs can't matter, it's that it's hard for them to matter consistently.
There have been countless studies done on the true impact of running backs in today's NFL. This is certainly not the first one. This isn't even the first one that I've done.
But with the hype surrounding the ultra-talented Saquon Barkley entering Thursday's draft, I figured it'd be beneficial to walk through the many reasons teams should avoid selecting him in the early parts of the first round. Most teams, at least.
And it has nothing to do with Barkley as a player. It has to do with his position.
You Throw to Win
Let's first establish something that should be generally accepted as fact in the year 2018: passing in the NFL is a lot more efficient than rushing is.
Over the last five seasons, the average quarterback attempt has resulted in 7.15 yards gained. The average running back carry has yielded 4.10 yards per rush. That's a pretty significant difference.
This goes beyond standard box score measurements, though. We've got a metric here at numberFire called Net Expected Points (NEP), which, like other expected point models, measures the number of points a player or team adds versus expectation. If, for instance, it's 1st-and-10 on the offensive team's 30-yard line, there's a particular point total we'd expect that team to achieve or score on that drive based on what's happened historically. A play that results in an incomplete pass, making it 2nd-and-10 from that same place on the field, would result in a lower expected point total on that drive. The players involved in that incomplete pass -- the quarterback and the receiver -- are then credited with the difference in expected points from one play to the next.
All of those individual plays, then, can be added up in order to tally a player or team's Net Expected Points total.
Like yards per attempt and yards per rush, Net Expected Points has shown that passing has been far more effective than rushing through the years.
|Year||Passing NEP per Drop Back||Rushing NEP per Rush|
What this is telling us is that, in a given season over the last five years, a drop back has never been less valuable than a rush. And, overall, a drop back has added at least 0.05 more points for a team than a rush has during this time.
Funny enough, the expected points per rush number decreases pretty dramatically when you look only at running back attempts. In turn, so does the difference between a drop back and a rush.
|Year||Passing NEP per Drop Back||Rushing NEP per Rush|
This intuitively makes sense. When you watch a football game, you're constantly seeing running backs getting dropped behind the line of scrimmage or rushing for only a yard or two. Compare that to a random drop back -- even factoring in big turnovers like interceptions -- and it's easy to see why passing would be more meaningful than rushing.
To this point, per drop back efficiency has also been more strongly correlated to winning since the turn of the century:
The orange dots and trendline above represent team regular season wins versus schedule-adjusted Rushing NEP per rush, while the blue dots are the same, but with schedule-adjusted passing efficiency. You can see how much steeper the trendline is for passing, and that's because there's a more dramatic increase in wins with an increase in Adjusted Passing Net Expected Points per drop back.
Teams are winning more when they're passing more effectively.
For you math dweebs out there, the R-value of per-pass efficiency versus wins is 0.66. With per-rush efficiency, it's 0.37. And when focusing only on the last five seasons, those numbers are 0.65 and 0.30, respectively. The delta is even more dramatic.
The same goes in the opposite direction, too: pass defense effectiveness correlates more strongly to wins than rush defense effectiveness does. But we're not as concerned about that here -- we're just trying to find the true value of a running back.
It's More Than Rushing
Of course, the function of a running back isn't just to throw his body around on the ground in order to grind out yards. Especially in this era's NFL, backs are pass-catchers, too.
And, if we're being honest, this is where the Saquon Barkley conversation gets really interesting. Your goal as an NFL team would be to enhance the pass in any way possible -- getting a quarterback, snagging offensive linemen who can pass protect, having strong wide receivers -- because it has a stronger correlation to winning than running the ball does. But what if your running back is a great receiver? Wouldn't that warrant an early first-round selection?
Over the last five seasons, running backs haven't compiled a very good yards-per-target average compared to wide receivers and tight ends:
|Position||Yards per Target, Last Five Years|
And when looking at Target Net Expected Points per target (that is, the number of expected points added on all targets divided by the number of targets seen), we get a pretty similar story:
|Position||Target NEP per Target, Last Five Years|
Naturally, this has to do with how these positions are used. A wide receiver has a better chance to gobble up yards (and expected points) because his depth of target is going to be higher than a running back's.
But that's also sort of the point here, isn't it? Running backs aren't used in ways to alter the game through the air like we see from a wide receiver and a tight end. Whether this is the fault of coaching, scheming, game planning, or just the position itself, it's how it is.
We're dealing with averages here, which means there are plenty of players who rank above the marks listed. So you may be asking, "Aren't there running backs -- maybe someone like David Johnson -- who are consistently providing value like we'd see from a wide receiver?"
No, not really. We've seen 108 instances where a running back had 50 or more targets in a season over the last five years. These are the best of the best pass-catching backs in the league. There were 58 unique running backs who made up those 108 instances, and 30 specific backs have seen 50 or more targets in more than one season over the last five.
So we've got 30 running backs who have had multiple high-volume seasons through the air. Of those 30, only 5 were able to hit the 0.25 Target NEP per target mark -- the average per-target efficiency at wide receiver and tight end -- across multiple seasons. And none have done it more than twice.
To think of this in a super basic way, consider Le'Veon Bell. He's had four seasons now with 50 or more targets in the Steelers' offense. Despite being considered the best receiving back in the NFL, he's hit the 0.25 Target NEP per target mark in just two of those four campaigns. In other words, from strictly a production standpoint, Ben Roethlisberger would've been better off throwing the ball to an average wide receiver versus Bell in two of the last four seasons.
Now, to reiterate, I'm very much a gray-area thinker when it comes to football analysis. This Le'Veon Bell example generalizes the topic a bit -- I understand that. For instance, Bell's value versus a replacement-level receiver like Eli Rogers in the Pittsburgh offense could be that he's often just a dump-off option if Roethlisberger is under severe pressure. That'll lower his efficiency. And Bell's presence on the field may change the way defenses play the Steelers, allowing for a more efficient offense overall.
Trust me, guys. I'm not here to be an idiot. I recognize that data won't tell us everything. But I do think large discrepancies within data should bring curiosity. It should make us question what's generally accepted.
Like, the fact that an average target to a wide receiver or tight end is two-and-a-half times more effective than the average target to a running back is pretty crazy. It tells us that teams either need to stop funneling targets to the running back position or that coaches need to be smarter with how they're utilizing their backs in the backfield. And until the latter happens, it's really tough to back an early first-round running back selection only because of his receiving prowess.
It's More Than Rushing and Receiving
"Hopefully you possess the ball more. You get more plays on offense, fewer plays on defense, and that allows your defense to play at a higher level. It's similar to how we played a couple years ago, when we ran the ball so effectively -- and I thought it had a really positive impact on everybody on our team. We believe Zeke gives us a chance to do that."
The following season, Dallas finished fifth-best in points allowed per game after ranking 16th the year prior.
All thanks to Ezekiel Elliott, right?
There's a very common thought that a good running game keeps a defense fresh, and when a defense is fresh, it performs at a higher level. But that's been debunked by Ben Baldwin.
In a Football Outsiders study from earlier this offseason, Baldwin drew the following conclusion:
Putting this all together, the main -- and perhaps only -- channel through which an offense can help a defense on a per-drive basis is through field position. Turnovers and quick three-and-outs make a team more likely to give up points on the following drive, but this appears to have everything to do with field position and nothing to do with defensive rest time. In other words, whether it's one minute or eight minutes, knowing how long a defense has had to rest tells one nothing about how the defense will perform given its starting field position.
If you recall, Dallas drafted Dak Prescott during that same 2016 draft. Prescott ended up leading the Cowboys to the fifth-best net yards per pass attempt average in the NFL during his rookie year and, for Dallas, it was an increase of 0.9 (6.2 to 7.1) year over year (2015 to 2016). Meanwhile, the team's yards per rush attempt went from 4.6 in 2015 to 4.8 the following year.
Most importantly, the Cowboys defense actually saw more plays against in 2016 than they did in 2015, they were barely better in yards per play allowed, and the offense didn't really give the unit any more rest:
|Yards Per Play Allowed||5.6||5.5|
|Off. Time of Possession per Drive||3:00||3:03|
The reason the Cowboys were better in 2016 wasn't because the defense was no longer tired. It's because the offense was actually competent, scoring a touchdown or field goal on 46.3% of their drives (fourth-best) versus 32.5% (ninth-worst) the year prior.
And since we know wins have a stronger correlation to passing efficiency than rushing efficiency, wouldn't it make more sense to credit the passing attack -- the unit that saw the bigger jump in overall usefulness -- for the team's success over the rushing attack?
That brings me to the next typically unmeasurable piece for running backs, and that's how they help quarterbacks become more efficient through the play-action pass. It's true that there's some correlation between passing success and rushing success (according to NEP, since 2000, the r-squared value between the two is about 0.15) but, once again, Ben Baldwin has us covered.
In a study done in February, Baldwin showed that, while play-action passing is indeed effective, it's not effective because of rushing success. It's just naturally effective. Perhaps this is because a human being playing football can't think to himself as a play is running, You know what? They haven't run the ball well, so I won't bite when the quarterback goes to hand the ball off to the running back. Whatever the case may be, it's another example of instead of just accepting something as truth, we should prove it to be true first.
The League Is Showing Us What's Valuable
If you told me that you liked wine more than beer but you spent five times the amount of money on beer versus wine, I'd call you a liar.
And across the NFL, front offices are telling you -- they're screaming at you -- that running backs aren't valuable.
Take a look at the table below, courtesy of spotrac.com, which shows the average salary for a player at each of the main positions on an NFL field:
Jeff Fisher may disagree with this list, and maybe Jon Gruden would bump the fullback position up a bit, but this is pretty telling. The only relevant position in football that has a lower average salary than running back is fullback, and it's not even really a relevant position.
This is a big deal when we're talking about selecting players early in the NFL Draft. Rookies now have a predetermined contract value, meaning we know, given a particular draft spot, how much these players will make on a rookie deal. If Saquon Barkley goes to, say, the Giants with the second-overall pick this year, then he'll make almost eight million dollars per season. Barkley would instantly become one of the highest paid running backs in football.
Maybe Barkley's skill will make him worth that much dough, but there's also an opportunity cost at hand here. If New York takes the running back, they then forgo the chance to have, for example, a quarterback on a cheap rookie deal. And that deal would include a fifth-year option.
Even if you think front offices are being silly with their lack of spending at running back, you still can't ignore it because of market value. A quarterback going second overall on a rookie deal would be making the 22nd-most money at the position. That's not at all the case for a running back.
No One Is a Guarantee
One of the more fascinating aspects surrounding Barkley's potential in the NFL is that no one ever seems to talk about a scenario where he doesn't pan out. This -- the NFL Draft -- is a game of probability, after all. No player is guaranteed to produce as a pro.
ESPN's Bill Barnwell went through a quick exercise where he listed the top running backs selected over the last 20 years in the draft -- the backs who were picked first in their respective classes -- noting that, of the 20, there were 7 who actually ended up (or will probably end up) as the class' best runner. And just 3 of the 10 running backs selected in the top-five have ended up as the best backs in their class.
This isn't to say that Barkley won't be the best running back among the group being drafted here in 2018, but let's not pretend there's zero percent chance he doesn't reach his fullest potential.
And to this point, having an awareness of "value" versus "safety" is really vital here. Is Saquon Barkley the safest bet to be one of the best at his position in this class? Yes. But that doesn't mean he's the most valuable pick.
Just because you know what a Wendy's cheeseburger tastes like doesn't mean you should be looking to buy one anytime you crave a burger.
Correlation Does Not Imply Causation
The Cowboys draft Ezekiel Elliott, and their win total increases by nine. Elliott misses time due to a suspension in 2017, and the whole offense goes to the you know what.
These connections have caused a lot of people to believe running backs are turning franchises around. But when you dig in and look at their individual contributions -- when you analyze things even on the player-by-player level -- it still doesn't tell a very positive story about the position.
I already talked through Elliott's influence on the Cowboys' win total during his rookie season, and you may have read it and said to yourself, If their success hinged more on Dak Prescott, then why was Prescott so bad when Elliott missed time in 2017?
Here's your short answer: because he played poorly.
What if I told you that the Cowboys didn't see a dip in their running back play with Zeke sidelined? What if I told you that there were signs that suggest Alfred Morris, his backup, was actually better?
Let's start with the basics.
|Player||Yards Per Carry||Rushing NEP Per Rush||Success Rate|
This doesn't tell us everything we need to know, though it does show that Morris and Elliott were nearly identical within our expected points model, even by looking at Success Rate, which measures the percentage of positive runs made by a player. Morris, then, crushed Elliott from a yards per carry perspective.
Was it because Morris saw easier defensive fronts? Well, per NFL.com's Next Gen Stats, Morris actually faced eight men in the box on 38.26% of his carries, while Elliott's rate last season was 35.12%.
That, though, can often be dictated by offensive personnel -- particular formations will force defenders into the box.
There was a study done by Pro Football Focus' Scott Barrett a couple of months ago that helped solve this problem a bit. Barrett didn't just analyze whether a player saw a stacked box or not. Instead, he looked at the number of blockers versus the number of defenders to determine the blocking advantage a player saw.
According to the data, Morris had one of the worst blocking advantages in football last season -- worse than Zeke's. Even still, he managed a yards per carry average that was far superior.
No, guys, I'm not suggesting that Alfred Morris is better than Ezekiel Elliott. I'm merely pointing out the rushing replaceability in football. Prescott's decline wasn't due to Elliott. In fact, even when Zeke returned in Week 17, Prescott managed just 179 yards passing and an 85.6 quarterback rating against an Eagles team that was resting their starters.
Prescott hit a slump. And it happened to occur around the same time Elliott's suspension hit.
The case with Le'Veon Bell is actually pretty convenient because we have a decent-sized sample of the Pittsburgh Steelers with and without Bell since he entered the league.
Have you ever wondered why the Steelers didn't miss a beat offensively when they lost Bell in 2015? If you recall, Bell suffered a torn ACL against the Bengals that year, and 32-year-old DeAngelo Williams stepped in as Pittsburgh's starter. Here's how the Steelers offense fared with and without Bell that year (the game where Bell was injured was not included in the data below):
|Split||Points Per Game||Yards Per Game||Expected Points Per Game|
The team averaged 13 more points per game without Bell, over 140 more yards and, according to Pro Football Reference's expected points model, they went from a below-average unit to one that was well above average.
Sample size is more than likely part of the blame here. But Bell's missed 16 games across his career (we won't count games where the team rested for the playoffs), and in those games, the Steelers have averaged 26.63 points per contest. In every relevant game that Bell's played, the Steelers have averaged 25.19 points.
Maybe this is why the Steelers have been hesitant to give him a long-term deal.
The Jaguars went from a team everyone made fun of all the way to the AFC Championship in just one season, and it was a season that followed their choice to pick up Leonard Fournette at the top of the draft.
How big of an impact did Fournette actually make?
|Player||Rushing NEP Per Rush||Rank||Success Rate||Rank||Target NEP Per Target||Rank|
Per Net Expected Points, among the 47 running backs with 100 or more carries, Fournette ranked 17th in per-rush efficiency. It's possible -- actually, it's probable -- that his rate was skewed by some big runs, as his Success Rate was far lower, showing us that he wasn't creating positive plays very consistently. His per-target numbers weren't awful through the air, but if you reference the chart above, he still wasn't as effective as an ordinary wide receiver or tight end.
Among these 100-plus attempt rushers, Fournette also ranked fifth-worst in the difference between his yards per attempt rate and his running back teammate's yards per tote average.
|Player||Yards Per Attempt||Teammate Yards Per Attempt||Difference|
Referencing Scott Barrett's study again, Fournette did have one of the worst blocker-to-defender ratios in the NFL last year. But when compared to the other running backs in similar situations (LeSean McCoy and Kareem Hunt, for example), his yards per carry average was still lower.
If we were to believe Leonard Fournette is the reason for the Jaguars' offensive turnaround, shouldn't his numbers have looked better than what we're seeing here?
Almost all players need on-the-field help. With the exception of maybe the most elite quarterbacks in football (Joe Flacco not included), a player's situation will either help or hurt him in some way.
After a breakout rookie season, Todd Gurley took a big step back in 2016. According to our numbers, Gurley had the third-worst rushing efficiency rate in the league among relevant backs, and his 35.25% Success Rate was well below average.
Fast forward to 2017, and he's the Offensive Player of the Year.
Gurley was dynamite this past year. Rarely do you see a high-volume running back produce positive expected point totals, but Gurley did that. And to top it off, he won where you need to win in today's NFL: through the air. He had a 0.30 Target NEP per target rate (that's better than an average wide receiver), finishing the season with the third-most expected points added on receptions.
Running backs can matter. Their production can influence the game. But it always takes some help.
The Rams completely overhauled everything from 2016 to 2017. Jeff Fisher was out, and millennial Sean McVay was in. They added receiving weapons. And, most importantly, they added key pieces to the offensive line.
As a result, it wasn't just Gurley who saw a significant bump in production. Jared Goff actually had the largest increase in adjusted yards per attempt from one year to the next that we've ever witnessed in the history of the game.
What's the point in bringing this all up? Impact. Very rarely are teams able to extract a Todd Gurley-like season without a significant, intelligent overhaul and plan. Do you think Alvin Kamara is able to do what he did in 2017 -- producing arguably the most valuable running back season of the year -- if he's not in the favorable Saints offense? If Gurley, as a player, is so dynamic and important, then why was the Rams offense so dreadful in 2016? Shouldn't a game-changing talent (which he is) be able to at least make his offense average?
It's true that coaching and scheme matters for any and all positions. Quarterback included. But if we expect Saquon Barkley to alter an offense completely, much more has to change in that hypothetical offense than just the running back position.
And that's why the position is a luxury, not a necessity.
Running Back Is a Luxury, Not a Necessity
There's a Boston Globe article from 2010 called How Facts Backfire, and it provides somewhat of a scary perspective on how people absorb factual information.
Facts don't necessarily have the power to change our minds. In fact, quite the opposite. In a series of studies in 2005 and 2006, researchers at the University of Michigan found that when misinformed people, particularly political partisans, were exposed to corrected facts in news stories, they rarely changed their minds. In fact, they often became even more strongly set in their beliefs. Facts, they found, were not curing misinformation. Like an underpowered antibiotic, facts could actually make misinformation even stronger.
I've done my best to present facts. Passing, according to research I've done, is a far more effective path to winning in the NFL than rushing is. A running back target -- and plenty of metrics will agree -- isn't nearly as valuable as a wide receiver or tight end target. There's no real link to running back play and defensive improvement. Teams aren't spending at the position. And even the best players to play running back aren't affecting the outcome of games like you may think.
These facts may not change your mind. Football is complicated, after all.
But, at the very least, these facts should open your mind.