This new approach to evaluating running backs shows there's no reason to pick one early in the NFL Draft

Somewhere around the 26th or 27th time that Saquon Barkley pushed 225 pounds off of his chest, it became fair to start wondering about hypotheticals: namely, is there any situation in which picking a running back with the No. 1 overall pick could be deemed justifiable?

We know all the reasons why not. Teams are running less and paying less than ever for running backs. Backs are seen as increasingly interchangeable, and besides, this year’s class is awfully deep, which negates even further the idea of drafting one particularly high.

At the same time, have you seenBarkley’s highlight film? The hurdles? The jets? The hands? Did you see him destroy the NFL Combine? Never mind positional labels ... when someone checks every single box, it doesn’t matter where you take him, right?

Barkley is indeed otherworldly in the open field, is a physical marvel, and is by all accounts a great teammates as well. More importantly, his diverse skill set — the 102 career receptions, the nearly 28-yard kick return average — could give the Penn State product more ways to make an impact than your typical running back.

Still, Barkley’s a running back. He’s usually going to be carrying the ball. And drafting a load-bearing back has had mixed results, to say the least.

NCAA Football: Fiesta Bowl-Penn State vs Washington — Fun fact: Saquon Barkley’s feet never actually hit the ground when he’s running.

As is the case for any position, stats will only tell you so much. The degree of difficulty is so much larger, and the responsibilities differ just enough for blurry translation.

As is also the case, though, we can learn something.

So let’s see what the advanced stats have to say about this year’s running back draft crop. To do so, we’ll lean heavily on two concepts: marginal efficiency and marginal explosiveness.

Marginal Efficiency: the difference between a player’s success rate* (passing, rushing, or receiving) or success rate allowed (for an individual defender) and the expected success rate of each play based on down, distance, and yard line.
Marginal Explosiveness: the difference between a player’s IsoPPP** (passing, rushing, or receiving) or IsoPPP allowed (for an individual defender) and the expected IsoPPP value of each play based on down, distance, and yard line.
For offensive players, the larger the positive value, the better.
* Success rate: a common Football Outsiders tool used to measure efficiency by determining whether every play of a given game was successful or not. The terms of success in college football: 50 percent of necessary yardage on first down, 70 percent on second down, and 100 percent on third and fourth down.
** IsoPPP: the average equivalent point value of successful plays only.

As we’ll see, efficiency translates from one level of the sport to another far better than explosiveness does.

Between 2009-17, 176 tailbacks rushed at least 200 times in a season, about 19.6 per year. Of that group, 146 of them (83 percent) posted a marginal efficiency between plus-1 percent and minus-10 percent. That’s a pretty significant difference — the difference between a 2009 Jamaal Charles (plus-0.9 percent) and a 2009 Brandon Jacobs (plus-9.6 percent).

It’s also, however, the difference of basically one extra successful carry for every 10 or 11 opportunities, at most about two extra successful rushes per game. Doesn’t sound like such a monstrous difference when you put it that way.

Among the 90 halfbacks who a) were drafted between 2010-17 and b) have been given at least 100 pro carries thus far, few have truly stood out.

Alvin Kamara, your 2017 offensive rookie of the year, is the only back to have produced a higher than plus-3 percent marginal efficiency rate, and only two others have been above even plus-1 percent: Montee Ball and Mike Gillislee. They were drafted in the third, second, and fifth rounds, respectively.

Of those draftees with at least 500 career carries, only two players have produced an even slightly positive marginal efficiency rate: DeMarco Murray (plus-0.8 percent) and Ezekiel Elliott (plus-0.3). The latter was selected around where Barkley will be; the former was a third-rounder who battled injury issues in college. Keep that qualifier in mind.

Green Bay Packers v Dallas Cowboys — DeMarco Murray’s marginal success rate peaked at plus-3.2 percent with the Cowboys in 2013.

RB draftees (2010-17) with at least 100 career carries:

Round	RBs	Total carries (first 4 years)	Marginal efficiency	Marginal explosiveness
Round	RBs	Total carries (first 4 years)	Marginal efficiency	Marginal explosiveness
1	12	6332	-6.4%	-0.20
2	18	6953	-5.5%	-0.25
3	14	4711	-6.2%	-0.16
4	18	4626	-5.8%	-0.30
5	13	3327	-6.1%	-0.21
6	12	4288	-6.8%	-0.20
7	3	580	-9.8%	-0.24

Running the ball in general is a marginally inefficient exercise. From 2009-17 in the NFL, the overall rushing success rate was 38.8 percent, while the passing success rate was 42.9 percent.

Teams acknowledge this on third downs — teams have run 72 percent of the time on third-and-1 since 2009, but that percentage drops to 33 percent on third-and-2 and 18 percent on third-and-3 — but they still attempt balance, running more than 50 percent of the time on first down and at least 50 percent on second-and-4 or fewer. Regardless, almost no backs finish with a positive marginal efficiency rate, and backs drafted virtually anywhere in the top five rounds produce similar marginal stats.

Carries per year and effectiveness

Round	Carries per pick (rookie year)	Avg. marginal efficiency
Round	Carries per pick (rookie year)	Avg. marginal efficiency
1	192.0	-6.8%
2	127.3	-5.6%
3	105.3	-5.8%
4	87.0	-7.8%
5	82.8	-7.1%
6	68.5	-4.7%
7	42.6	-13.8%

If a team spends a high-round draft pick on you, you’re more likely to get more opportunities to figure things out. But in terms of value, it’s hit or miss, no matter where you’re picked.

The thesis of my recent quarterback draft analysis piece was that while stats can’t necessarily tell you how a player will perform, they can define that player’s ceiling. And just as quarterbacks aren’t going to exceed their college efficiency values, neither are running backs.

The correlation between your marginal efficiency in college and the first four years of your pro career is about 0.29, while the correlation between your college career and your first pro season is about 0.32. Those aren’t enormous, but they are not bad as projection factors. And as you see below, nobody exceeds their college rate stats, especially out of the gates.

Sixty-six halfbacks were drafted between 2010-17 and carried at least 75 times in their rookie seasons. Only about seven came close to matching their college marginal efficiency.

Only nine of the 66 in this sample came within 3 percentage points of their college efficiency level. So if you’re inefficient in college, that’s not going to change, no matter the line in front of you, the quality of opposition, etc.

We’ll come back to that point.

The correlation between your marginal explosiveness in college and in the pros is almost nonexistent. And to the point that it does exist, it’s negative.

Correlation between marginal explosiveness in college career and in first pro season: -0.11
Correlation between marginal explosiveness in college career and in first four pro seasons: 0.03

It’s almost as if, the more explosive you are in college, the longer it takes you to figure out how to be explosive in the pros. Combine measures like broad jump (correlation with pro marginal explosiveness: 0.18), bench press reps (-0.20, meaning the more you bench, the less explosive you’re likely to be), and vertical jump (0.13) are more closely tied to big-play potential than how many big plays you broke at the college level. And they’re still pretty loose correlations.

(By the way, the correlation between 40 time and marginal explosiveness in the pros? 0.02. No correlation whatsoever.)

Let’s look at this year’s RB draft prospects with the above context in mind. Here are the players most likely to be drafted, along with both their college marginal efficiency average and four factors related to explosiveness: marginal explosiveness, broad jump, vertical jump, and bench reps.

What does this data tell us?

Derrius Guice and Royce Freeman are probably the safest bets.

Despite battling injury in 2017, Guice’s career marginal efficiency was third among these prospects, and one of the guys above him was NC State’s Jaylen Samuels, whom we’ll discuss in a moment. Guice was on the higher end of the explosiveness scale, though his vertical jump blurs that picture.

It’s the same story with Freeman, who wasn’t as explosive and didn’t wow with his explosiveness exercises at the combine.

Explosiveness is secondary to efficiency, though, and of this year’s prospects, Guice and Freeman are perhaps the most likely to approach a marginal efficiency of zero.

Barkley, on the other hand? Hard to say.

Due to some combination of iffy line play and Barkley’s own playmaker instincts — which would sometimes lead to him sacrificing a three-yard gain to attempt something greater — Barkley was not an incredibly efficient back at Penn State.

He did improve a bit during his junior season, moving from a minus-1.1 percent marginal efficiency in 2016 to plus-2.4. And these numbers ignore his receiving ability.

But his style is such that if he doesn’t have a speed advantage over the opposition, he’s going to bounce his way into a lot of two-yard losses.

In a way, his acing of the combine complicates things — his 41-inch vertical was absurd, but typically we associate a ton of bench reps with burly guys like Samaje Perine or Le’Veon Bell, neither of whom topped a 33-inch vertical.

So Barkley defies type. And his explosiveness profile probably translates to the pros. But it’s going to be at least a little bit of a concern until it isn’t.

Nick Chubb and Rashaad Penny are in similar boats.

Penny pretty much broke the marginal explosiveness scale, posting only decent efficiency numbers as a junior (plus-2.5 percent, close to Barkley) but still averaging nearly eight yards per carry. Only guys like Jahvid Best (college marginal efficiency: plus-0.34) and Tevin Coleman (plus-0.29) can compare.

Neither Best nor Coleman saw many big-play opportunities as rookies, and both have been average at best at NFL explosiveness. And Penny’s below-average broad and vertical jumps are a red flag.

Chubb, meanwhile, falls into the Barkley category; he wasn’t as explosive in college, but his vertical and broad jumps were tremendous ... and he was too strong on the bench to project as explosive. That might be a semi-weak correlation that becomes a completely weak correlation.

Lockheed Martin Armed Forces Bowl - San Diego State v Army — Rashaad Penny on one of his many long 2017 jaunts.

There are red flags for Kalen Ballage.

A product of Peyton, Colo., Ballage made a name for himself early in 2016 when he scored eight touchdowns in a single game against Texas Tech. He ran a 4.46 40 at the combine — certainly not bad for a 227-pounder — but again, 40 time and explosiveness don’t correlate. His jumps were average, and his efficiency numbers at ASU were abysmal.

Like Barkley, he brings some receiving ability (he caught 76 passes over his final three seasons, though he gained only 620 yards on them), but if your college stats are your ceiling, then the best you can hope for from Ballage is a marginal efficiency similar to that of a Jeremy Langford or Bishop Sankey.

If you’re going to draft an NC State back, go with the guy who’s supposed to be a tight end.

Samuels was one of college football’s most fascinating players. He lined up as a tailback, a blocking back, an H-back, a tight end, a wideout ... he played basically the JAYLEN position.

Honestly, though, NC State might have been overthinking things. He was one hell of an efficient ball-carrier, but he carried only 166 times, while catching 196 passes, over his final three seasons. And at 5’11, 223 pounds, he should, in my opinion, be viewed as a potential halfback. That is some extreme efficiency potential there.

His teammate Nyheim Hines, on the other hand? Less efficiency potential. His minus-4.0 percent marginal efficiency was fourth-worst among the players listed above, ahead of only Ballage, Justin Jackson, and Ralph Webb.

Just take the Derrick Henry clone.

Bo Scarbrough isn’t quite as big as Henry, a fellow former Bama back who produced a plus-1.4 percent marginal efficiency with decent explosiveness as a rookie. But at 6’1, 232, he’s still awfully big, and his upright running style makes him seem about 6’4. He had some standout moments in college, most notably at the end of his sophomore year, when he rushed for 180 yards and two touchdowns in the CFP semifinals against Washington, then went for 93 yards and two more scores before getting injured in the finals against Clemson.

But there’s the “I” word. He’s been hurt a lot. One never knows for sure how to deal with that moving forward. Adrian Peterson was injured frequently at Oklahoma but missed only nine games in the first seven years of his NFL career. And Murray only played one full season at OU but still managed to post a patently absurd 392 carries in one year (2014 at Dallas).

Scarbrough is an obvious injury risk, and that’s part of the reason he’s projected as only a mid-round pick. He’s had ACL tears, knee sprains, broken bones, you name it. But that doesn’t automatically doom him to future injuries. And hey, buying an injury-prone big man at a slight discount worked pretty well for the Sixers ... eventually.

No matter what, stats always seem to circle back to Bama, don’t they?