It’s never shocking to hear that one of the top prospects in the NFL Draft happened to be a fantastic athlete when it came to the pre-draft testing circuit. Myles Garrett and Khalil Mack were two of the best athletes we’ve seen go through the combine process, and it’s been fairly unsurprising to see them turn their physical skills into dominant NFL careers.

But NFL teams pay attention when a prospect who has gone under the radar posts elite numbers. We regularly see players without much college production — but who show supreme athleticism in their testing — get drafted earlier than initially expected.

Sometimes, this works out — like for Jimmy Graham, Donovan Peoples-Jones or Danielle Hunter. With counterexamples like Stephen Hill, Justin Hunter, Jordan Willis and others, it’s difficult to say whether this approach really works.

But if we limit our definition of “athleticism” to workouts that historically correlate to performance at each position, we can see that there is some advantage to overdrafting for athletic ability. I used data from 2007 to 2014 to come up with rough models for positional athletic ability and looked at players from the 2015-19 drafts to see if they outperformed their competitors at the same draft slots.

We’re focusing on players drafted outside the top 100 because it’s not particularly interesting to draft a super-athlete that everyone already knows is good.

Only 25 percent of players drafted outside the top 100 end up starting at least one year in the NFL, compared to 36 percent for players identified as better prospects by the model outside the top 100. Athletes were nearly three times as likely to make the Pro Bowl (6.7 percent as opposed to 2.7 percent) as their similarly drafted less-athletic counterparts and start more games (23 percent of potential starts, as opposed to 15 percent).

One has to be careful: Teams won’t want to pass up on Za’Darius Smith to draft Geneo Grissom. But there’s a lot of value in identifying athletic potential and prioritizing it. To find under-the-radar players, we’ll look at players who are expected to be drafted but sit outside Dane Brugler’s top 100 who tested extremely well at each non-specialist, non-quarterback position.

Running back

At running back, there are two potential paths to success: as a traditional bell cow and as a scatback. While modern running backs now do a lot more pass catching than before, it’s still important to create two athletic models to account for the fact that running backs over 215 pounds will win in ways that those weighing 190 pounds generally won’t.

For bigger backs (210 pounds and above), the broad jump and 10-yard split matter the most and it actually helps to be a little taller, despite conventional wisdom. For smaller backs, the 10-yard split still matters, but the three-cone and “flying 20″ — the back half of the 40-yard dash — have historically correlated with success. In the recent past, this model has looked favorably upon Aaron Jones, Tony Pollard, Marlon Mack, James Conner and Nyheim Hines.

This model likes three backs outside Brugler’s top 100 that many people still expect to be drafted: Ohio State’s Trey Sermon, Louisville’s Javian Hawkins and Michigan’s Chris Evans. Sermon’s workout numbers didn’t draw a lot of plaudits, and his 4.61-second 40-yard dash won’t raise any eyebrows, but for a big back like him, his incredible 1.49-second 10-yard split (a 97th percentile score) and 10’5” broad jump speak to his potential success. And as a bonus, he’s not stiff, having run a 6.84-second three-cone.