The simplest aspect of college evaluation might be the skeleton key to improving accuracy for NFL scouts. At the basest level, how a player plays in college—one would think—would provide the foundation for draft evaluation.
As always, it’s not quite that easy. Just a list of Heisman Trophy winners can be enough to dispel the idea that playing well in college clearly projects to playing well in the NFL. Traits matter more than the fact of good performance, and seemingly the 2016 defensive rookie of the year (Joey Bosa) proved this with 10.5 sacks in 11 starts despite grabbing a mere 5.0 a year prior for Ohio State.
But evidence suggests that production has been undervalued by NFL scouts and that there’s an edge to be gained when combining film grades with production scores. Critically, those production scores need to be more nuanced than total receiving yards for receivers or total touchdowns for linebackers.
Instead, production statistics that take into account team role, strength of schedule, strength of cohorts and frequency do a better job. Bosa is a good example; edge defenders who do well in the NFL don’t necessarily have high sack totals. But they do generally have a standout year somewhere in their college careers. For Bosa, that was the year prior to his final stint in Columbus, when he racked up 21.5 tackles for loss and 13.5 sacks.
We’ve gone over production profiles for a number of other positions, and those demonstrate the concept equally as well. For a better idea, be sure to check them out:
- Offensive Linemen
- Off-ball Linebackers
- 4-3 Defensive Ends
- 3-4 Outside Linebackers
- Defensive Tackles
Today, we’ll be taking on one of the best positions in the draft, and one of the deepest such classes we’ve seen in years: safety.
Safeties are tasked with a number of difficult demands, all of which add to the complexity of evaluating their impact. Some of the best safeties in the NFL, like Seattle’s Earl Thomas, play far away from the ball and have an enormous unseen impact on the game that might not get noticed until their absence.
Others, like Reshad Jones for Miami, play a critical role in the run game and as a short-area enforcer—dealing blows to opposing receivers and running backs close to the line of scrimmage. There are safeties who excel in man coverage, like New England’s Devin McCourty, and those who are a big part of blitz packages, like the Giants’ Landon Collins.
And of course, there are players tasked with all of those roles—like Minnesota’s own Harrison Smith.
They all play the same position.
In order to deal with this problem in some way, I made sure to use the production numbers to create not just grades, but categories of safety. Those production numbers were already detailed in earlier pieces—using the tackle share metrics from the linebacker study and the coverage metrics from the cornerback piece, I was able to create ratings that respected the role that safeties play in the run game while also keying in on the fact that they are primarily coverage defenders.
As a reminder, the tackle share metrics are three-fold. The first is total percentage of team solo tackles, the second is total percentage of team tackles for loss and the third is the percentage of tackle attempts that were missed. Players with teammates expected to go high in the draft were given some credit for their teammate’s tackle production, with about a third of a first-round teammate’s tackles credited to them, a fourth of a second-round teammate’s tackles and so on.
The coverage statistics are more straightforward: ball-hawk rate is the percentage of targets that a defensive back either intercepted the ball or broke it up by getting their hand in the way and adjusted yards per target measures how many yards, first downs, touchdowns and interceptions a player allowed or created in coverage. Coverage statistics were further modified using strength-of-schedule adjustments.
Sports-Reference.com provided the tackle and strength-of-schedule statistics while CFB Film Room provided the target data and missed tackle data. Where CFB film room did not have data, Pro Football Focus did.
With passing grades and run defense grades already distinct from each other, I was able to use the difference between the two grades to create three categories: free safety (those with much better pass defense scores than run defense), strong safety (the opposite) and hybrid safety (players close enough in both types of scores).
I don’t know if that’s how they project in the NFL or if that’s what evaluators see as their overall use, but it does provide us with a quick means of distinguishing between the different types of safeties available. Teams with talented single-high safeties may want to find a complementary piece with a player who tested as a better run defender, while those with thumping box safeties might want to find coverage guys.
This isn’t to say a player categorized as a strong safety is necessarily poor in coverage or can’t play as a strong safety—only that the most ideal circumstance would have them playing strong safety because it maximizes their talents.
Speaking of, let’s start with the strong safeties.
The strong safety group is young to start off. Between the 12 total safeties, the average age is 22.5 (again, these are ages at the end of their rookie year), so seeing three safeties benefit right away from premiering at a younger age than the average safety at the top of the list speaks well to the classes’ prospects.
Jamal Adams is one of two safeties this year placing in the top-ten overall prospects, something we haven’t seen in quite some time. The LSU product earns a stunning 120.5 score here and proves that he’s more than just a fourth linebacker in the box.
Adams sports the best tackle share among the safeties in the class and above-average scores in missed tackle rates and tackle-for-loss rates. He does have the lowest ball-hawk rate among safeties, but shuffles along with a decent adjusted yardage allowed along with one of the most difficult passing schedules among his peers.
He’s not as versatile as the safeties in the hybrid safety category, but certainly has versatility capability, if the numbers are to be trusted.
Marcus Maye and Jabrill Peppers share the same, functionally average, score. Peppers, from Michigan, is a good deal more controversial. It’s easy to see why; Peppers has the worst pass defense rating of the 12 safeties in CBS’ top 150. Not only that, Peppers (along with Washington’s Budda Baker and Alabama’s Eddie Jackson) might benefit too much from teammate corrections.
There are so many defensive Michigan players in the draft (Taco Charlton, Jourdan Lewis, Chris Wormley, Ryan Glasgow, Channing Stribling and Ben Gedeon) that Peppers gets credited for 18.2 percent of his team’s tackles when in reality, he earned 11.2 percent. That is pretty low among his contemporaries at the same position—and he played linebacker.
But the score is what it is for now, and it marks him as a good run defender—and to his credit, he has a very low missed-tackle rate.
Peppers allowed the highest adjusted yards per target in the set and had a relatively weak schedule, so the criticisms of Peppers might be worth something.
Maye doesn’t actually have many statistical problems, he’s just going to finish his rookie year at over 23 years old while others in his class will do so at 21. An average schedule, tackle-for-loss rate, and coverage score are buttressed by a great ball-hawk rate and best-in-class missed tackle rate. He receives a small ding for his overall solo tackle market share but mostly loses all his gains with his “advanced” age.
The negative standout among the strong safeties is Obi Melifonwu from Connecticut. Of the seven statistical markers that make up these scores, he only generates one positive mark: team solo tackle share, where he’s above average among safeties. Besides that, he has the lowest rate of tackles-for-loss, the third-lowest adjusted yards per target (behind Peppers and one of the hybrid safeties) and an average missed tackle rate.
All of this was done against the easiest schedule in the group and at an older age than the other safeties. He might end up making ground at the NFL combine, but there’s enough reason to be concerned.
Budda Baker will receive size demerits from scouts, but he shouldn’t get a knock for his production. Like Peppers, his tackle share is boosted by teammate corrections, but they are much smaller—his gains from that process are half that of Peppers’. Overall his run production, including tackle-for-loss share and missed tackle rate, is above average across the board. His pass coverage scores even to about average, but has opposite indicators.
He allowed few adjusted yards per target but had a low ball-hawk rate against a relatively weak passing schedule.
Though these sound like weak endorsements at best, the scores come out to be very positive in part because it’s so unusual to produce at that level at such a young age. Generally speaking, we should see improvement over time and the fact that he’s the second-youngest player in this group of safeties is a big positive.
Here are the hybrid safeties, safeties who are equally good (or bad) at run defense and pass coverage, according to the production statistics that we’ve molded together. Because there’s more of a weight placed on pass coverage, free safeties should have an inherently higher grade than hybrid or strong safeties, but that’s not necessarily the case.
There’s a big red no for Rayshawn Jenkins from Miami. Jenkins does have an age penalty, but it’s relatively minor. Instead, what gets him are across-the-board poor scores, doing his worst in adjusted yards per target and doing least-worst in tackle share, which is still significantly below average. The system ends up giving him some credit for the quality of passing opposition he played against, but he’s largely been unproductive in college and may not be worth a pick in the top-200, much less the top-150.
The only other hybrid safety with a concerning score is Texas A&M’s Justin Evans. Evans, by a pretty large margin, has the highest missed tackle rate of the safeties and a poor overall tackle market share. Despite that, he earned a good portion of his teams’ tackles-for-loss and did an average job generating pass breakups. Still, despite being a little better in coverage than against the run, he comes out with some production concerns.
Both Josh Jones from North Carolina State and Malik Hooker from Ohio State ended up with positive scores. Hooker and Jones both ended up with clearly better coverage scores than run defense scores, but the difference wasn’t stark enough to pull either of them out of the “hybrid safety” category.
As you’d expect, Hooker had above-average scores as a ball-hawk and in preventing yards to his targets, and did so against pretty good competition. He also generated a fairly good amount of his teams tackle total, though has a concerning missed tackle rate and was responsible for a low percentage of the team’s tackle-for-loss rate.
His age gives him a fairly large boost and it recommends him well for the possibility of future improvement—and run defense was something scouts saw improve throughout the course of his year.
Jones is a little different in that he had a slightly better tackle share score and worse passing scores, but did it against much tougher competition. His passing metrics aren’t significantly worse and the opposition adjustment makes up for it, but it does imply he’s slightly more well-rounded if less talented.
Finally, the free safeties. No free safety has a production profile that was easy to break down, which you’ll see in a moment.
Marcus Williams‘ score is too high. I don’t mean that like hyperbole, I mean that his score is probably skewed in some way worth drawing suspicion.
It’s not impossible to grab scores above 130 or anything—after all, a group of 150 players should have some players two standard deviations above the mean (in fact, there should be four)—but that half of his rating is built on 15 targets. Sure, it’s a credit to him that he was only targeted 15 times, but it also makes his per-target stats particularly suspect.
It might be how Utah played him or how weak his supporting cast was, but the few times he was targeted an unbelievable -10.5 adjusted yards per target. For context, the next-best safety (Malik Hooker) allowed +0.58 adjusted yards per target. The range among safeties not named Marcus Williams was +0.58 to +9.25—with reasonable steps in between (the next-best scores were +0.68, +1.87, +2.63, +3.38, +3.62 and so on).
For the purposes of calculating scores, I excluded Williams in the ratings for the other safeties and reintroduced him to get his rating relative to that group.
Still, it’s very likely that he’s a very good coverage defender and deserves high marks; his 2015 scores were very similar (and an equally low number of targets) and his ball-hawk rate was pretty high. He was very good at avoiding missed tackles (the second-best in the entire safety group) but had a low overall tackle share and tackle-for-loss share.
All of this points to a safety who might potentially have been asked to play extremely high up and away from the line of scrimmage on a large number of snaps, which is a rare usage function in the NFL. People who have actually watched him may know better.
For now, his statistical profile is extremely friendly, if slanted for a unique style of free safety.
There was a separate problem with Eddie Jackson from Alabama. He missed enough games due to injury this year that it was important to evaluate his 2015 statistics, with the understanding that his tackle share would be supplemented by teammates from the 2016 draft and this upcoming draft (with a round discount for those in the 2017 draft applied retroactively).
In order to give him credit for age-specific production, his age score was given a year “bonus” even though his age listed above will be his actual age at the end of December of this year. In other words, he was graded on 2015 production as if his age was going to be 24.1 at the end of his rookie year.
That’s still a pretty significant age penalty and he overcame it, though it comes with an enormous amount of tackle credit in the run game because of teammate effects. Peppers benefited from an additional seven points in tackle share, but Jackson gets a whopping 10. His original tackle market share of 6.8 percent is by far the worst among his peers, so one might want to be a little hesitant on how his production translates into run success if one already had concerns.
Still, it needs to be said his production as a coverage defender was so good that he overcame a severe age penalty, which is rare. Again, being an older prospect isn’t bad—it’s being a prospect who does not produce commensurate to their age that’s the issue. Players like Marvin Harrison and Keyshawn Johnson have proven that age-adjusted production in college matters more than age alone, and Jackson might have outperformed his expectations in a similar manner.
He had the second-best ball-hawk rate in the class, the third-best adjusted yards per target (second-best after excluding Marcus Williams) and did all that while facing the strongest passing slate of any safety. There are issues with his high missed tackle rate, but what really brings him down to only “a little bit above average” is his age.
We, of course, have no relevant data on Lorenzo Jerome from Saint Francis of Pennsylvania. A cursory look at his college statistics suggest some concern, as he only grabbed 8.2 percent of his team’s tackles (the lowest in the set, assuming the teammate corrections for Jackson remain pristine) and 5.8 percent of his teams TFLs (below average, though not as low as a number of other safeties).
He grabbed six interceptions and five additional pass deflections, which doesn’t tell us much without target data but can be compared to what his teammates produced. Cornerbacks Jalen Wells and Malik Duncan produced more total pass deflections (15 and 16, respectively) and though we know that CBs produced more deflections than safeties, it should concern us that the only defensive prospect from the school falls this far behind his teammates.
Again, he might have only been targeted on rare occasion, so that can’t tell the whole story, but there are some questions to answer. Though four of his defensive teammates made all-conference, he was the only defensive back among his teammates featured there. Not only that, he was the only player in the conference history to be selected first team in four consecutive years.
He followed that up with two interceptions in the NFLPA bowl, and two interceptions in the Senior Bowl, as well as a forced fumble. He’s also clearly athletic, as he was an all-conference returner and was second in the FCS in kickoff return average.
Given those conflicting pieces of data, it may be better to dispose of production analysis for Jerome specifically. People who have concerns will just as easily find problems with his production as his proponents will find reasons to support him as a draft pick.
Another stab at a more speculative look revealed stark divides in the safety group. There are effectively only three safeties with average grades, and one of those grades was workshopped from a prior year’s effort.
Otherwise, the safeties operate at extremes, with three safeties occupying a danger zone and five signaling green lights. From a production standpoint, it seems like a unique and varied safety class—a conclusion that reflects well on the generally well-regarded defensive backfield group in this draft.
Very soon, we’ll take on simpler offensive positions—tight end, wide receiver, running back and quarterback.