A big component of draft evaluation is production. It’s not the biggest part and it’s not determinative—but it is helpful. Spending a first-round pick on a receiver who grabbed 200 receiving yards in his college career seems unwise. Production can fill in missing gaps in our film review and athletic testing; if we think of a position’s technical requirements, we can too often miss on an effective but unorthodox player.
Sebastian Vollmer, for example, is an extremely effective, but unusual offensive tackle. He handles the foundational skills well—he has to—but his approach to pass protection is not something one would expect from a top-level right (and sometimes left) tackle. He stays high, keeps his hands low and wins consistently.
Getting stuck in orthodoxy is a good way to miss winning players, and finding players who produce regardless of how they do it is a good way to win. So, finding players who produce in college can help us find players who produce in the NFL.
Obviously, one has to be careful. Trevor Insley and Troy Edwards are #1 and #2 in the FBS single-season receiving leaderboard. Insley went undrafted in 2000 while Edwards was a first-round pick for the Pittsburgh Steelers in 1999. Between the two of them, they’ve played for 11 professional football teams, NFL Europe and Arena Football included—with 2569 total NFL receiving yards.
A more comprehensive statistical analysis, however, would have argued that Insley was undervalued as a UDFA (and, by having made any yards in the NFL, he was—especially given that it was injury that ended his career) and that Edwards was overvalued as a first-round pick (which, having never broken 800 yards in a season, he clearly was).
Edwards would finish his rookie year a little over 23.5 years old, and he only grabbed 38.5 percent of his team’s receiving yards (a percentage referred to by fantasy quants as “market share”) that year despite being somewhat old for a draft prospect.
The average market share for a receiver in CBS’ top 32 this year is 42.5 percent. For the top 50, it’s 46.6 percent. And those receivers are all younger, too.
Long story short, Edwards’ age and relative production put him below the generalized pick value he was eventually selected at.
This just means that production can enhance film grades, not supplant them. The production profiles in this piece will be used to adjust expectations relative to already established film grades; if Player A has a fifth-round grade but better production than a first-round graded player (Player B), it is unlikely that Player A is actually a better receiver than Player B—just that it is more likely that Player B should have been drafted higher than he was.
There’s a direct relationship between age, market share and future expected performance for receivers in the NFL. One must account for how dominant a receiver is within the confines of his own offense and how old he was within that offense.
But beyond that, I’ve also found that one must account for highly-drafted teammates, and one reason I argued that Michael Thomas had a strong—not weak—statistical case as a receiver despite having a low market share. Playing with Braxton Miller, Jalin Marshall, Ezekiel Elliott and Nick Vannett reasonably suppressed his market share.
Here, we’ll be looking at offensive linemen. Production for offensive linemen used to mean games started, team rushing yards and occasionally a total number of “pancake blocks” as determined by team media guides. Now, we have excellent resources like CFB Film Room and Pro Football Focus that track how many sacks, hits or hurries an offensive lineman gives up.
Most of the numbers for this piece will come from CFB Film Room, with PFF filling in the numbers where CFB Film Room doesn’t have them. They seem to track these a little differently, though not too much—there’s only so much disagreement one can have over how many sacks a player gives up. I adjusted the PFF scores, which are generally harsher to offensive linemen, to match the general pressure rates produced by CFB Film Room.
It’s important to remember then, that these won’t be useful for parsing finer details; two linemen with very similar ranks could end up with individual ranks higher or lower than each other in either system, but will generally be in the same ballpark.
After that, I threw in a demerit for penalty yardage, about ten percent of the weight that pressure rate received.
The most important adjustment might be for the offense. Generally speaking, two things have an enormous impact on quarterback sack rates: the quarterback and the offense they operate in. A QB who can sense pressure or adjust to blitzes may do more to impact a sack rate than a lineman’s skill, and a quick-strike offense can do even more.
So to get around that I incorporated team sack rates as a penalty to the lineman’s individual pressure rate. The team that led the country in sack rate last year was Troy, where Antonio Garcia comes from. While Garcia may individually be a good prospect, I find it unlikely that his four teammates along the line and his quarterback are similarly good prospects—Troy probably wouldn’t have a bottom-of-the-barrel passing offense if that were true.
Similarly, Nathan Peterman at Pitt does a pretty good job of avoiding pressure and that helps offensive linemen like Dorian Johnson and Adam Bisnowaty when it comes to their pressure numbers. It’s not an offense that uniquely enables better sack rates but Peterman himself does.
Accounting for that on a case-by-case basis is difficult, but providing an adjustment—something like a 25 percent weight within the pressure formula—using team sack rates for all of the draftable linemen will provide a clearer picture of their production.
As a reminder, before reading the table, production isn’t everything, especially for these offensive line metrics.
Unlike the receiver metrics mentioned above, I haven’t back-tested any of this data, and there’s only two years of it anyway given that PFF and CFB Film Room only rigorously started charting college players since 2015.
Update: I’ve since done cursory testing of the PFF college data against PFF’s NFL data. I don’t think they have an incentive to fudge their grades to match their college production metrics but it’s important to note. The results of that testing are mentioned in my production profiles piece on cornerbacks, and also quoted below:
For what it’s worth, a system I cobbled together for offensive linemen has seen early positive returns; there’s an extremely strong relationship (for the math nerds: an r of .79 and r2 of .62) between that system and first-year offensive linemen grades in PFF.
The system said Willie Beavers was the worst offensive lineman to be drafted, and he was cut despite being a fourth-round pick. It argued that Jack Conklin was the best, and he had the highest PFF grade of the ones who were drafted. The three best statistical grades of the drafted OL earned the top four PFF grades, with the only “miss” being the highly-drafted Ronnie Stanley, who earned a mere “above-average” statistical grade and ended up with the third-best PFF grade.
That’s extremely suspect given that it has worked within a small sample, but it’s at least noteworthy enough to continue the experiment.
The methods here are just applications that make sense to me and not a rigorous understanding of offensive line metrics as applied to projection overall. Also, these only test pass protection and penalty avoidance, not run game production.
We have no idea if penalties project to the next level or if age is as significant for offensive linemen as it is for wide receivers, but for now it’s worth including as a means of understanding how those linemen perform.
The production ratings are on a baseball-style 100+ scale, where 15 points above 100 is one standard deviation above average while 15 points below is, as you’d imagine, one standard deviation below average.
As an example, Matt Ryan’s passing performance this year in a 100+ system using adjusted net yards per attempt, was 143—nearly three standard deviations above average (45 divided by 15 is three). Sam Bradford, at 114, was nearly one deviation above average and Tyrod Taylor was dead even at 100. Jared Goff, at 61, was more than two below average.
The ranks and projected rounds used below were grabbed from CBS’ official rankings from a week or so ago. The ages were compiled to the best of my ability and represent a player’s age at the end of December 2017.
First are the centers:
The only unremarkable score is Tyler Orlovsky‘s, which is functionally average. All this means is that one wouldn’t change his grade based on his production. He’s generally been quite good at pass protection relative to his peers, but his massive penalty rate (17 penalties this year, more than any other offensive linemen featured in this piece) and somewhat advanced age drop him.
Kyle Fuller has even better pass protection numbers (giving up pressure on a mere 0.8 percent of pass-blocking snaps) with only one penalty this year, but played in a very pressure-friendly scheme and will be extremely old relative to his peers; the oldest of the linemen evaluated for this piece.
Ohio State’s center Pat Elflein is a gifted run blocker, which this doesn’t measure. But his pass protection has been subpar and the worst of the four centers in CBS’ top 150. In fact, he has the worst pass protection grades of centers in CBS’ top 300; more sacks (3) and more total pressures (13). Given that Ohio State passed the ball more often than a lot of its contemporaries, that’s not awful—but it’s worse than his peers, and his minor penalty issues add into it.
Ethan Pocic, therefore, stands above the rest. Not only is he younger by more than a year, he gave up no sacks or quarterback hits—only five pressures, per CFB Film Room. His pressure rate isn’t spectacular, but it’s good and looks a lot better after adjusting for the offense he was in and his draft age.
Next, we’ll look at the guards who played guard in college:
There are a lot of guard prospects I like in the class, but it turns out that some of my favorites—Dan Feeney, Dorian Johnson and Isaac Asiata—have underwhelming production. For Asiata, age seems to be the biggest issue. His raw pressure total isn’t good (four sacks, two hits and nine hurries), but given the number of pass protection snaps he had (496) and how difficult it was to prevent sacks in that offense, he gets a little leeway even after incorporating a high penalty rate. His age has the biggest impact on his outcome and he needed a stellar pressure rate to overcome it.
Johnson had a good pressure rate but was blocking for one of the easiest quarterbacks to block for; Pitt only suffered a sack on 3.0 percent of dropbacks. Out of 128 schools, that ranks seventh-best. Feeney simply had a surprisingly poor pressure rate; no additional context needed there. He’s fine with regards to penalties and his offense while suffering a small age penalty.
One player who I didn’t think was particularly good in the Senior Bowl, Danny Isidora, ended up with a pretty decent production rating. He had an average pressure rate but in an offense that was a bit difficult to block for and without many penalties. He functionally gets a bonus for how long the quarterback held the ball.
Two players that stand out in big ways, Nico Siragusa and Damien Mama, accomplished their ratings in different ways. Siragusa had the single-best pressure rate among guard prospects while also playing in the most difficult offense to block for; San Diego State allowed a sack on 9.45 percent of their dropbacks—one of the highest rates in the country.
The only reason Siragusa’s score isn’t higher is his age, which isn’t too damning anyway. Damien Mama, on the other hand, had a pretty good (but not great) pass protection score in an offense designed to prevent sacks; they had fewer than even Pitt—USC had the third-most sack averse offense in the FBS. What boosted his score was his young age.
Next, we can look at players who played tackle in college but are projected by CBS to be guards in the NFL:
There’s not much to be said about Forrest Lamp and Dion Dawkins; it’s unlikely that they’ll be under or overvalued after considering their production. They are both a little old for prospects—like Feeney, Siragusa and Isidora—but Lamp committed one penalty all year without giving up a sack (three hits, one hurry) in an OL-friendly offense.
Dion Dawkins gave up significantly more pressures (two sacks, two hits and seven hurries) and had a penalty problem, but did so in an offense that was tough on offensive linemen (in fact, one of his positives coming out is his experience in a “pro-style” offense).
Zach Banner isn’t just old for a prospect; he was not particularly talented. He had a fine pressure rate among college peers, but it’s below the standard of top 150 prospects (three sacks, two hits and 12 hurries on 412 pass-blocking snaps) and it came in USC’s very easy-to-block offense. His penalty issues aggravate the problem.
On the other hand, David Sharpe of Florida had an even pass-blocking performance in a fairly difficult offense to block for, and did it at a very young age compared to his draft prospect contemporaries. Sharpe is probably a long-shot but his production and size alone should be enough to cast a long look.
And now for tackles, all of whom naturally played tackle in college:
It’s an unusually old offensive tackle class—the average age for prospects in CBS’ top 150 is 22.9 years old, and only two of the 13 tackles listed above are below that age. Those two tackles, Roderick Johnson and Cam Robinson, consequently end up with very positive production grades.
You can see the numbers in the header image of this story, but what’s interesting about Robinson is that he posted good pass protection numbers despite scouting reports referring to him as “raw” and “unpolished” in comparison to Ryan Ramczyk’s “pro readiness.” He does have a penalty issue and operates in an offense that prevents sacks, but his age gives him a big boost.
I generally dislike “project” offensive tackles like Robinson is purported to be, but if it comes with as much success as Robinson has already shown, one could make an exception for him.
At Florida State, Roderick Johnson hasn’t exactly flown under the radar, but he hasn’t received much mainstream traction. Still, he’s a somewhat polarizing prospect for those who have watched him. His scouting report sounds like an endless stream of what coaches might love to have; he’s long, big and comes from a pro-style offense. He needs to be coached up in some ways of course, but he has all kinds of tools (evidently).
A generally good pass protection score might be confusing given the four sacks given up, but with over 450 total snaps and fewer hurries overall, he ends up having a fairly average total pressure rate—and at a young age while playing for a team that gave up a lot of sacks; they ranked 101st of the 128 FBS teams. Being the lone star of that kind of offense can be meaningful.
Ramczyk, Antonio Garcia, Adam Bisnowaty and Erik Magnuson all ended up having largely average grades but for different reasons. Garcia famously hasn’t given up a sack this year and he has great pressure numbers but Troy was the number one team in the country at preventing sacks. That means a larger penalty than was given to USC or Pitt offensive linemen.
In what is essentially (for projection purposes) a gimmick offense, Garcia gave up as many pressures as his teammates did. On top of that, he’s just a hair older than the average prospect and has penalty issues.
Bisnowaty is even older with a worse pressure rate in a similarly friendly offense (though he’s better about penalties). Between the two of them, their offenses hid their production more than other offensive linemen and any predraft talk about how many sacks they’ve given up should be treated with a lot of suspicion.
Ramczyk’s Wisconsin team and Magnuson’s Michigan team had much more standard offenses, though Wisconsin was a tougher assignment. Between the two of them, they have combined for one penalty for seven yards. That should make their age and relatively average pressure rates more than palatable for their projected round slots.
Sitting in the middle of the remarkable scores of Robinson and Johnson and the average scores of Ramczyk, Garcia, Bisnowaty and Magnuson are two tackles that are at opposite ends of the spectrum. Dan Skipper and Taylor Moton. Skipper has the worst pressure rate of the listed tackles and committed a fair deal of penalties, so it may be interesting to see him on the positive side of the ledger here.
The Razorback benefits from two things in this situation: first, the obscene sack rate of the Arkansas offense; it’s not as bad as Siragusa’s San Diego State team, but it is one of the worst in the FBS. Second, because Skipper’s scores come from Pro Football Focus, a correction was made to adjust PFF’s more harsh scores against offensive linemen to the scores closer to the CFB Film Room numbers.
The other tackle with an above-average score is Taylor Moton. Moton comes from the same school as Vikings fourth-round draft pick Willie Beavers, but is significantly different. Beavers was PFF’s least-favored and lowest-graded tackle. Moton was at one point their highest-graded tackle.
Unlike Beavers’ 2015, Moton didn’t allow a sack this year and only allowed nine total hurries. He is coming from a spread-ish offense, which helps with sack numbers and his age isn’t great, but his pressure numbers are good enough to put him well above average.
After that are two players next to each other in round projection, Will Holden and Chad Wheeler. They both also have fairly disappointing scores. Holden’s awful 7/8/11 sack/hit/hurry line is buoyed in part by the PFF to CFB Film Room correction but mostly by the same things that helped Skipper and Johnson; a bad sack protection offense. Still, his age doesn’t help.
The same is true of Wheeler, a player with fine statistics but again in an extremely friendly offense. His age hurts things, and that’s the reason he sits below the middle.
There are two extremely poorly scoring tackles as well: Conor McDermott and Garrett Bolles. McDermott has a combination of factors working against him; mediocre pass-protection scores, an offense that could avoid sacks with some ease and age issues. McDermott is the second-oldest of the tackles and the third-oldest of the offensive linemen featured in this piece.
The second-oldest is Bolles, who doesn’t have the solid track record that the oldest player (Kyle Fuller) has. Bolles isn’t just an old prospect, he’s given up a surprising amount of pressure in half the pass blocking snaps of his peers. In an average pass-blocking offense with those age problems, he has the lowest score of any of the offensive linemen.
This sort of work isn’t meant to rank offensive linemen; it’s meant to give us a clearer picture of one aspect of evaluation. It might also give us redlines and in that way it means one could take Isaac Asiata, Garrett Bolles, Conor McDermott and Kyle Fuller off the board.
One might also then give Roderick Johnson, Cam Robinson, Ethan Pocic, Damien Mama and Nico Siragusa extra consideration.
It might also give us clues for which linemen to watch for in the combine. Players like Taylor Moton, Dan Skipper, Forrest Lamp, Dion Dawkins and Danny Isidora could rise into more serious consideration with a good combine.
It might also be the final blows for athletes with already poor performances. Dan Wheeler, Zach Banner and Pat Elflein might be due for a serious drop with a disappointing combine.
