Some takes are scary to actually put out into the world.
OK, here it goes: *whispers* Aaron Donald is average against the run.
The Los Angeles Rams defensive tackle's utter dominance in every aspect of his role on the field is seemingly the only thing the entire internet agrees on. Or it was, anyway. But I've got the numbers to back this up.
That's the big takeaway from our new run stop win rate (RSWR) and run block win rate (RBWR) metrics based on NFL Next Gen Stats player tracking data, which were released on Tuesday.
When my colleague Brian Burke -- the creator of these metrics -- first saw Donald in the middle of the pack among interior linemen, his assumption was the problem was the model, not Donald. It had to be a bug.
But the more we watched Donald's play, the more we came to the conclusion that the model was picking up on something. The plays marked as losses were losses, and Donald was not winning on run plays at an exceptional rate. According to the RSWR model, winning is possible by a defender:
Beating his blocker such that it puts him in better position to stop the runner or ...
Disrupting the pocket/running lane by pushing his blocker significantly backwards or ...
Containing the runner, even if blocked, such that he must adjust his running lane or ...
Recording a tackle within three yards of the line of scrimmage.
Given that criteria, Donald recorded a RSWR of 29% last season. That's 1 percentage point below the qualifying interior linemen average of 30%, and well short of the category's leader -- Broncos nose tackle Mike Purcell, at 40%. Now, that might not be a fair comparison to have a gap shooter like Donald against a nose tackle, but even if we remove every qualifier that played at least 50% of their snaps at 0-technique or 1-technique -- directly over or just off the center -- the same result still held (although I'll note that Donald almost never plays nose tackle, while many other DTs play some snaps there).
So where's the disconnect? Among all defensive tackles, no one averaged a larger "push" -- upfield distance covered in the first two seconds of the snap than Donald -- at 1.3 yards. That should be a good thing, and yet his RSWR is only mediocre.
Sometimes all that push results in Donald recording utterly destructive run stops. It's possible that he actually does have a higher-than-average number of highlight-reel run stops that garner an outsized amount of attention. But on a per-play basis, he doesn't win against his run blocker more than the league average.
One theory on why Donald is below average in RSWR: Donald sometimes sells out to stop the pass, which leads to vertical overpursuit. Need an example? Watch No. 99 in this NFL Next Gen Stats animation:
Plenty of other times, though, Donald loses the way any other defender does: just getting blocked.
Regardless of the why, however, there were clues pointing to this conclusion on Donald in more basic statistics. Last season he recorded a tackle on 9% of run plays. That ranked 130th out of 158 defensive linemen who played at least 300 run snaps (Miami's Davon Godchaux and Atlanta's Grady Jarrett led the category at 20%).
Opponents also did not run away from Donald: Runs to each side of the field were roughly equal when he lined up on both the left and right side. In addition, he was double-teamed at an above average but not exceptionally high rate relative to other defensive tackles.
In fact, over the past three seasons, the Rams' run defense averaged the same expected points added per play with Donald on the field as with him off the field. We would expect an elite run defender to have stronger numbers.
Should the Rams worry about the 29-year-old Donald? Not in the slightest. He's coming off a 12.5-sack season and is clearly the game's best interior pass-rusher. Run defense is less important, and if it's a trade-off between the two he's making, that's a fine one to make.
Still, Donald's lack of elite production against the run is perhaps the hottest take from our RSWR and RBWR metrics. There are plenty of other things it can tell us heading into the 2020 season, however:
Can the Ravens replace star guard Marshal Yanda?
When Yanda retired this offseason, there were questions about how Baltimore could fill his void at right guard, particularly for an offense that relied heavily on the run. And indeed, Yanda was the top-ranked guard last season in our RBWR metric, which hammers home how crucial he was in front of Lamar Jackson & Co.
The good news for Baltimore? They're bringing back their other four starters along the offensive line in what was our top-ranked run blocking offense. That includes left guard Bradley Bozeman, who ranked second only to Yanda among guards in RBWR.
In addition, former Seahawk D.J. Fluker -- who ranked 17th in RBWR last season among guards -- is the favorite to win Yanda's old spot. That would make him the weak run-blocking link along an exceptionally strong Baltimore line, but he would hardly be a liability.
Ezekiel Elliott ... matters?
We're not going to settle the value of the running back debate here. But by comparing each team's RBWR to its expected points added per rush, we can see which teams overachieved in the ground game relative to their blocking. There's a decent correlation between the two (r-squared of .28).
Team-level run block win rate compared with run play EPA/P, for the 2019 regular season.
— Seth Walder (@SethWalder) September 7, 2020
RBWR: A new stat from ESPN using @NextGenStats data. pic.twitter.com/UkfWrzVj7n
And it's hard not to notice that the Cowboys' production far exceeded the blocking the that running backs received (team RBWR also includes tight ends and fullbacks). That's a sign that Elliott (and maybe Tony Pollard) are adding something.
The same could be said for the running back corps in Arizona, Indianapolis, Tennessee and Minnesota. On the flip side, Christian McCaffrey (Carolina), Saquon Barkley (New York Giants) and Alvin Kamara's (New Orleans) teams all performed only around expectation given the blocking they received.
So that's why the Giants tagged Leonard Williams
I was sort of dumbstruck at general manager Dave Gettleman's decision to franchise tag Williams, who seemingly has never lived up to his potential. As a pass-rusher, Williams was average last season, although he faced an above-average number of double-teams. That's not a franchise tag-caliber level of play, however.
But where Williams was elite was in the run game, where he ranked fourth in RSWR among defensive interior linemen. His push -- the distance traveled vertically in the first 2.5 seconds -- was also strong, ranking 11th.
Is he worth $16 million on a one-year deal? I doubt it. But at least we can see the Giants' thinking now.
Who is elite against the run and the pass?
There are some pass-rushers who really just rush the passer. Chandler Jones (Arizona) and Preston Smith (Green Bay), for instance, racked up double-digit sacks last season but didn't add much as run defenders, according to RSWR.
But then there are those who excel in both phases. Most notable: DeMarcus Lawrence (Dallas) and Joey Bosa (Los Angeles Chargers), who both rank in the top five in RSWR among edge defenders. The former recorded only five sacks last season, but he ranked third in pass rush win rate (PRWR) among edge rushers. Bosa had 11.5 sacks and ranked sixth in PRWR among edge defenders.
On the other side of the ball, David Bakhtiari (Green Bay) and Ronnie Stanley (Baltimore) were the two-way standouts, each finishing in the top five among tackles in both pass block win rate (PBWR) and RBWR.
Brian Burke contributed to this article.