By BRYCE ROSSLER
After Aaron Donald won his third Defensive Player of the Year award in four seasons, another three-time DPOY took to Twitter to plead his case for another player: his younger brother.
Aaron Donald is an absolutely incredible player. I love watching him play & he’s headed to the Hall of Fame without question. This has nothing to do with AD personally.
This is me saying what my brother won’t.
TJ played 1 less game and STILL led the NFL in every major category. pic.twitter.com/m1vzrD88WU
— JJ Watt (@JJWatt) February 7, 2021
Most Steelers fans cited T.J. Watt’s slight edge in sacks, but sacks don’t tell the whole story. And that’s probably why his big brother included pressures in his statistical appeal. Although the younger Watt had only a 1.5 sack lead over Donald, the former’s unsourced pressure numbers dwarfed the latter’s. This might be compelling when you consider that pressures are more predictive of sacks than sacks themselves, but we mustn’t forget that Watt and Donald play different positions.
Setting arguments about positional value aside, it seems intuitive that Aaron Donald is a better interior defender than T.J. Watt is an edge. And now we can measure that in a straightforward way with our Pressures Above Expectation (PAE) stat.
2020 Leaders, Pressures Above Expectation
(minimum 200 plays, regular season)
Player | Pressures* | xP | PAE | Plays |
Aaron Donald | 57 | 34.7 | 22.3 | 460 |
Chris Jones | 47 | 27.3 | 19.7 | 343 |
T.J. Watt | 59 | 42.4 | 16.6 | 397 |
DeForest Buckner | 43 | 28.1 | 14.9 | 396 |
Shaquil Barrett | 52 | 39.7 | 12.3 | 361 |
John Franklin-Myers | 36 | 23.8 | 12.2 | 296 |
Leonard Williams | 43 | 32.6 | 10.4 | 405 |
Cameron Heyward | 41 | 30.7 | 10.3 | 415 |
Stephon Tuitt | 39 | 29.1 | 9.9 | 409 |
DeMarcus Lawrence | 40 | 30.4 | 9.6 | 295 |
*Screens, RPOs, and plays with a QB hit but no hurry or knockdown are excluded
Those of you who are statistically-inclined are probably familiar with the basic concept of expectation-based metrics like Completion Percentage Over Expectation (CPOE). For the uninitiated, these metrics are just a way to contextualize performance based on what actually happened on the play.
There are no black boxes here; everything is based on common football sense. We know that it’s harder to complete a pass 40 yards downfield than it is to flick the ball to a running back wide-open in the flat. And just as CPOE accounts for play-level factors like these, PAE operates similarly for pass rushers.
Completions are not created equally, and neither are pressures. It’s harder for a nose tackle to record a pressure on a first down play action than it is for an edge defender to record a pressure on 3rd & Long when they suspect a pass is coming. You already know that. We’re quantifying it.
PAE considers the quarterback’s drop type, the down and distance, the score, the use of play action, and each defender’s alignment to determine their likelihood of recording a pressure, which we call Expected Pressures (xP).
We then measure the outcome of the play against that xP. That means if a pass rusher had a 20% chance to pressure the quarterback and did so, they’re rewarded with 0.8 PAE for that play because a pressure represents 100% (100%-20%=80%, so .8). The math and methodology here are very simple.
PAE and xP combined are more predictive of pressures than pressures themselves. Put more simply, efficiency and opportunity allow us to predict future performance better than past performance can.
This is not to say that pressures are now obsolete. Just as sacks weren’t discarded when we collectively learned that pressures were more predictive, we shouldn’t discard pressures because of the findings we’ve outlined. We can, however, leverage these new statistics to continue pushing forward our quantitative understanding of pass rush.