Pro Football Focus has always had a few understandable issues with subjectivity. Grades are not hard and fast metrics, and they can occasionally be wrong as a result; however, I do believe PFF’s goals were both honorable and valuable. It is incredibly difficult to separate out performances in football, as running backs are only as good as the lines blocking for them, and even good corners can appear terrible without some semblance of a pass rush in front of them. PFF attempted to leverage time-tested
scouting processes to rank individual players regardless of their surroundings.
I can already hear people, including younger versions of myself, yelling that PFF has always been bad, but that’s really not true. Ben Baldwin, who helped to develop much of that analytics framework available to the public on advanced stats like EPA and CPOE, and who maintains RBSDM.com used to run statistical analysis of PFF’s quarterback grades versus more objective statistics, and PFF grades would generally come out on top in terms of predictive value, or at least very close too it.
And for positions like offensive line, where traditional stats don’t even really exist, PFF grades have always been useful, or at least better than nothing. But several years ago, PFF’s focus started to move away from the consumer-facing side of the house, and it sure seems as if their grades have gotten less attached to reality. Now, just because I think a grade is bad doesn’t mean the grade is necessarily bad. I’m certainly not anything close to perfect at this whole football thing. But we’re also starting to see some consistency issues within PFF itself, where their own internal statistical tracking really disagrees with the grading.
The most obvious example on the Packer side of things is Dani Dennis-Sutton, who is having a sensational rookie campaign, especially for a fourth-rounder. DDS has recorded 20 total pressures on just 143 pass-rushing snaps. The raw total ties him at 24th overall out of 150 qualifiers, and the only players who have more pressures in fewer snaps are Trey Hendrickson and Green Bay’s Brenton Cox. DDS also ranks highly in hurries, and he’s also only one of thirteen edge rushers with a pass broken up on the season.
Despite all of this, his 60.4 pass rushing grade is just 87th, and his overall grade is barely better. If you wanted to argue that DDS gives up some value on run defense, I am perfectly willing to believe you, but having him in the middle of the pack on pass rush is insanity. We here at Acme Packing Company pay special attention to Dennis-Sutton as he was the rare draft crush that actually ended up on the team, but we’re not exactly biased in terms of calling out Packer players when they are terrible players or people.
Another recent example is linebacker Ty’Ron Hopper’s grade from the Panther’s game on Thursday. Hopper is in for the injured Zaire Franklin and/or Edgerrin Cooper, and frankly, many of us didn’t think much of Hopper entering the game. He was great in every facet on Thursday; he filled up the stat sheet with a pick, a sack, and three passes defended. He was a monster. He was also the third-lowest-graded Packer defender ahead of only Javon Bullard and Carrington Valentine. Sometimes interceptions are lucky and just fall into your hands, but Hopper was in great coverage positions several times, a sure tackler, and most importantly for an off-ball linebacker, he showed incredible versatility. And yes, stats can be deceiving, but please check out this breakdown of his play by Tyler Brooke for an explanation of why the stats were, in fact, dead on.
I might be willing to listen to an argument that Hopper was just average and benefitted from the efforts of those around him. I still think that’s wrong, and that he was one of the best players on the field, but I would get it. Arguing that he was the third-worst Packer defender is ridiculous.
And that’s the problem in a nutshell. A good model will return results that are 80% exactly what you think they should be, and 10-20% surprises (that you need to double-check). If it’s too far off, the model is probably broken. I think we’ve reached a point where things are getting a bit too far off.
PFF also says that Tyler Shough and Geno Smith have been better than Jordan Love (AND Caleb Williams), with Love as the 15th overall QB and Williams at 13. Love’s statistics are down this year and based purely on numbers, ranking him 15th is somewhat justified, but the PFF ethos is supposed to be about seeing the Packers’ offensive line for what it is and giving the poor guy some extra credit. At this point, I’m not really sure what value PFF grades are adding. It’s a shame.
