There was a brief moment of time on Twitter earlier this week when the topic of advanced metrics in football was trending. The subject trended sometime before Edward Snowden’s Twitter debut and after everyone went to bed gushing over Aaron Rodgers’ Monday Night Football performance. Everyone except Pro Football Focus (PFF), that is. Their evaluation of his stat line fell somewhere between room temperature and lukewarm, and people were outraged.
@PFF “Everyone in this room is now dumber for having listened to it. I award you no points, and may God have mercy on your soul.”
— Matthew Parvis (@MatthewParvis) September 29, 2015
Noted NFL media personalities took notice.
In a result obvious for everyone who watched #Packers QB Aaron Rodgers last night, he scored a -2.3 rating from @PFF pic.twitter.com/C3b5o9AS3M
— Ian Rapoport (@RapSheet) September 29, 2015
And there was even some LOL-ish input.
I know @PFF is smart and all, but I suspect the Packers trading away Rodgers is a knee-jerk reaction they’ll eventually come to regret.
— Dave Dameshek (@Dameshek) September 29, 2015
The crux of it all was that the -0.8 grade PFF gave Rodgers for his 68.6%/333/5/0 slash line was the final piece of proof that advanced stats are hogwash. Never mind the fact that PFF wrote a piece explaining their grade, and also offering up some of their faults in not being able to capture everything. The issue at hand is the resistance towards accepting advanced statistics into our nomenclature because they aren’t what we’ve been attuned to. It’s hard to give PFF’s report any credence when Rodgers’ stat line translates to 50 fantasy points or makes someone the newest millionaire in DFS.
To be clear, I’m not agreeing with PFF that Rodgers was in the bottom half of NFL quarterback performances this weekend. I’m highlighting that it is a long, and maybe endless road, ahead for the adoption of advanced stats. First of all, the way these numbers are presented play a huge role in how we interpret them. In this case, people saw a negative and assumed it was a bad grade, but their grading scale rates this negative number as average. Unlike QBR (0-100 with 50 being average) and passer rating (0-158.3), PFF’s grade ranges from a low single digit negative number (Rodgers’ -0.8) to a high of a single digit positive number (Carson Palmer’s 7.2 this week). The uproar was less about saying Rodgers was bad, and more about the negative sign in front of it. Case in point, ESPN’s QBR for Rodgers was 78.0, just enough for 9th best QB of the week behind Luke McCown, but no one made any fuss over that.
Second of all, most people are going to see the quick graphic on Twitter and dismiss the explanation behind it, and use that to make their assumptions on the faultiness of advanced stats. The fact is stats like these are just way too “inside baseball” for the casual fan to care about. “The other three touchdowns, however, were passes thrown short of the end zone on speed outs to Randall Cobb. Were they bad throws? No, they were expected throws with the credit going to Cobb for fighting through contact or defeating the coverage with speed to the edge. That makes these zero-graded throws.” The previous sentence is an excerpt from PFF’s justification of Rodgers’ three TD’s to Cobb. Something tells me the barstool fan doesn’t take into account speed outs, contact, or edge speed into their judgement of an NFL performance.
Vice Sports covered this topic in part on Monday when talking about MLBAM’s Statcast. In short, Statcast can tell the number of times a pitch rotates, defensive angles of pursuit and average leadoff distance among hundreds of other things. In other words, the same level of minutiae that PFF looks at in their evaluations. Towards the end, the article talked about the presentation of Statcast data and how they hope it eventually gets used. It talked about how MLBAM is planning on releasing this data into broadcasts from a fundamental level (home run distances) to more advanced levels (exit velocity) as the viewing audience becomes more comfortable with the data. The PFF data is on the advanced end of the curve and did no one any favors by going viral without any context or introduction behind it.
Lastly, I’d be willing to bet this leads to an adjustment in their grading system. No one thinks that Rodgers was in the minority of QB performances PFF put him in this weekend, but advanced stats hint he wasn’t a lock for the top three either. So the answer, as often happens, probably lies somewhere in between. Smart, reasonable people at PFF who are passionate about this stuff will undoubtedly be tinkering with their formula to make sure they have it right, and adjust the metrics if it isn’t. This would at the least show flexibility and adaptability to faults in their logic, and acting on it to correct any mistakes. The feedback from the initial report shows an overwhelmingly ignorant and stubborn response from the general public. While I don’t advocate for every sports fan to consume stats like rocket scientists, I hope we become more open-minded to the advancements in sports data and less reliant on the same old metrics that we’ve been force fed over the past 50 years.