Aaron Rodgers is having a renaissance season. As I touched on recently, he and Patrick Mahomes are in a two-way race for the MVP. Both are putting up ridiculous seasons in both traditional and advanced metrics. The case for which is having a better season is splitting hairs for the most part.
Our ability to evaluate players, but in this case specifically quarterbacks, is often limited by the data we have publicly available. Unlike baseball, which has a long and storied tradition in statistics, football does not. Baseball has multiple websites dedicated to the tracking and public availability of all sorts of advanced statistics. Sites like Fangraphs, Baseball Prospectus, and Baseball Reference provide many of the traditional and advanced metrics in common use, while MLB’s very own Statcast site runs through Baseball Savant — think of Statcast like the MLB’s version of NextGenStats. A big reason for the advances in baseball analytics over the past three decades has been the public availability of data. When data is publicly available, more people can work with it to come up with newer and more in-depth insights.
Football has a major problem with this. There have been great efforts put forth to try and overcome this. Unlike baseball, which has interactive websites that almost anyone can use, much of football’s useful data is buried either behind a paywall, requires a private contract to access, or requires programming knowledge that most fans do not possess.
That is why I am grateful for Football Outsiders, who still have maintained a publicly available DVOA database (you just need to create a free account). Ben Baldwin’s rbsdm.com has been very helpful for many folks who want to access EPA data without learning how to program. Sports Info Solutions offers some data in their publicly available datahub, even if most of the really good stuff is hidden behind a tens of thousands of dollars paywall. I would be remiss if I didn’t also include Josh Hermsmeyer from FiveThirtyEight, who also has tried to make his data as publicly available as possible. On the other hand, my issues with NextGenStats have been well-documented on my Twitter account for years now, as their database is not only largely trivial statistics, but often not even updated at appropriate times.
The best stuff is often not available. Every now and then though, we get a peek behind the curtain. This week was one of those times. I highly recommend that you follow Anthony Reinhard on Twitter, as every now and then you’ll get something like this that includes data from Sportsradar that normally costs a small fortune to obtain.
Ben Roethlisberger leads the league in both dropped passes and expected points lost due to dropped passes so far this season, which is driven mostly by the Steelers recent problems in this area. Since Week 9, they've dropped 17 passes costing an estimated 39.6 expected points. pic.twitter.com/609KGujldk— Anthony Reinhard (@reinhurdler) December 9, 2020
This is the good stuff. One of the major issues with evaluating quarterbacks is understanding the degree to which their supporting cast helps or harms them. A big talking point for Packers fans over the past two years has been the drops that Aaron Rodgers has had to endure. Of course, not every catchable ball will be caught. Drops happen. Some quarterbacks are harmed more than others, and that’s what we see in the chart above.
Anthony was kind enough to send me over the table that included the specific EPA losses by drops for each quarterback. Note that this also includes the expected yards after catch (or xYAC), which makes this much more useful. Each drop has its own potential. Dropping a pass when you’ve beat the coverage deep is much different than dropping a pass with someone between you and the end zone.
As you can see in the chart above, Aaron Rodgers has lost the third-most EPA to dropped passes this year. What happens when those numbers are added back in and put out on a per-play basis? If we use the EPA/play stats from rbsdm and the figures from above, Aaron Rodgers starts to really separate himself from the pack.
Rodgers and Mahomes are still in their own tier, with a large drop-off to play-action fiend Ryan Tannehill. One of my preferred measurements to use is the CPOE + EPA composite index from rbsdm. It takes into account accuracy but also impact. When combined into one chart, Aaron Rodgers starts to look like a more clear favorite as the best QB of 2020.
You’ll have to pardon the mess of quarterbacks in the above average category, but again, Rodgers outpaces Mahomes in both categories.
This mostly comes down to this: if the Packers replaced Marquez Valdes-Scantling with a deep threat with better hands, would Rodgers’ season stand out even more? MVS is 2nd in drop percentage of players with at least 40 targets, per Sports Info Solutions. To be fair to Mahomes, Mecole Hardman is 3rd, but on average his targets are seven yards shorter than Valdes-Scantling’s, so he loses far less in EPA.
The mainstream thought is that Mahomes is correctly the leading MVP candidate. I’m not sure that it actually is correct. In addition to drops, the Packers have faced a fairly tough slate of defenses so far, particularly compared to Kansas City. Per Football Outsiders, the Packers have faced the sixth-toughest schedule of defenses while Kansas City has faced the 26th toughest. Rodgers also narrowly has Mahomes beat in PFF grade at 94.7 versus 93.3.
The race hasn’t changed a lot since I last wrote about it, but as more data comes in, I increasingly become convinced that the wrong person is in the lead. If the season ended today, I think the more compelling case for MVP belongs to Aaron Rodgers.