This week, PFF released their 2020 quarterback rankings. These rankings use a combination of PFF’s own grades and “estimated points added” numbers to model the range of outcomes and likelihood for every NFL starting quarterback.
Additionally, they use Bayesian updating, which is basically a strategy that uses previously known information to predict future outcomes. In this case, a player’s draft position plays a role in their ultimate outcomes, but the model weighs recent events more heavily. This latter point is true season-to-season and game-to-game. I.E., Baker Mayfield‘s performance in the final game against Cincinnati in 2019 is more important to the model than the one against Tennessee.
Below are PFF’s predicted rankings with error bars extending from the 10th to 90th percentile.
Should We Care?
In short, yes. Bayesian updating is a pretty common technique in forecasting and the model’s results meet the eye test.
Ostensibly, PFF projects the most likely range of outcomes for Baker to be a moderate net negative to a borderline top-five quarterback. Nonetheless, the most likely scenario is that Baker is a very slight net positive.
What Do I Believe?
I find it hard to believe that Baker does much worse than his mean outcome. As I have written previously, Kevin Stefanski’s offense emphasizes traits Baker has demonstrated massive success with in the NFL. Thus, Baker Mayfield should look like the best version of himself.
The question is, as the sample size increases, will Baker perform at his mean outcome or better? Another PFF study gives me hope.
In this study, PFF uses principal component analysis – a technique for condensing large amounts of data using generic distance and relatedness between observations. They find:
a relationship between difference in scheme and difference in offensive performance, which is enhanced when there is a play-caller changehttps://www.pff.com/news/nfl-pff-data-study-how-scheme-changes-influence-offensive-performance
Other work from PFF finds that outside zone schemes are successful since their introduction in the 1990s.
Ultimately, I believe the change to a more effective and friendly scheme means that Baker Mayfield’s average outcome is slightly higher than that projected by PFF. In essence, the coaching change could regress Baker upwards compared to where he was last year.
Nonetheless, the nature of a “mean outcome” is that neither side of the distribution of outcomes is more likely. It is very possible that Baker regresses downwards and the Browns are, once again, looking for a starting quarterback.