Per request, I thought I would share the quarterback ratings for this season. These ratings shift somewhat, but the rate at which they change depends on where the quarterback is on the aging curve. Basically, quarterback effectiveness changes most rapidly on either end of the career. This means that Blake Bortles and Derek Carr both gained rapidly in the ratings; we expect that there may be significant improvements at this stage in their career. Similarly, Peyton’s rating went into free fall when the stats made it clear he’s at the end of his run.
I’ve included two numbers that I think are a good snapshot of quarterback effectiveness.”NYA” is Net-Yards per Attempt, which is a metric that adjusts yards per attempt for sack yardage. TD/INT is the ratio of touchdowns to interceptions. These two numbers are the two best summary statistics for how good a quarterback has been.
The “Rate” figure is the number that matters in the PETE model. This is a linear adjustment; it’s a simple add and subtract process to switch between quarterbacks. These ratings are generated with a set of nested regressions that account for team context like pace and the running game.
The number takes the form of a simple “value above replacement” context, where 0 represents a league average quarterback. In some cases, the team backup is not listed in this table. In these cases, when there is an injury to the starter, I select a rating for the backup based on a nearest neighbor algorithm which selects the most similar quarterback based on attributes like starts, seasons of NFL experience, and draft selection.