Adjusted Quarterback Efficiency: Daniel Jones Breakout Can't Be Ignored

A few more tweaks to AQE, but more on the margins than major revisions. Mahomes still head-and-shoulders above the rest, Daniel Jones breaking into the top-7

For this week’s AQE update, I didn’t make major revisions to the methodology like previous weeks. For now, I feel good about the number of contextual factors the metric adjusts for, and the overall magnitude of the adjustments.

Data enhancements

The last couple weeks I’ve been working with FTNData to expand the availability of time-to-throw and time-to-pressure numbers for dropbacks that don’t end in an actual pass, e.g. sacks and scrambles. They were super responsive getting that data into their systems, which gives better context for how long quarterbacks who run more often and take more sacks are holding the ball.

The “time-to-throw” metric can be either thought of as purely the time before an actual pass, or the total time a quarterback holds the ball before it leaves their hand, they play ends via sack, or they pass the line-of-scrimmage on a scramble. I prefer the expansive definition, mostly because it better captures each quarterback’s willingness to hold the ball under all circumstances.

We might think of scrambles as ways to turn sacks into positive plays, but the rates of each are positively correlated, i.e. players who scramble more often are generally sacked more often. It’s selection bias to talk about the positive effects of a scrambling quarterback in terms only of scrambles, which aren’t counted if the line-of-scrimmage isn’t breached, eliminating attempts that turn into sacks.

With this enhanced scrambling and sack timing data, a handful of the “Blocking” adjustments moved around, giving more or less credit to teams for protecting quarterbacks for longer when they’re extending plays.

Equalizing the effects

The other change I made for each adjustment was to make them all fully bi-directional, calculating the average positive or negative per play adjustment across the season for all quarterbacks, and then moving the adjustment for each quarterback’s factor adjustment up or down based on their play sample sizes.

This is really more tweaking around the edges, but I think helpful to be able to view each adjustment in the table and know that a positive number can be interpreted as above league average, and negative below.

2022 AQE results through Week 17

*You can find all the detailed data for AQE, in a downloadable format, in the Unexpected Points Google stat sheet for paid subscribers.