Last year Buffalo Rumblings took a crack at predicting some of Buffalo Bills quarterback Josh Allen’s key stats based on his 2018 performance combined with a deep data dive (see below). As it turned out, the statistically based predictions turned out to be quite accurate to last season’s actual result. So let’s do it again.
Methodology
In the five measurements you’ll see below, the aim was to create a statistically based model of expected change. Specifically, this meant examining quarterbacks drafted in the ten years prior to Josh Allen (2008 to 2017) and determining the average amount of change that occurred from one season to the next. From there, the average change is applied to Josh Allen’s actual statistics from the prior year (2019) to predict the next year (2020).
In an attempt to create a better parallel for Josh Allen, only quarterbacks that started games in their rookie year were considered. To control for potential outliers, quarterbacks were removed from the list if they failed to log 50 or more attempts in each of their first three seasons.
For the statistically inclined the answer to your first question is “yes.” That does mean the sample group for this year’s look is smaller than last year. It’s no surprise that fewer quarterbacks made it to 50 passes three years in a row than would have made it through two seasons with that many attempts. Specifically, there were 37 players in the sample last year compared to only 27 this year. This does decrease confidence in the result.
Yards per game
:format(webp):no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/19750052/Year_2_to_3_projections___Ypg.png)
One of the other concerns that potentially arose from lopping off data for quarterbacks failing to hit that 50 attempt marker is that it should pull the results toward the positive end of the spectrum. This is based on the expectation that fewer “bad” quarterbacks would be allowed to come off the bench in year three.
If that’s true, it barely registers for our first metric. The average quarterback improved only 5.6 yards per game between year two and three, which is hardly noticeable. This change for Allen is represented as the “Perfect average” bar that you’ll see is still below 200 yards per game.
As we did last year, the standard deviation was calculated for each measure to produce a range of “normal” change. Let me know in the comments if you’d like a longer explanation, but the short version is that the bars labeled “low average” and “high average” represent the lower and upper limit of change that could still be considered “normal.”
For yards per game, there’s an incredibly volatile range of normal. Allen could improve all the way to 242.3 yards per game and it shouldn’t be shocking. He could also regress pretty badly and be within expectations. For fun, each chart adds the best- and worst-case scenarios. These are simply the amount of change each of the listed quarterbacks experience added to Josh Allen’s 2019 actual performance. Let’s hope Allen leans more toward Matthew Stafford’s end of the spectrum.
Yards per attempt
:format(webp):no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/19750082/Year_2_to_3_projections___YPA.png)
This chart should scare Bills fans. On average, third-year quarterbacks got a tiny bit worse when it comes to yards per attempt. This won’t make you feel any better either, but even at the “High average” outcome of 7.5 yards per attempt, Allen would have only been 14th best in the league. Thoroughly mediocre.
Matthew Stafford is the man you want to emulate again. If Allen were to hit 8.7 yards per attempt that should put him near the top of the league. In 2019 it would have put him in second place, directly ahead of...Matthew Stafford (8.6 Y/A).
Completion percentage
:format(webp):no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/19750128/Year_2_to_3_projections___comp_per.png)
Josh Allen’s biggest leap between year one and two was in this metric. Let’s hope for another big year as the average improvement was very slight. It’s good news that the projected possible decrease (low average bar) wouldn’t be a major drop off for Allen. A climb to 65% or so would be respectable.
Touchdown percentage
:format(webp):no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/19750151/Year_2_to_3_projections___TD_per.png)
This chart is not one to inspire a lot of hope. Not only did the average quarterback get worse in this measurement, the normal range shows things can get a whole lot worse. An increase to 5.4 would likely result in a top-ten ranking here, and this would still be considered a “normal” amount of change.
Nick Foles represents our worst case scenario but don’t worry. It’s not that he threw zero touchdowns. He had an excellent year two with 8.5% of his passes being a touchdown. He plummeted to 4.2, which gives us a worst-case scenario change 4.3% dropping off. Applying that to Allen’s 4.3% rate in 2019 gives us a clean zero here.
Interception percentage
:format(webp):no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/19750173/Year_2_to_3_projections___int_per.png)
Just a reminder, lower is better in this stat. Like we saw with touchdown percentage, the average quarterback change for interception percentage went the wrong way, with the perfect mean change being a 0.4% increase.
The high average seem lofty, whereas the low average seems more possible. Diving deeper though, a 4.0% interception rate would have been the second worst in the league this past season. Allen showed consistent improvement in this area during the 2019 season and hopefully this trend continues.
Summary
I hate to say it, but we really need to root hard for Josh Allen to do better than “normal” when it comes to how he changes from this last year to the next. Even the more optimistic end of normal for most of these measurements would lead to an average season. If Allen were to strictly improve the perfect average amount, it wouldn’t be improper to start questioning if the team needed to start over at the position.