In 2019, Buffalo quarterback Josh Allen averaged 5.71 Adjusted Net Yards per Attempt, which ranked 23rd among all NFL quarterbacks. The league average was 6.16 ANY/A in ’19, which means Allen was a bit below average as a passer: he averaged -0.45 ANY/A relative to league average.

This year, with two games to go, Allen has significantly improved. He is averaging 7.55 Adjusted Net Yards per Attempt, and will likely finish at least in the top 8 in the pass efficiency rankings. He is averaging +1.53 Relative ANY/A, given that the league average through 15 weeks is 6.40.

Allen is just 24 years old. Last year, he had a below average ANY/A, and this year, he’s increased his Relative ANY/A by at least +1.25. So how will Allen do next year? On the one hand, it’s easy to think that a young quarterback who makes big strides is on the way up, and will only get better. But on the other, we know that regression to the mean is a strong force, and somehow who makes a huge jump year to year probably benefited from some unsustainable things going his way. On the, uh, third hand, how do we know what Allen’s mean even is?

So let’s try to investigate by finding somewhat comparable examples throughout history. For Allen, we are going to label 2020 his Year N, which would make 2019 his Year N-1 and 2021 his Year N+1. I looked at all quarterbacks since 1990 who:

• Threw at least 224 passes in Year N-1, Year N, and Year N+1.
• Were below average in Relative ANY/A in Year N-1.
• Improved by at least 1.25 RANY/A from Year N-1 to Year N.
• Were 26 or younger in Year N

Those filters should identify quarterbacks who were full-time starters, young, and made a big improvement in one season from a starting point that was below average. Now, this methodology does have a small bit of survivorship bias which would work in Allen’s favor: four quarterbacks who would otherwise qualify are not included: Steve Walsh, Josh Freeman, J.P. Losman, and Craig Erickson all otherwise qualified but failed to throw enough passes in Year N+1 to make the cut. They arguably should be included, and would make the prospects for Allen’s 2021 look worse. ^[1]The Freeman/Losman examples are the most interesting. Josh Freeman was actually on here twice; as a rookie, he ranked 29th in ANY/A, but then shot up to 6th in his second season! In his third year, … Continue reading It’s always possible that Allen doesn’t throw enough passes to qualify in 2021, but it would presumably only happen due to injury. And there is one positive quarterback who is also excluded: Dak Prescott (2019) would otherwise qualify and had a great Year N+1 (i.e., 2020), before getting injured and just falling shy of the minimum cutoff.

So let’s looking at those who did qualify. Since 1990, there are 21 quarterbacks ^[2]Excluding those who did not play much in Year N+1 who were “kind of like Josh Allen will look entering the 2021 season”; that is, young and just coming off a season of huge improvement. One of those, in fact, was Josh Allen prior to 2020. He was very young entering 2020 and just coming off a season where his RANY/A improved by 1.50. In looks like he is going to respond by… improving again by about 1.50! But that is rare: it was the second largest improvement in Year N+1 in this sample. Overall, 9 of the 21 quarterbacks improved in Year N+1, and 12 declined. More importantly, the declining group had a much bigger dropoff, with all 12 declining by at least 0.50 RANY/A. In fact, overall, these quarterbacks went from about -1.23 Relative ANY/A the year before the breakout, +0.81 RANY/A in the breakout year, and then back down to +0.27 RANY/A in the follow-up season. Here is the full list. Someone like Carson Wentz is an easy cautionary tale: he went from -1.12 RANY/A in 2016 to +1.53 RANY/A in 2017 back down to +0.57 RANY/A in 2018.

What does this mean for Allen’s 2021 campaign? Not much, of course. He is his own unique player. But it does mean that young quarterbacks who improve a lot don’t always improve just because they got better: a lot of times they have a great year for the same reason any quarterback has a great season. Some bounces go their way, players stay healthy, the schedule is easier, some interceptions are dropped, some long touchdowns are completed, and so on. In Allen’s case, adding Stefon Diggs has obviously been a huge boon to his production, which should be a good thing for Allen assuming Diggs stays healthy next year. But in short, when a player has a great season, usually some unusual things go their way and those are unlikely to repeat the next year. Even for young players.

Now, what about Allen? Are there signs that his play this year is unsustainable? Not in the sort of ways you might think. For starters, his interception rate is good but not a large outlier. None of Allen’s statistics are huge outliers: he’s simply producing above-average statistics across the board. His best stat is probably his completion percentage, which is actually a pretty stable statistic year to year. One thing that’s unique about Buffalo’s offense is over 82% of the team’s receiving yards have gone to their wide receivers; no other team is at 80%.

Overall, I’d say the signs are positive that Allen’s success is sustainable. But studies like this help to reign in the optimism that is all too easy to generate with young players. Allen has exactly 4,000 passing yards and 30 passing touchdowns this season, with two games left to play, after having about 3,000 passing yards and 20 passing touchdowns last season and about 2,000 passing yards and 10 passing touchdowns his rookie year. No, I don’t think 5,000 passing yards and 40 passing touchdowns are coming next year: if I had to guess, I’d say Allen will finish lower in the ANY/A rankings in 2021 than he did in 2020, but higher than he did in 2019.

What do you think?