Josh Allen is the archetypal shotgun quarterback: 6'5", a moving pocket, a downhill runner. So it is a little jarring to look up and notice that the reigning MVP's offense has, in two seasons, been rebuilt around a formation that was supposed to be dying.

In 2023, Buffalo ran 43.8% of its handoffs from under center. In 2025, it was 75.1% — a 31.3-point jump that was the second-largest in the league behind Chicago (+33.4). Their under-center pass rate more than doubled over the same window, from 11.7% to 25.8%. The Bills went 13-6 in the regular season and finished with the league's second-ranked offense by EPA/play (+0.138), all while doing this against the prevailing analytical grain.

The shift, in three numbers

Buffalo's run-formation splits, 2023 → 2025:

Same run volume, radically different alignments. Buffalo added 191 under-center handoffs in two years without meaningfully changing how often they ran the ball.

The league moved in the same direction, but not remotely this fast. Leaguewide UC run share climbed from 47.4% in 2023 to 56.0% in 2025 — a real trend, but Buffalo's shift was 3.6× the league average. Only Chicago (+33.4) moved more; Jacksonville (+29.9), Baltimore (+28.2) and Indianapolis (+27.6) rounded out the top five.

Running against the analytics

Here's the part that reads against type. In 2025 the league still ran the ball better out of shotgun than under center:

• League shotgun runs: +0.024 EPA/play, 4.89 yards per carry
• League under-center runs: −0.032 EPA/play, 4.19 yards per carry

That's a 0.056-EPA gap per rush in favor of shotgun. The old orthodoxy — spread the box, hand it off, let the numbers work — still holds on average. Buffalo did the opposite, and their under-center run game outperformed the league from both formations:

• Bills UC runs: +0.084 EPA/play, 4.94 YPC (437 attempts)
• Bills SG runs: +0.187 EPA/play, 5.84 YPC (145 attempts)

Their shotgun runs were more efficient still — that gap hasn't gone away for them either — but the under-center version was already comfortably above the league shotgun baseline. In 2023, Buffalo's UC runs sat at just +0.012 EPA/play on 3.95 YPC. Two years later the same formation family gained nearly a full yard per carry and roughly seven times the EPA per attempt.

It's about the handoffs, not Allen's legs

The temptation is to credit Allen's rushing. That's not what the play-by-play says. Isolating Allen's own carries in 2025:

• Allen under center: 58 rushes, 3.90 YPC, +0.533 EPA/rush
• Allen shotgun: 57 rushes, 8.32 YPC, +0.528 EPA/rush

Allen split his own runs almost 50/50 between formations, and his shotgun scrambles/designs were far more explosive per carry. The under-center bet is on the running backs. Strip out Allen's carries and Buffalo ran 379 non-Allen handoffs from under center — 81% of the RB-carry pie — at 5.10 YPC and +0.016 EPA. The 88 non-Allen shotgun runs went for 4.24 YPC at −0.034 EPA. When the ball leaves Allen's hands, Buffalo wanted to be under center.

Passing followed the same door

The play-action bootleg universe that lives inside an under-center identity showed up in the pass game too. Buffalo threw 160 passes from under center in 2025 — more than the 79 and 90 of the two prior years combined — and posted +0.187 EPA/dropback on those plays. That's down from a small-sample 2023 (+0.358 on 79 attempts) and 2024 (+0.427 on 90) but on nearly double the volume, and it stayed comfortably positive while the offense as a whole finished second in the league in pass EPA (+0.164/play).

Put another way: Buffalo more than doubled the frequency of a play type and lost very little per-attempt efficiency doing it. That's the marker of a real structural change rather than a defense-picks-its-poison mirage.

What I think this means

Two seasons is a short window, and one team's success in a formation the rest of the league is (on average) still worse from doesn't rewrite the analytics. Chicago made a bigger jump and finished eighth in offensive EPA. But the Bills' version is the one worth watching because of who runs it. If the "shotgun-spread era" really is rolling back at the margins then the most interesting proof of concept is a Josh Allen offense that stopped looking like a Josh Allen offense and got better.