Joe Burrow

My QB1 on value, and Josh Allen outscores him by 34 points. 12.8 games means the weeks he misses get streamed at replacement, and the roster spot still yields more than anyone else's — that's VOR doing exactly what it's built for.

Rate anchored on his own record, 22 Aug, at Matt's read that Burrow and Nix are the wrong way round. THE MODEL CHARGES HIM FOR THE SAME INJURIES TWICE. His games are cut to 12.7 off a real 51-of-68 availability record across 2022-25, which is fair. But his RATE is then dragged to 18.31 by the two seasons he spent hurt -- 2023 (10 games, 15.52/g, wrist) and 2025 (8 games, 17.43/g, injured in week 1) -- while his two healthy seasons read 22.67 and 23.05. The games term already prices the absence; the rate should not price it again by averaging in the weeks he played injured. 19.72/g is his own 2023-25 at the model's QB decay of 0.80, games-weighted. It is deliberately NOT his healthy rate of ~23: some of the games he does play, he will play hurt, so a blend is right -- just not one sitting 1.41 below his own weighted history. For contrast the same calculation puts Bo Nix at 18.61 against a model 18.86, so the model is 0.25 ABOVE Nix and 1.41 BELOW Burrow, a 1.66/g swing on rate alone. 250.5 = 19.72 x 12.7. He clears Nix on VOR (42.6 against 38.2) while scoring 38 FEWER POINTS, and that is the streaming definition doing exactly what it is for: a roster spot yielding 12.7 elite games plus replacement is worth more than 15.3 merely good ones. Lands QB2. GAMES NOT TOUCHED, on purpose. 12.7 is a faithful read of a genuinely fragile record and is not the thing in dispute; the rate is.

This is a view, not a correction — the model said something different and was overruled. See how it works.

Pace extrapolates the selected weeks over 17 games. It is what that stretch was worth, not a projection — no availability, matchup or regression adjustment is applied.