Pittsburgh Steelers vs New England Patriots — Win Probability

Model Underestimated New England Dominance in Blowout Over Pittsburgh

The model projected a New England victory with a 5.4-point margin and gave the Patriots a 66% win probability before Sunday's game against Pittsburgh. The market agreed on the direction—New England was favored by 5.5 points—but the actual result was a decisive 17-point rout that exposed a significant gap in the simulation.

New England scored 20 points and held Pittsburgh to 3, producing a margin that was 11.6 points larger than the model's expected spread. The Patriots' pregame projection had them scoring 23.5 points; they finished 3.5 short. Pittsburgh fared much worse than anticipated, scoring only 3 points against a projected 18.1—a 15-point miss in the defensive direction.

The model correctly identified New England as the likely winner, but by a substantial degree it underestimated the quality of that victory. A 5-point win and a 17-point win are fundamentally different outcomes in football, and this gap suggests the model's pregame assessment either overstated Pittsburgh's offensive capability or understated New England's defensive strength. The market line of 5.5 points landed nearly identical to the model's 5.4-point projection, meaning both the in-house simulation and the professional betting market misjudged the matchup by roughly the same margin.

Both teams enter Week 3 at 1-1. The model's directional call—favoring New England—proved sound, and the 66% win probability was well-calibrated to a game that unfolded as a clear home victory. However, the magnitude of the miss (11.6 points wider than projected) represents one of the larger forecast errors available from this weekend's slate and warrants review of the underlying assumptions about both defensive pressure and offensive execution in this particular matchup.