Dallas Cowboys vs New York Giants — Win Probability

Giants Stun Cowboys as 8-Point Upset Defies Model Projection

The New York Giants pulled off a Week 1 upset over Dallas, winning 28-20 as a 41% win probability underdog in the model's pregame assessment. The victory marked a significant deviation from both the simulation's expectations and the market's read on the matchup.

Pre-kickoff, the model projected Dallas to outscore New York by 3 points (expected margin: -3 for the Giants), with the Cowboys favored at a 3-point spread. The model gave the Giants only a 41% chance to win. The actual result — an 8-point Giants victory — represented a 11-point swing from the expected margin, a substantial miss that places this well beyond the range of normal projection variance.

Dallas was the market favorite, and that preference appeared well-founded on paper. The model's expected scoring breakdown had the Cowboys projected to put up 25.8 points while holding New York to 22.7. Instead, the Giants outscored Dallas's actual output of 20 points with 28 of their own, suggesting the model's assessment of Dallas's offensive capability or New York's defensive strength — or both — was significantly misaligned with what transpired on the field.

For a model whose core function is to project game outcomes with precision, an 11-point swing between expected and actual margin is a meaningful miss. This wasn't a close game that fell just outside the confidence interval; it was a reversal of the projected winner and a substantial point total gap. The Giants' upset victory suggests either a material shift in team composition or performance level that the pregame model did not adequately capture, or execution on Sunday that exceeded baseline expectations. Either way, Dallas's failure to cover and the Giants' refusal to perform as a vulnerable underdog represents a clear divergence from what the numbers anticipated.