Tennessee Titans vs New York Giants — Win Probability

Giants' Narrow Win Fits Model Projection, but Scoring Fell Short

The New York Giants defeated Tennessee 12-7 in Week 3, a result that validated the model's directional read but revealed a significant scoring gap.

Pregame, the model projected a Giants win by 2.5 points with a 57% win probability—a lean toward New York rather than a strong conviction. That aligned precisely with the market's published spread of 2.5 points, suggesting the two assessments were in sync.

What actually happened came closer to the model's pick than the margin alone suggests. The Giants won by 5 points, not 2.5, so the model's win call was correct and the upset flag did not trigger. Tennessee was favored to stay winless at 0-3, and did. But the offensive output from both teams fell well short of expectations.

The model had projected a combined 37.5 points (20 for New York, 17.5 for Tennessee). The game delivered 19 total points—a 18.5-point swing downward. The Giants managed exactly 12, a full 8 points below their projected 20. Tennessee scored 7, 10.5 below their 17.5 projection.

This is a rare case where accuracy in the outcome—picking the right winner and roughly the right direction—was paired with a major miss on how the game would be played. Neither offense approached expected efficiency. For a model built on scoring projections, this gap matters for calibration, even though the win probability nailed the result.

The 5-point margin exceeded the 2.5-point spread, so the Giants covered more ground than either the model or the market anticipated. But that outperformance came entirely through defensive/special teams play, not superior offensive execution. The model saw a close game and got a close game—just a significantly lower-scoring one.