Detroit Lions vs Carolina Panthers — Win Probability

Panthers' 6-Point Upset Over Lions Exposes Model Miss on Detroit

Carolina arrived in Detroit as a clear underdog by every measure. The market favored the Lions by 4 points, and Matrix's pregame model gave Detroit an even wider edge—projecting a 27.7–23.8 Lions win, a 3.9-point advantage. The model assigned Carolina only a 38 percent win probability.

The Panthers won 32–26.

This was an upset, and by a significant margin relative to the projection. The model expected Detroit to win by 3.9 points. Instead, Carolina won by 6 points—a 9.9-point swing from forecast to result. That swing is large enough to represent a genuine model failure on this matchup, not measurement noise.

Carolina outscored the pregame expected scoring by 8.2 points. Detroit fell 1.7 points short. The gap wasn't driven by one team's explosion; both offenses moved the needle in the same direction—sharply away from the model's read.

The market's 4-point line proved closer to reality than the model's implicit 3.9-point projection, though both instruments got the direction of the game wrong. Neither anticipated a Carolina win at all, much less by 6. For an upset of this magnitude, the model's 38 percent win probability for the eventual winner represents a material underestimation of the Panthers' true chances.

This game illustrates the difference between a miss that looks unlucky in hindsight—where the favorite was favored for sound reasons and got beaten by variance—and a miss where the model's assessment of the underlying matchup appears to have been simply wrong. The 9.9-point differential from expectation here falls into the latter category. Both teams finished 2–2 after the game, underscoring that the upset did not emerge from one team being dramatically over- or under-talented relative to record.