Baltimore Ravens vs Cleveland Browns — Win Probability

Model Underestimated Cleveland in 13-10 Upset Over Baltimore

Cleveland's 13-10 victory over Baltimore on Sunday was a clear upset, and the pregame model significantly underestimated the Browns' chances. The model projected Baltimore to win by 2.9 points, with the Ravens favored at a 59 percent win probability. The market agreed, posting a 3-point spread in Baltimore's favor. Neither projection accounted for what actually unfolded.

The Ravens scored only 10 points against a projection of 21.2, a 11.2-point shortfall. Cleveland managed 13 points against an expected 18.3, a 5.3-point miss, but crucially in the winning direction. The model's expected margin of -2.9 for Cleveland was off by 5.9 points—a meaningful gap in a low-scoring game.

Where the projection failed most was in Baltimore's offensive output. The model expected the Ravens to generate more than 21 points; they produced less than half of that. Cleveland's defense performed far better than simulated, while the Browns' offense, despite coming in under the projected total, did enough when it mattered.

The upset wasn't a coin-flip result hiding behind normal variance. A 3-point swing in one direction would have aligned with the projection; this was a 5.9-point gap. Both the model and market properly identified Baltimore as the more likely winner—Baltimore's 59 percent win probability was legitimate—but the magnitude of Cleveland's advantage on the field was underestimated.

After the result, both teams sit at 4-2. The model's error here centered on Baltimore's scoring potential, not on assigning the game to the wrong team. For a low-scoring outcome, however, that distinction matters less. The projection was substantially off.