Dallas Cowboys vs Houston Texans — Win Probability

Dallas upsets Houston as model underestimated Cowboys in 34-30 shocker

The model projected Houston to win by 3 points and gave Dallas just a 41% win probability before kickoff. The market agreed, installing Houston as a 3-point favorite. Dallas had other ideas, winning 34-30 in an upset that exposed a significant gap between the pregame projection and reality.

The expected margin favored Houston by 3 points (model: 25.8 Houston, 22.8 Dallas). The actual result flipped that entirely—Dallas won by 4. That's a 7-point swing from projection to outcome, well beyond the range of normal variance.

What made this particularly notable: the model's win probability for Dallas (41%) and the market's spread (Houston -3) were aligned in their pessimism about the Cowboys. Neither expected Dallas to pull this off. Yet Dallas scored 34 points against a model projection of 22.8, a 11.2-point outperformance. Houston scored 30 against an expected 25.8, a 4.2-point beat in their own right, but it wasn't enough.

For a site built on factual honesty, the important observation is straightforward: the model significantly underestimated Dallas's offensive capability in this matchup. The Cowboys didn't just win—they won convincingly despite being underdogs in both the model's simulation and the market's assessment. Houston's 0-4 record after this loss and Dallas's 2-2 mark underscore the real-world stakes.

This was not a close miss. A 7-point margin error is substantial. The model saw Houston as the clear favorite and Dallas as the underdog; instead, Dallas controlled the game enough to win by a field goal. Whether this reflects a personnel adjustment, schematic advantage, or simple execution edge on the day, the gap between projection and outcome was too large to ignore.