Tampa Bay Buccaneers vs Dallas Cowboys — Win Probability

Model Missed Badly on TB-DAL: Upset Win Widens Projection Gap

Tampa Bay's 24-16 victory over Dallas was a decisive upset that exposed a significant flaw in the pregame projection. The model gave the Buccaneers just a 24% win probability and expected Dallas to win by 9.5 points. The market agreed, listing Dallas as a 9.5-point favorite. Tampa Bay won by 8.

That 17.5-point swing between the expected margin (Dallas +9.5) and the actual result (Tampa Bay +8) represents a substantial miss. The model not only got the winner wrong—it got the magnitude of the outcome wrong in the opposite direction.

On raw scoring, the model projected Tampa Bay would score 19.5 points and Dallas 29. Tampa Bay actually scored 24; Dallas scored 16. The Buccaneers beat their projection by 4.5 points while the Cowboys fell 13 points short of theirs. That 17.5-point swing in the differential is where the upset lived.

This wasn't a close call that came down to a final field goal or a missed extra point. Tampa Bay controlled the outcome decisively enough to flip the entire game script. The model's 24% confidence in an upset didn't capture what the Buccaneers would actually do on the field, nor did it anticipate Dallas underperforming by such a margin.

For a brand built on factual accuracy, this game is a clear data point: the pregame model underestimated Tampa Bay's ceiling and overestimated Dallas's floor by a combined 17.5 points. Whether that reflects a systematic issue with how the model weighted the matchup or a singular miscalibration remains open, but the numbers here are unambiguous. Tampa Bay won bigger than any single projection suggested it could.