Baltimore Ravens vs Atlanta Falcons — Win Probability

Model underestimated Baltimore; Ravens stun favored Falcons in Week 5

Baltimore entered Sunday as a 3.3-point underdog against Atlanta, and the model was even more skeptical. The simulation projected the Falcons to win by 3.2 points, with Atlanta scoring 23.4 and Baltimore just 20.1. That gave Baltimore only a 40% win probability—a clear lean toward the Falcons despite the market's smaller spread.

The Ravens executed a different game entirely. Baltimore won 30-23, securing a 7-point margin that flipped the projection on its head.

The gap between expectation and result: the model expected Atlanta to win by 3.2 points. Instead, Baltimore won by 7. That's a 10.2-point swing—a significant miss in both direction and magnitude. The model was wrong about which team would win and substantially underestimated how decisively wrong.

This qualifies as a clear upset. Baltimore's 40% pregame win probability means the model saw this as unlikely but plausible; the market's 3.3-point line was slightly more bullish on the Ravens. Neither forecast captured what actually happened. The Ravens' 30 points exceeded the projection by 9.9, while the Falcons' 23 matched their expected output almost exactly. Baltimore's offensive performance was the driving factor in the miss.

For a site built on factual precision, this game represents a notable blind spot. The model's read on team strength, game flow, or matchup dynamics fell short of the task. Whether Baltimore's execution, Atlanta's underperformance, or a structural model weakness caused the error remains a separate analysis—what's clear is that the simulation confidently leaned the wrong way in a game where the market itself was closer to the mark.