Everton and Manchester City served up a six-goal spectacle that bore little resemblance to the controlled affair our model had anticipated. Jérémy Doku's 43rd-minute opener seemed to set City on course for another clinical victory, but the hosts mounted an unlikely comeback through goals from Tomiás Barry (68', 81') and Jarell O'Brien (73'). The momentum swung decisively in Everton's favour until City reasserted themselves late on, with Erling Haaland equalising at 83' and Doku completing his brace in the 90th minute to secure a 3-3 draw.

Our prediction of a 1-2 City victory missed the mark entirely. The model flagged City's defensive solidity (0.81 conceded per game), Everton's blunt home attack, and a historical trend of low-scoring encounters between these sides, all pointing toward an Under 2.5 situation. What materialised instead was a breakdown in defensive structure from both teams, particularly City's typically impenetrable rearguard. Everton's spirited second-half performance defied their inconsistent form and mid-table lethargy—the very factors we'd cited as evidence of reduced motivation. City's title-race intensity clearly proved insufficient to close out proceedings despite leading twice.

The draw represents a significant variance from our probability assessment, where we'd assigned City just an 11% chance of dropping points. Everton's resilience and City's uncharacteristic defensive lapses transformed what looked set to be another demonstration of City's superiority into an unexpectedly open contest. For our model, this match underscores how high-performing teams can occasionally deviate from their established patterns in ways standard metrics struggle to capture.