West Ham vs Everton

West Ham's 2-1 victory over Everton on Saturday proved a sobering miss for our pre-match model, which had backed an Everton win with a predicted scoreline of 1-2. The Hammers struck first through Tomáš Souček's 51st-minute finish, courtesy of Jarrod Bowen's assist, before Everton leveled through Kiwior Dewsbury-Hall's 88th-minute header off John Tarkowski's delivery. Crucially, West Ham found a late winner in the 90th minute when Crysencio Wílson converted another Bowen assist, securing three points in a finish that belied the underlying patterns we'd flagged beforehand.

Our analysis had weighted the prediction heavily toward Everton (56% win probability) based on their superior win rate at that stage of the season—40% versus West Ham's 30%—and a historical pattern of low-scoring encounters between these sides. The pre-match context suggested Everton's mid-table position offered more stability than West Ham's relegation-zone desperation, a logic the model applied with reasonable confidence. What we underestimated was West Ham's capacity to impose themselves late in matches despite their documented struggles with conversion efficiency. Bowen's two assists demonstrated the kind of creative output that conversion metrics alone don't fully capture when clinical finishes materialize.

The late drama also highlighted a limitation in our approach: the model had flagged BTTS as plausible given West Ham's attacking urgency in a relegation battle, yet assigned low probability to their outright victory. Saturday's result suggests we may have overcorrected for historical caution when assessing sides under genuine pressure. West Ham's desperation proved a more potent force than their underlying form indicators implied.