Tottenham came from behind to beat Coventry 3-1 in a match that painted a very different picture to what we'd sketched beforehand. Maddison opened the scoring early on with an assist from Kudus, but Coventry hit back in the second half when Rudoni levelled things up courtesy of Thomas-Asante's pass. That's when Tottenham took control. Marmoush restored the lead shortly after with Robertson setting him up, and Davies added a third deep into stoppage time to seal it, finishing off a move started by Solanke.

Our model leant toward a Tottenham win at 48% probability, but we'd backed a specific scoreline of 1-0 and got that wrong. We'd flagged the stakes as sky-high for both sides locked in relegation trouble, which pointed toward caution and a draw mentality from the players. The form data suggested a scrappy, low-scoring affair. What actually unfolded was a more open game—three different Tottenham scorers, Coventry dangerous enough to level it briefly, and an extra goal in injury time to seal the points. The match didn't play out with the defensive solidity we'd anticipated, though the eventual result direction—a Tottenham win—landed where our model's strongest lean sat.

When a scoreline misses by as much as ours did here, it's worth a reminder that single-goal predictions are inherently tight calls in football. We read the situation one way, the game wrote it differently, and that's the nature of the sport. Tottenham got the win, Coventry went home empty-handed, and the story moved on.