Santos came away with a 2-1 win over Cruzeiro in a match that didn't go to script. Our model had leaned towards a 1-1 draw—the most likely outcome at 36%—but Santos went on the offensive early and secured the points through goals in quick succession. Benjamín Rollheiser broke the deadlock in the 28th minute, then Fabrício Bruno's own goal five minutes later put Santos two up before half-time. Cruzeiro pulled one back through Matheus Pereira in the 90th minute, assisted by Luciano Rodríguez, but it came too late to change the result.

Before kickoff, our model had expected a tight, low-scoring affair. The underlying metrics—xG readings that were nearly level between the sides—suggested both teams would create chances at a similar rate, and a draw felt like the natural resting point. Santos' win probability sat at 37%, which meant the model wasn't ruling them out, just seeing it as less likely than a share of the spoils. What happened on the pitch was different: a more decisive Santos performance that capitalised on early opportunities and held firm despite Cruzeiro's late rally.

The model missed on the result direction here, plain and simple. That's football, though—narrow probability leans can and do get it wrong. The match itself unfolded in a way the pre-match shape didn't quite anticipate, and that's a reminder that even well-balanced expected goals figures don't guarantee how a game will actually play out.