Borac Banja Luka won this UEFA Conference League knockout tie 1-0, settling it early and then holding firm through a second half played against ten men. Savic broke the deadlock in the 14th minute, converting from Hrelja's assist to give the home side the lead they'd need. The match then turned decisively in Borac's favour when Vitebsk's Mohammadi was sent off in the 33rd minute, leaving the visitors to defend the rest of the way with a numerical disadvantage they couldn't overcome.

Our model had leant toward a Borac win with a 56% probability, though it'd picked a much livelier scoreline—a predicted 3-1. That direction was right, but the match unfolded far tighter than expected. Before kickoff, the model had flagged Borac's home form as a strength and Vitebsk's away record as volatile, alongside the knockout stakes that would sharpen both teams' focus. The early goal and the red card certainly changed the texture of the contest, compressing what looked like it might be an open affair into a tighter, more controlled affair once Vitebsk went down to ten. Borac managed the advantage well, and a single-goal margin held.

It's a reminder that even when a model calls the winner correctly, the actual path to victory can differ sharply from what the numbers suggested. Knockout football's binary nature—elimination on the line—makes it inherently less predictable than the league game, and a sending off early enough to reshape ninety minutes will always be a wildcard no pre-match model can fully price in.