Gornik Zabrze 1–1 Ferencvarosi TC: AI predicted 2–1 ✗

A knife-edge second leg where the aggregate state shapes everything

All five models converge on a Górnik Zabrze home win, but the confidence levels — ranging from 40% to 46% — tell the real story: this is a near three-way toss-up with a slight home tilt. The agreement is notable but not convincing, given these models' shared tendency to lean toward favourites and the fact that their 30-day result accuracy hovers around 47-48%, barely better than chance in a three-way market.

Where the experts agree and why it matters

The consensus rests on four pillars: Górnik's home advantage and raucous crowd at the Ernest Pohl Stadium, their solid domestic form (3-1 at Radomiak, 2-1 vs Śląsk), Ferencváros' worrying domestic start (4-2 loss at Paks, 0-0 vs Vasas), and the structural imperative that Górnik must score to keep the tie alive. GPT-5.6's point about rotation is particularly sharp — Górnik held back key attackers (Sadílek, Ismaheel, Prekop, Ikia Dimi) for the weekend, and Peter González's two-goal haul at Radomiak gives manager Michal Gašparík a genuine selection upgrade over the blunt front line that failed in Budapest.

Where they diverge — and the sharper read

The scoreline split is instructive: Grok, Claude, and GPT predict 1-0 (a result that only forces extra time), while GLM and Kimi predict 2-1 (which would still only level the aggregate). The 1-0 camp is better supported by the evidence. Claude's analysis — weighted highly given its 55% recent result accuracy — identifies the core problem: Górnik's attack is their weak link, producing only 8 shots per game in Europe, with last season's top scorer Liseth goalless this term. They scored once in three European matches against serious opposition. GLM and Kimi's BTTS assumption relies on Ferencváros scoring, but the Hungarian side's incentive structure — defending a 1-0 lead with a Conference League safety net — argues for a more conservative approach than their domestic leakiness suggests.