Lyon 3–0 Sparta Praha: AI predicted 2–1 ✓

Synthesis: Lyon Favored but Fragile

All five expert models converge on a Lyon home win, with four of five predicting a 2-1 scoreline. This convergence is meaningful but not self-validating — the models share a known favourite-leaning bias, and the market's 1.48 price (≈63–64% fair) is noticeably more aggressive than every expert's calibrated estimate (56–59%).

The tie state is the real architect

Sparta Praha carry a 2-1 aggregate lead from Prague, where they were genuinely the better side — 62% possession, three big chances to one, and their goals came from open play while Lyon's only goal was a Suchomel own goal. Tolisso's missed penalty at 1-1 kept the tie alive, but it also confirmed that Lyon's attacking machinery was malfunctioning even before accounting for the absences of Šulc, Merah, and Bidstrup — the creative spine Fonseca needs to unpick a low block. Sparta, by contrast, need only a draw to advance and can play the game they want: compact, patient, countering into the space Lyon's advancing full-backs inevitably leave.

Lyon's defensive crisis is the hardest fact in the dataset

The single most reliable data point across all models is Lyon's concession record: at least two goals conceded in five consecutive competitive matches (2, 4, 2, 2, 2). This is not a dip — it is a structural openness that persists across different opponents, formations, and game states. Niakhaté's return from suspension helps, but Tagliafico remains banned, leaving the same left channel that Ryneš exploited in Prague still improvised. Lyon will score at home — they have 7 goals in their last two home wins — but the question is whether they can keep Sparta off the scoresheet, and the evidence overwhelmingly says they cannot.