Fipster got the big names right, but the round belonged to the ones it did not see coming

The round started with a fairly sensible idea of who would win, then spent the evening reminding us that sensible is not the same as complete. Fipster landed two exact scorelines and got enough of the home favourites right to keep the ledger respectable, but several of the loudest moments came from players it had not called at all. That is usually where a model’s neat little picture gets fingerprints on it.

Liverpool and Real Madrid were the cleanest reads

The best call was Liverpool against Atletico Madrid. Fipster had it at 2-1, and that is exactly how it finished. More importantly, the read behind it held up: a narrow home win rather than anything extravagant, with no need for a parade of scorers to make the point. In a round where several matches ran away from their pre-match shapes, that one stayed close enough to the model’s centre of gravity to count as proper work rather than luck in a smart coat.

Real Madrid against Inter was another exact hit. Again, Fipster settled on 2-1 and again that was the final line. The value there was not just in picking the winner, but in trusting a tight game rather than overcomplicating it. Sometimes the model earns its supper by resisting drama. Very rude of it.

The miss came where control mattered more than volume

FC Porto against Manchester City was the clearest wrong turn. Fipster leaned towards a home win and even had a 2-1 shape in mind, but Manchester City won 2-0 and did so through Erling Braut Haaland twice. The model over-weighted the home side’s chance of keeping things tight and under-read how quickly City could turn one good opening into two goals. Diogo Costa still posted an eye-catching eight saves, which tells you Porto were busy enough defensively; it just also tells you they spent much of the evening chasing shadows they never quite caught.

The round’s decisive names were mostly off Fipster’s sheet

This was not a night for tidy scorer calls. Raphinha scored twice for Barcelona, while Lamine Yamal added one and supplied two assists; none of those names were among Fipster’s pre-match scorer picks for that fixture. Ferran Torres went further for Paris Saint Germain with a hat-trick, with Masour Dembélé adding two goals and two assists beside him, and again Fipster had looked elsewhere entirely. Erling Braut Haaland’s brace for Manchester City was another clean example of an unforeseen scorer deciding matters without much ceremony.

'Called' names did matter when they landed. Archibald Brown scored for Fenerbahçe after being flagged before kick-off, which at least gave one player call something to cling to in a 1-1 draw with AS Roma. But most of the round’s standout performances belonged to players outside the pre-match shortlist: Martin Ødegaard nicked Arsenal’s winner at Napoli, Danijel Šturm scored twice for Slavia Praha even in defeat, and Michael Olise finished Bayern München’s rout with two goals and an assist after coming through uncalled. The pattern is hard to miss: Fipster saw enough winners coming, but not always who would actually do the damage.

That leaves a simple brief for next time: keep trusting the broad direction when the probabilities are firm, but be wary when player-level data is thin and the goals can arrive from almost anywhere else. The round showed that Fipster can still find enough correct outcomes to stay useful; it also showed that Champions League nights have no interest in being polite about individual heroes. A model can live with that. It just has to keep room in its pockets for names it did not fancy.