A round where the model found the right shape, then kept missing the furniture

The round mostly confirmed the direction of travel, even if the scorelines were rude about it.

Fipster leaned heavily towards home wins across the card, and that part of the read was not absurd. Three calls landed, but only one of them really looked comfortable on paper before kick-off. The rest of the round had a habit of punishing narrow-score thinking. A couple of matches stayed close enough to flatter the model’s instinct for tight margins; others blew straight past it once an unexpected scorer arrived or a favourite simply failed to control the game. Braga against Gil Vicente never got played, so there was at least one match that refused to join the argument altogether.

The cleanest hit was Sporting’s win over Guimaraes.

This was the sort of call that makes sense after the fact because the pre-match shape held up: Sporting were still favoured, and the game finished as a narrow home win rather than drifting into something stranger. The model had framed it as a 2-1 type of contest, and that basic read survived even though Sporting needed three goals to get there. The important bit is that this was not a lucky guess at the result line; it was a correct call on the match staying competitive while still going Sporting’s way. Fotis Ioannidis also mattered here in a way Fipster did not explicitly call as a scorer, which is usually how these things go when a prediction is right for the wrong reasons and then right again for better ones.

The miss that stung most was Casa Pia against Benfica.

This was not just a wrong scoreline; it was a wrong story. Fipster expected a close home win and instead got a rout built around players it never put forward pre-match. That is where the model over-weighted balance and under-weighted how quickly Benfica could turn one good opening into several. Vangelis Pavlidis did most of the damage, and he did it without any pre-match scorer call from Fipster to lean on. Once that kind of performance starts rolling, there is not much left for tidy scoreline logic to do except watch from a distance and take notes it should have taken earlier.

The round’s decisive names were mostly the ones Fipster failed to flag.

That is probably the clearest lesson from this set of fixtures. Vangelis Pavlidis stood out as the headline performer, with his hat-trick doing exactly what hat-tricks tend to do: making every cautious forecast look faintly embarrassed. Patricio Pérez gave FC Porto an early edge in their away win, while Fotis Ioannidis shaped Sporting’s victory with both goals and assists in a match that still ended up close enough to keep everyone honest. José Lima Cardoso and Leonardo Rodrigues dos Santos also decided Nacional’s win without any help from pre-match scorer calls, which is becoming something of a theme here. Even when Fipster got the result right — as with Arouca or Nacional — it often missed who would actually write the script.

There is still something usable in this round, which is more than can be said for some scorelines.

Three correct calls in nine is not pretty, but it is not random either. The model found enough structure to keep pointing at home sides in several places and to land on two tight wins correctly. What it did not do well was separate ordinary pressure from actual finishing power. That matters more than ever in rounds like this, where one striker gets hot, one underdog finds its moment late on, and suddenly all those neat 1-0 ideas are sitting in traffic. Next round needs less faith in symmetry and more respect for who is likely to do the scoring rather than merely occupy the correct half of the pitch.