Alexander Isak and Tyrick Mitchell made a mess of the tidy scorelines

Two right calls, eight that got away

The broad story of the round was simple enough: Fipster liked a set of narrow home wins, and the league promptly spent most of the weekend ignoring the neat bits. Arsenal and Manchester City were the only sides to land exactly where the model wanted them, which at least stopped this from becoming a complete write-off. But the rest of the card was full of awkward little corrections — late equaliser here, goalless draw there, an away win where a home edge had been pencilled in with rather too much confidence.

The one call that held its nerve: Manchester City 1-0 Coventry

This was the cleanest hit because the model read it as a controlled home win rather than a shootout. That mattered. Manchester City did not need to chase the game or turn it into a scrap; Erling Braut Haaland settled it in the first half, and that was enough to keep the scoreline inside the lane Fipster had marked out. The useful part is not just that the home side won, but that they won in the manner expected: one goal, no fuss, no need for heroics from anyone else. A rare moment of administrative competence from football.

Where we overreached: Ipswich 0-2 Liverpool

This was the miss that stung most because it was not just wrong on outcome; it was wrong on tone. Fipster leaned hard into an Ipswich home win and treated it as if Liverpool would be there to be managed rather than to do the managing. Instead Alexander Isak scored twice inside nine minutes and shut down the whole idea before it had properly started. The model over-weighted home advantage and under-read how quickly Liverpool could turn the match into something Ipswich could not recover from. Once Alexander Isak had struck twice, there was no neat second-half correction available. Just a very brisk lesson.

The names who actually ran the weekend were mostly uninvited ones

The round’s loudest individual story belonged to Alexander Isak, whose brace was also one of the clearest examples of an unforeseen scorer doing real damage to a prediction. Tyrick Mitchell did something similar for Crystal Palace, scoring twice in a game Fipster had expected Fulham to control; that is exactly the sort of player-level blind spot that turns a tidy-looking forecast into an untidy result. Erling Braut Haaland was another useful reminder that some threats do not need much invitation. He was not called pre-match either, yet he still supplied Manchester City’s winner and gave one of the round’s two correct scorelines its shape.

There were other names worth noting too: Vitaly Janelt for Brentford, Bryan Mbeumo and Benjamin Šeško in Everton’s draw with Manchester United, and Ainsley Maitland-Niles arriving late for Everton as well. None of them were part of Fipster’s pre-match scorer calls because there were no such calls to lean on in those fixtures, which is its own problem when a round starts producing goals from places you have not bothered to price in. The lesson is plain enough. The model got its best work done when it trusted structure over noise; it got punished when it assumed familiar patterns would hold without asking who might actually decide matters.

Next up: less faith in tidy margins

That leaves Fipster with a useful sort of discomfort. The two ticks show the model can still find a narrow win when the match state stays obedient; the eight misses say it needs more respect for late swings, surprise scorers and games that drift away from their first reading. The next round will ask for less certainty about slim home edges and more humility about who actually has teeth in front of goal. Which is annoying, but cheaper than pretending otherwise.