Nine games down, one to come on Monday night, and the opening weekend has already made a decent number of people look silly. This column included. Here is what actually happened, and which bits of it are likely to still matter in May.
Seven of the nine completed fixtures went to the home side. One away win, one draw. Whatever pre-season told us about squad depth and new signings, the first thing this weekend said is that playing at home in front of a full ground in August is worth a lot.
Our season preview gave Hull a 92.5% relegation probability, the bleakest number the model produced for any club in any league. They beat Manchester United 2-0, with Semi Ajayi and Nobel Mendy scoring, having gone into the game as 12.7% outsiders with six players unavailable.
That is the single biggest gap between projection and result anywhere on the weekend, and it is a useful reminder that a 66% favourite loses a third of the time.
Only three of the nine games saw both sides find the net. Six featured a clean sheet, and the round produced 25 goals at 2.78 a game, which is a perfectly normal total arriving in a very abnormal shape: a handful of comfortable home wins rather than end-to-end draws.
Anyone who spent the weekend on both-teams-to-score markets, and there were plenty of us, got a lesson in the difference between goals and shared goals.
Tottenham lost 3-0 at Brentford in Roberto De Zerbi's first league game in charge, with Keane Lewis-Potter, Vitaly Janelt and Michael Kayode scoring. The model leaned Brentford at 41.3% against 30.2% and the game was over long before the hour.
New system, unfamiliar shape, three attackers unavailable. It was always the hardest possible opening assignment.
Brighton were missing Baleba, Mitoma and Ferguson and still beat Aston Villa 4-0, Jack Hinshelwood scoring twice. Villa had João Gomes sent off and had arrived with the best closing form of any away side in the division.
The model backed Brighton's underlying numbers over Villa's form line. Occasionally the boring answer is the right one.
Hull beat United, Ipswich beat Sunderland 2-1, and only Coventry came away empty-handed after losing 3-0 at Arsenal. Two wins from three for the sides every model in the country expects to go straight back down.
City beat Bournemouth 2-1 through Marc Guéhi and Josko Gvardiol. Erling Haaland, the most expensive player in FPL history and the model's top-projected footballer this season, did not score.
The over 2.5 goals call landed, which was the model's loudest number of the round. The route it took there was not the expected one.
Newcastle 2-2 Liverpool was the game of the weekend and the one bet the model was most confident about: both teams to score at 58.4%, the strongest such call of the round. Szoboszlai and Gakpo scored for Liverpool, Elanga and Joe Willock for Newcastle.
A rebuilt Newcastle side projected twelfth took a point off a projected top-five team. That will feel like a win on Tyneside.
Nottingham Forest lost 1-0 at home to Leeds, Anton Stach scoring the only goal. Forest were 44.8% favourites, Leeds 26.3%. It was the round's only away win.
Honesty section. The projections model called the winner correctly in six of the nine completed matches, missing on United at Hull and Forest at home, with the Newcastle draw splitting the difference.
Our own selections on the predictions page did worse: four winners from nine, dragged down by a run of both-teams-to-score bets in a round that produced six clean sheets. Those are settling into the running record on that page, where they will stay whether they flatter us or not.
Fulham against Chelsea on Monday night closes the round.