FPL Gameweek 3 Review: We Got The Captain Right And Still Scored 33
Our captain scored 9. The ten players around him scored 22. The average FPL manager scored 51. Here is where it went wrong.
Two pieces a week: the side the model picks before the deadline, and an honest read of how last week's calls actually went. What we believed, what we measured, what we changed because of it — including the attempts that failed.
Our captain scored 9. The ten players around him scored 22. The average FPL manager scored 51. Here is where it went wrong.
Haaland is our highest projection of the season at 7.9, at home to the worst team in the league. Six of our top eight are Man City and you can only own three. Full squad, best value and differentials.
We projected 8.3 for our GW1 captain. He scored 2. This week the model has him higher — 9.4 — plus the full £100m squad, the best value and the GW1 scorecard.
We projected 8.3 points for our GW1 captain. He scored 2. The model's scorecard, the real differentials, and the GW2 moves the data actually supports.
We checked every 2025/26 appearance against its fixture difficulty rating. A player scores 4.93 in a rated-1 fixture and 2.47 in a rated-5. But the effect is far bigger for defenders than forwards.
Every Gameweek 1 fixture with its difficulty, the model's highest-projected player in each position, the captain call and the best value on the board. First of a weekly series.
We ran the same £100m squad optimisation at one, three, five and eight gameweeks. Between the one-week and three-week answers, seven of the fifteen players change. Only seven survive every horizon.
We took twelve FPL stats from the first half of 2025/26 and checked which ones predicted the second half. Points per 90 came seventh. Threat and xGI came first.
Both £100m squads our model would draft for Gameweek 1 — one built around Haaland, one that fades him. The gap is 6.2 points over eight weeks, and the real argument is not about points.
Live 2026/27 ownership across all 564 players. Haaland is in 75.1% of squads, Man City fills two slots in the average team, and 371 players sit under 1%.
Team name ideas built from this season's actual squad lists, how to set up a mini-league properly, and the rules worth knowing before the Gameweek 1 deadline.
The bonus points system was rebuilt to stop defenders double-dipping. We ran the new rules over last season's data to find who loses most. Senesi loses 63 BPS.
Tottenham's defensive overhaul repriced up, Man City's breakout stars costing more than the one who actually arrived, and Arsenal's biggest names cut by up to £1.6m — what it means for your draft budget.
Last season's points against this season's prices, for all 558 players. The best points-per-million in the game is a £4.5m defender — and the premium market has halved.
Everything that changed for Fantasy Premier League 2026/27 — the rules, the prices, the fixtures and the eight new managers — with the numbers behind each, from our own database.
Blank and double gameweeks decide when FPL chips pay, and they are knowable in advance. How the 2025/26 structure dictated Bench Boost GW33, Free Hit GW34 and Triple Captain GW36 — and how to read 2026/27's structure when it lands.
The 2025/26 Fantasy Premier League season reviewed from the database — final top scorers, the best points-per-million picks, the differentials that paid, and three lessons worth carrying into 2026/27.
A plain-English walkthrough of the FPL Analytics expected-points model — what goes in, what it predicts, how it scored against simple baselines across 31 gameweeks, and exactly where it still fails.