First things first: the Gameweek 3 deadline is tonight - Friday 4 September, 6:30pm UK. Liverpool's trip to Ipswich drags the whole gameweek forward, so the usual Saturday-morning tinker is a Friday-evening one. Here is what the Statz projections say to do with it - who comes in, whether the wildcard should come out, and which teams to build around for the next fortnight. All numbers are from the Statz FPL projection model as of Friday lunchtime; check team news before you commit, because the late press conferences can still move expected minutes.
Erling Haaland is the week, frankly. Coventry at home projects to 7.5 points - two clear of anyone else - built on 1.16 expected goals, the highest single-fixture xG the model has produced this season. If you do not own him, that is the transfer; if you do, he is the captain and the rest of this article is detail.
Alexander Isak (5.8 projected at Ipswich, 80 expected minutes) is the strongest route into the Liverpool attack, and the fixtures are with him: Ipswich away tonight, Fulham at home next week. Dominik Szoboszlai (5.1) is the midfield version of the same idea. Liverpool's results have been scratchy, but the projections price the underlying numbers, and the next two opponents are the kindest pair on their calendar.
In midfield, Antoine Semenyo (5.5 projected, 81 expected minutes) rides the same Coventry fixture as Haaland at a fraction of the price, and Bruno Fernandes (6.1 at Everton) is Manchester United's set-piece-and-everything hub as usual. At the back, Josko Gvardiol (5.0), Nathan Collins (4.9, Brentford home to Sunderland) and Virgil van Dijk (4.8, 90 expected minutes) lead the board.
Two warnings from the same data. Morgan Gibbs-White projects well (5.0) but is a listed doubt at the time of writing - do not buy a press-conference risk on a Friday deadline. And Rayan Cherki's 58 expected minutes is the model telling you City's rotation is real; the output per minute is elite, the minutes are not safe.
The honest framework: count the starters you would not pick again - flagged, benched, or bought in a GW1 panic. Four or more and the wildcard is working for you; fewer and it is working against you. Two gameweeks is not enough evidence to torch a plan you believed in three weeks ago.
There is also a genuine timing argument for waiting one more week. Chelsea's fixtures turn kind from Gameweek 4 (Hull at home projects at 81% - the most one-sided fixture on next week's card), Manchester City's do the opposite (the derby at Old Trafford), and Spurs' injury pile-up - no Vicario, no Romero, Kulusevski out, Maddison and Kudus doubts - will look very different in a week. A Gameweek 4 wildcard gets all of that information, a full week to use it, and none of tonight's 6:30 scramble. If your team is broken, break the glass now; if you are merely restless, next week is the better week to be brave.
The two-gameweek picture, from the match projections:
Full projections for every player are on the FPL projections page, the predicted lineups live on team news, and the Team Planner will do the arithmetic on your actual squad.