FPL GW5 Review: 69% of Transfers Changed Nothing

Gameweek 5 is checked, and the most useful number in it has nothing to do with the armband.

Of the moves The Algorithm recommended to members' real squads, 69% finished the week level with what the manager was already going to do. 21% came out ahead. 10% came out behind. Across everyone, the whole exercise was worth 1.64 points per manager.

That is the transfer market in one line. Most weeks, the move you make is noise. The cost of making it is not.

The transfer that mattered, and the one that cost

Same squad, same week, hits included, points of the player in minus the player out.

The best of the round was Tarkowski in for Virgil, a six-point swing on 14 against 8, recommended in 6% of plans. Barry in for João Pedro was the same swing on 6 against 0, in 4% of plans.

The one that cost: Dewsbury-Hall in for Ndiaye, 3 against 5, recommended in 5% of plans. A move made for later fixtures can lose a week and still be right, but it lost the week.

Notice the shares. The moves that actually swung something were each recommended in single figures of plans, because they depended on what the manager already owned. The modal outcome, by a distance, was level.

Which is why 47% of plans rolled the transfer and 1% took a hit. When the median recommended move is worth about a point and a half, a four-point hit is not a close call.

The receipts

The Algorithm's XI, frozen before Friday's 18:30 lock and belonging to nobody, made 64 against the game's average of 48. It projected 65.3.

The shape was 3-4-3: Raya, Tarkowski, Gabriel, Murillo, Gibbs-White, B.Fernandes, Semenyo, Dewsbury-Hall, Haaland, Wissa, Isak, with Haaland in the armband.

Four rounds, ahead of the game's average in all four. Two of them landed inside four points of their own projection, and both of the misses went upwards, by 11.5 in GW4 and by 42.3 in GW2.

The armband is where the week went wrong. Haaland was the recommendation in 82% of plans and returned 6. Semenyo was the recommendation in 7% of plans and returned 17. The least-backed of the three named captains was the one that paid, and no amount of a 16-point margin over the average makes that a good call.

89% of members agreed with the captain. Among the minority who did not, the model's pick came out ahead in 71% of those cases and the manager's in 20%. Both of those are true about the same gameweek.

Playing the whole recommendation rather than their own week beat the manager in 37% of cases and lost to him in 14%, worth 2.31 points per manager.

Among our members, 52% beat the game's average of 48, and the best of them sits inside the top 5,000. They are engaged managers who chose to link a team, so read that as a description of who they are rather than proof of anything.

What was planned before the deadline

Tarkowski was the most-recommended buy of the week, in 21% of plans. Ndiaye was the most-recommended sale, in 10%.

Triple Captain appeared in 38% of plans. Haaland returned 6. That is the week, on the record.

18 days, then a morning deadline

Gameweek 6 locks on Saturday 10 October at 11:00. A morning deadline means your team is due before the football rather than after it, so the last press conferences are the ones that count.

It also means nothing you read about a knock this week survives contact with the next three weeks. You have one free transfer and eighteen days to decide what it is for. Spend the break on fixtures.

The sale worth making: Groß

Groß (£5.8m), Brighton

Fixtures: at Sunderland, then Crystal Palace at home on a 44.9% clean sheet, then Liverpool away and Man City away.

He is the leading scorer in the game on 47 points, form of 9.4, every one of Brighton's 450 minutes, 2.63 xGI. He is also absent from our top 40 for Gameweek 6 to Gameweek 13.

Those two facts do not contradict each other. Points scored are not points coming, and Brighton's run turns hard: as it stands they read 11.0% at Liverpool in GW8 and 9.1% at Man City in GW9.

Bottom line: this is the free window to move the best-performing midfielder in the game before his fixtures collect the bill, and at 27.2% owned you will not be alone in doing it.

The buy worth making: Saka

Saka (£9.5m), Arsenal

Fixtures: Leeds at home on a 52.6% clean sheet and 2.18 expected goals for, the best pair of numbers in Gameweek 6.

32 points from 416 minutes with 4.20 xGI, the highest attacking output at Arsenal. He reads 5.7 for Gameweek 6 and 41.1 across the eight, third in the game behind Haaland and B.Fernandes.

Minutes are not the question here. Fixtures are: 4.0 at Liverpool in GW9 and 4.2 at Man City in GW12 are the two holes, so he is a hold-through-two-bad-weeks buy rather than a punt on the restart.

Bottom line: third on our eight-gameweek board and owned by 13.5%, which is the part I cannot explain.

Holds, and where the armband goes

Haaland stays. £15.6m, 73.6% owned, 4.42 xG in five rounds, 47.5 across the eight, the best total in the game.

The armband week is not Gameweek 6 though. He reads 4.86 away at Liverpool, his second-lowest of the eight, against 7.0 at home to Ipswich in Gameweek 7, which is the highest single number anywhere on our board.

B.Fernandes reads 6.2 in Gameweek 6, with Man Utd at home to Spurs on a 32.1% clean sheet and 1.91 expected goals for. At £11.9m and 38.8% owned, that is the armband I would use in the first week back.

Tarkowski still reads well after the move that worked: £6.1m, 15.7% owned, 43 points, every minute, 32.1 across the eight with 5.1 in Gameweek 6 at Hull, where Everton read 35.1%. The flaw is 0.22 xGI, which is clean-sheet money with no attacking floor, and GW8 at Arsenal drops him to 3.1.

The cheap-versus-expensive Arsenal defence question has not moved since the GW5 preview: Gabriel reads 40.6 across the eight at £8.0m and 22.8% owned, Calafiori 31.2 at £5.8m and 50.6% owned.

The board, Gameweek 6 to Gameweek 13

Projections as of 21 September 2026. The full table of every player's expected points for the next eight gameweeks is free and updated daily.

Two availability notes from the official field, both to recheck before 10 October. João Pedro (£7.7m, 66% owned) carries a knee injury at a 75% chance of playing, and his Gameweek 6 figure already discounts a 29.8% chance of no minutes. Palmer (£9.7m, 26.3% owned) carries a muscular injury at 75%, and his doubt is one week wide: 24.2% for Gameweek 6, then 5.2% from Gameweek 7.

FAQ

Should I use my free transfer now or roll it?

Roll it unless a flag forces you. The median recommended move last week was worth about a point and a half, and you have eighteen days of team news before Gameweek 6 that you have not seen yet.

Should I sell Haaland over the international break?

No. He is the best eight-gameweek total in the game on 47.5 and he has played every minute. Move the armband for one week, not the player.

Who is the Gameweek 6 captain?

On today's numbers B.Fernandes at 6.2, then Saka at 5.7, then Haaland at 4.86. That is close enough that the 9 October team news decides it. If you would rather go against the crowd's armband, start with how to pick a differential that pays.

The Algorithm plans the next eight gameweeks for your own squad, and FPL Pro Coach at £10 a month is the only tier with the 14-day free trial: start one here.