2025/26 is a reconstructive replay (cutoff = first kickoff − 90 minutes, not a live two-hour freeze). 2026/27 pages are an illustrative sample of the live format. What that means

FPL Replay

FPL 2025/26 reconstructive replay

A full 38-gameweek reconstructive replay from the FPL lab. Three approaches run on rebuilt pre-deadline inputs, then scored on official points. This is not a live two-hour freeze.

Optimiser
2010 pts
Template
1990 pts
Same-state evidence
2010 pts
Chips played
None

What this season shows

Same-state evidence copied the optimiser in all 38 gameweek folders, so the evidence column equals 2,010. An exploratory post-season fork that kept a different squad from Gameweek 12 finished 2075 hybrid net (+65 from GW12–38). Direct same-state evidence contributed only +16 points. The evidence fork is an exploratory historical path recovered after outcomes were known. It is not a preregistered live freeze and is not a headline skill claim.

No chip was used. In Gameweek 34 both the optimiser (28 net) and the template (33 net) took an 8-point hit in a blank. Read the case.

Download season JSON · Season review

Gameweeks

38 weeks
GWCaptainPlanProjectedOptTemplateHash
GW1Palmer3-5-2 · Hold5656sha256:5bdbb4bf…39ab
GW2M.Salah3-5-2 · Hold60.05959sha256:b4893944…713a
GW3M.Salah3-5-2 · Hold63.45959sha256:dc98975a…2e2e
GW4M.Salah3-5-2 · 2 tr63.25762sha256:ad5bce4c…bf75
GW5M.Salah3-5-2 · Hold64.84649sha256:9e9c4872…e483
GW6M.Salah3-5-2 · Hold66.53135sha256:4e3bbb78…8931
GW7M.Salah3-5-2 · 1 tr65.04032sha256:3113a56c…e37f
GW8M.Salah3-5-2 · 2 tr66.75236sha256:37a129bb…81d7
GW9M.Salah3-5-2 · Hold67.85354sha256:83312c51…9617
GW10M.Salah3-5-2 · 2 tr67.86555sha256:ca001de9…298c
GW11M.Salah3-5-2 · Hold67.73544sha256:a1519482…0aeb
GW12M.Salah3-5-2 · Hold73.32939sha256:3328be88…4e75
GW13M.Salah3-5-2 · 1 tr67.23632sha256:49b5ea37…c4dc
GW14Haaland3-5-2 · 2 tr68.56345sha256:dbb77519…09bc
GW15Haaland3-5-2 · 1 tr68.15578sha256:446ba2ce…ed74
GW16Haaland3-5-2 · 1 tr69.25956sha256:6c274351…9bb4
GW17Haaland3-5-2 · 1 tr70.68844sha256:76667596…e8be
GW18Haaland3-5-2 · 1 tr71.43364sha256:aa407d93…7da4
GW19Haaland3-5-2 · 1 tr69.63644sha256:48651a0c…8706
GW20Haaland3-5-2 · 1 tr66.26543sha256:6b3e3c4a…07b9
GW21Haaland3-5-2 · 1 tr69.55538sha256:6b8ac10a…a8b7
GW22Haaland3-5-2 · 2 tr67.54771sha256:3a0da477…d9bb
GW23Haaland3-5-2 · Hold71.24167sha256:25fb0409…ed98
GW24Haaland4-4-2 · 1 tr67.46061sha256:dae38723…630c
GW25Gabriel3-5-2 · 1 tr67.05866sha256:5bb74b6f…bbc5
GW26Gabriel4-4-2 · 1 tr91.66847sha256:1a0a103d…a930
GW27Haaland4-4-2 · Hold72.74339sha256:124d9dd3…d4f5
GW28Haaland3-5-2 · 1 tr69.96735sha256:a9c0f6d6…1aac
GW29Semenyo4-4-2 · 1 tr67.55398sha256:af439a8b…1aa1
GW30Haaland4-4-2 · 1 tr68.04547sha256:54349e56…2ad5
GW31B.Fernandes3-5-2 · 3 tr53.66368sha256:d5b0ba9e…34b3
GW32B.Fernandes3-5-2 · 1 tr64.45337sha256:a63af706…d50b
GW33Haaland4-4-2 · 3 tr100.47961sha256:6e0ea922…3a87
GW34B.Fernandes3-4-3 · −8 hit · 3 tr35.62833sha256:b457f602…0659
GW35B.Fernandes3-5-2 · Hold60.44440sha256:ffc043e7…03c9
GW36Haaland3-5-2 · 2 tr87.38957sha256:a19ad730…e7a5
GW37Gabriel3-5-2 · Hold58.57063sha256:d630606c…969c
GW38Haaland4-4-2 · 2 tr59.83076sha256:36cfad56…72e8

Did the evidence approach beat the optimiser in 2025/26?

Not on the published per-week plans. Same-state evidence matched the optimiser every week. The +65 figure is an exploratory forked squad path from Gameweek 12, recovered after outcomes were known.

Were any chips played?

No. See the chip hubs and the Gameweek 34 hit case for what that gap means.