Gaffer · Last Man Standing

Round 8 — one club, and you only get it once

DEADLINE Fri 23 Oct, 17:30 UTC 93h TO GO RULES DRAW IS OUT · 1 LIFE TEAM RATINGS 69 matches in UPDATED 19 Oct 2026, 20:01 UTC FULL FPL BOARD →

This round

The recommendation is not the club most likely to win on Saturday. It is the club whose use this week leaves the best route through the rest of the season, which is a different answer often enough to be the point. Spending the best team on a week the second-best team would also have survived is how most entries go out in October, so every candidate is priced by re-planning the whole season behind it: cost is the share of the run you give up by taking it now.

Chelseathe pick for GW8
79%chance it survives the round
3.3%chance of surviving all 8 planned rounds
0clubs already spent
This round, priced over the whole route
ClubFixtureWinDrawSurviveRouteCostField
Chelseav Spurs79%14%79%3.3%—57%
Man Cityat Aston Villa52%26%52%3.2%-0%3%
Arsenalv Everton62%26%62%3.2%-1%6%held for GW11
Crystal Palacev Newcastle47%23%47%2.9%-9%2%
Man Utdv Bournemouth44%26%44%2.6%-20%2%held for GW14
Brightonat Liverpool55%25%55%2.2%-33%4%held for GW12
The route · 3.3% chance of surviving all 8 rounds
ChelseaGW8 · v Spurs79%
BrentfordGW9 · v Nott'm Forest58%
FulhamGW10 · v Newcastle63%
ArsenalGW11 · at Newcastle70%
BrightonGW12 · v Newcastle90%
LiverpoolGW13 · v Sunderland53%
Man UtdGW14 · v Coventry City61%
Ipswich TownGW15 · v Newcastle54%
Clubs used · one life, a draw is out
nothing yet

How the numbers are built

Win, draw, survive. Every club carries an attack and a defence rating fitted to results, so a fixture's expected goals is attack × opponent defence × home advantage. A pair of goal rates is a distribution over scorelines, and a distribution over scorelines is a win probability. The four lowest scorelines carry the Dixon-Coles correction, because independent Poisson under-counts 0-0 and 1-1 — and in a format where a draw eliminates you, the draw is half the risk rather than a rounding error.

Route. The chance of surviving every round on the plan, not just this one. Eight rounds at 60–70% a time multiplies down to a couple of percent; that is the format, not a fault in the model, and the page prints it rather than hiding it. The number worth watching is the margin over taking the best available team every week, which is quoted on every run.

Field. The one column here that is modelled rather than measured. No pool publishes its picks, so it is a model of where a typical field lands — a softmax over win probability, flattened because everyone else has their own used list. Treat it as a shape, never as evidence.