Every week a solver picks a fifteen-player squad. It goes up here before the deadline, with the hash of the snapshot it used, alongside the squad I actually entered. Both are scored after the week is played.
Version v1.0-phase0: no bonus-points term, 81% of selectable players matched to prior history, and 18 of the 100 club-gameweeks in the horizon priced from a published difficulty rating rather than from expected goals. The method page has the rest.
| Week | Deadline | Model | Entered | Gap | Scored |
|---|---|---|---|---|---|
| GW1 · 2026/27 | 2026‑08‑21 17:30 UTC | 186.87 | 178.24 | -8.63 | not yet played |
| GW2 · 2026/27 | Published here before its deadline. | ||||
Model and Entered are the model’s own objective — a decay-weighted five-gameweek total for the best legal eleven each week, plus captain and bench shares. Not a points forecast for the gameweek. How it is computed.