Transfer what-ifs · runs-added, but for goals
The Window
Pick a club. Sell, sign, move roles, drag minutes — across 2,500 players in ten leagues — and get the window priced: this scenario is worth +X expected goals ≈ +Y points, decomposed change by change, with every assumption on the table and none of them hidden.
The club
Ten leagues, 224 clubs. Squads list every player who qualified for the season corpus — depth pieces under the minutes floor aren’t priced here, and the ledger below says what that leaves out.
An estimate, not a prediction — what this model claims, and what it refuses to
The squad
Each row is a lever. Minutes re-books playing time. Role switches to the player’s own rate in that position — only offered where he actually played it (3+ apps). Usage is your haircut for a smaller share of the ball: Cunha’s touches dropped 30% when he swapped Wolves for United — the model cannot know that in advance, so the dial is yours. 2-season blends this season with last, minute-weighted, where last season is on record.
He joins his position’s band below with the same levers. An import arrives with his rate scaled by the source and target league coefficients, and his set-piece half re-priced at this club’s dead-ball rate — both printed on his row.
The depth chart
Pick a shape and it asks for eleven shirts. Who holds each one, and who covers, is the whole control — minutes follow the pecking order rather than the other way round. Seeded from the season that actually happened, so switching it on moves nobody until you do.
The market
Where this squad is weakest by the site’s own published rule, and what the Recruiter’s candidates for those roles would actually add here — priced by the same model as everything else on this page, against your spend ceiling.
The verdict
A or B
Pin a scenario, go and try another, then read the difference. The pin keeps the numbers this model gave it at the time and the link that rebuilds it — so nothing here is a second estimate of the same thing.
The other end
Attack is only half a window. This is what your changes do to the goals the squad concedes — from a model fitted on 288 team-seasons, and honest about how much of defending it cannot see.
What it’s worth in the table
Re-run the season’s Monte Carlo with this squad’s attacking output scaled by your scenario — the same Dixon–Coles model behind The Justice Table, 10,000 seasons each way, in your browser.
What else moves
The window is priced in expected goals above — but it moves the rest of the squad’s volume too. As played → your scenario, across every covered player.