Tools · Studies 04–05
Run the models yourself.
Two of the five studies ended in something you can use: a way to find players who play alike, and a way to estimate what survives a move. They answer different questions — and only one of them is a forecast.
Study 04 · Discovery
Player similarity
Who plays like him?
- What it does
- Finds the players in any top-five league whose context-adjusted profile — chance creation, involvement, shooting volume and shot profile — looks most like the player you pick, compared only within the same role.
- What it can't do
- Predict how a player will do after a move. Study 5 tested exactly that: similarity added nothing to the forecast. It also can't see defending, pace or anything outside attacking output.
- Data
- Profiles as of the 2025/26 season · every result precomputed with the Study 4 engine
Transfer calculator
Will his game travel?
- What it does
- Predicts a forward or midfielder's first-season xG + xA per 90 at a chosen club in another league, with an 80% range and the chance he keeps at least 75% of his output — and shows which inputs drive the number.
- What it can't do
- Price, fit, injuries or defensive work. Individual forecasts are uncertain (a typical 80% range is ±0.2), and low-output central midfielders moving into the Premier League are under-predicted.
- Data
- Model fitted on 428 real summer moves · tested once on 142 unseen moves
Case study
What would the model have told Manchester United in June 2024?
Standing at 1 June 2024 with only what was known then, every striker and central midfielder in the other four leagues was ranked. Zirkzee: predicted 0.46, actual 0.47 — the right number, but 48th of 106 strikers.