How role rankings work (GK, CB, AM, WINGER, ST)

Most football sites sort everyone into four buckets: attacker, midfielder, defender, goalkeeper. That hides half the argument. A defensive midfielder and a number ten share the word “midfielder” but not the job. TackleZone role rankings split the Top 5 pool into eight roles: GK, CB, FB, CDM, CM, AM, WINGER, and ST.

Assignments start with match lineups. We read formation grids and coarse positions to estimate where a player actually played across the season — not where a marketing card says they belong. When lineup evidence is strong (think a winger who stays wide all year), that role wins even if raw stats look more like an attacking midfielder.

When lineups are split — a deep-lying playmaker tagged as both AM and CDM — we apply plurality rules for regulars with large samples so split seasons do not drop elite creators off the board entirely. Stats-only predictions fill gaps when lineup data is thin.

Each role gets its own TZ Score: a 0–100 composite normalised within that role across the Top 5 leagues. Centre-backs lean on duels, interceptions, and aerial work; wingers blend per-90 output with season goal and assist totals; attacking mids weight key passes and creativity. The score is relative to peers in the same role, not a universal “best player” number.

Eligibility still matters. Completed seasons use a substantial minutes floor (around 900) and a minimum rating so one hot month does not crown a board. Early in a new campaign we lower the minutes bar progressively — the same logic as our position rankings.

Use role boards when the debate is positional: who is the best pure striker, who is carrying a title race from defensive midfield, which full-back combines defending with chance creation. Pair a role board with Compare or a player profile when you need the full stat line behind a single name.

Role assignments and scores update as new lineups sync. If a player is missing, they may be below the usability threshold or outside the Top 5 sample — not because the model “dislikes” them.