Generated 2026-10-16. Seasons 2025-26 and 2026-27 (to date). For each club and match, the formation is predicted from data before kick-off and compared with the back line the manager actually started: four defenders or not, counted by FPL position (GK/DEF/MID/FWD). Only matches where all eleven starters have a known FPL position count (FPL positions exist for current players only): 324 team-matches.
| Rule | Back line right | XI overlap with the actual XI |
|---|---|---|
| Observed shape (current): the (defenders, forwards) count the manager started most often in the last 6 matches picks the formation family; the best XI is chosen within it, FPL defenders only fill back-line slots, and positions a player started at in 10%+ of his games since 2023-24 (transfermarkt-datasets lineups) count as natural | 75.3% | 74.2% |
| Strongest fit (previous): the formation that fits the strongest XI best | 61.4% | 73.5% |
| Always a back four (base rate) | 61.1% | — |
The previous rule was no better than always guessing a back four: it chose a back three whenever a squad had three good Transfermarkt centre-backs, even when one of them plays full-back (Liverpool 2026-27: Araújo). The window of 6 matches was set before this test and not tuned.
Limits: FPL lists some wing-backs as MID, so only "back four or not" is scored. Who plays which slot cannot be scored (FPL does not record it); the "defenders stay in the back line" rule changes slots, not who starts, which is why the XI overlap barely moves. Since the played-positions change a centre-back who also starts at right-back (Ronald Araújo: 14% of his starts since 2023-24) fits right-back fully; this season's positions arrive through API-Football's daily lineups.