Player vs player
Head to Head
Put any two players on the same bench, across all ten leagues — and pin each side to its own season. A category radar, a metric-by-metric percentile duel, a computed similarity score, and each player's archetype. Compare two players, or the same player across two years: the page opens on Bruno Fernandes' 2024/25 against his 2025/26 — an 82 → 95 bounce.
The matchup
Each side keeps its own colour, and every chart on this page holds that convention. Same name on both sides, two different seasons, traces one player's arc.
Loading…
Season movers · 2024/25 → 2025/26
Biggest Statz Rating swings among outfielders with ≥ 1,500 min in both seasons. Click one to load it into slot A (at 2025/26).
Risers
Fallers
Category radar
Each spoke is a category score (0–100) — the mean of that category's sub-metric percentiles. The overlaid shapes show where each player's game is weighted; the table beneath carries the exact numbers, with the gap between two players or each one’s distance from the group average. Switch it to Bars for the mirrored head-to-head view. Flip to Pizza for the scouting terminal's percentile pizzas, side by side — each player on his own position's slice set, one hue per side, with the same football / technical label registers.
Where they live
Every qualifying player in the pool as a muted cloud, the compared sides pinned in their slot colours — pick any two axes to see where each player's game actually sits. Axis baselines differ by family: the duel, on-ball, deep-event and finishing axes are z-scored within position group and pool, while the style composites and the PCA axes are z-scores across all outfielders — the caption under the chart says which pair you are looking at. Duel/physical axes read the Sofascore / FotMob / Opta boards.
The matchup on the map
Two chosen axes
Metric by metric
Only metrics every selected player records are shown — mirrored back-to-back for two, an aligned table beyond that — with the per-metric leader marked. Percentiles are ranked within each player's position cohort (AM and W ranked separately); switch to raw per-90 for a single cross-position scale.
Shared metrics, mirrored
Bar length = percentile within the player's own cohort · 0–100
How alike are they?
A similarity score of 100 means two statistically interchangeable profiles; 0 means opposites. Where the site has already paired two players in its neighbour graph we show that exact number, otherwise we compute one live. With three or more sides this becomes the full pairwise matrix.
Similarity SCORE is cosine similarity between two players' z-scored per-90 feature vectors, clipped at 0 and scaled ×100 (features in meta.pca.features_outfield). When a pair is not already stored in the neighbour graph, this page computes an approximate alignment in-browser from the metrics the two share — directionally the same idea, but read it as a guide, not the canonical number. Start share is a proxy for "% minutes as starter" (minutes ÷ appearances ÷ 90); true starts are not in these sources.