Chelsea

The authoritative final table (full season), centred on Chelsea.

Net xG (created minus conceded) in each 15-minute window, summed across the whole season's matches — above the line is time they dominate for chances, below is time it hurts them. Real on-clock shot timing.

Expected goals piling up over 8 matches: Chelsea against what their opponents created. A line pulling away upwards is a side creating more than it gives away; the gap at the end is the season’s balance (-1.2 xG). Hover a point for that match.

Each gene is a 0–100 score — the team's league percentile across the metrics that define that phase of play. It distils the whole report into an identity fingerprint: the taller the gene, the more that trait defines them.

Fifteen style metrics across five domains — Possess, Disrupt, Finish, Press, Defend — each a percentile against the rest of the league. Metrics marked ~proxy use the closest available event-data signal where the standard definition needs data this feed doesn't carry (explicit pressure events, offside/keeper-sweep incidents).

Field tilt = share of final-third touches between the two sides across Chelsea's matches, the standard proxy for territorial dominance. Plotted against points per game for every side in the league — the real question isn't just who controls territory, it's whether that control actually converts into results. Dashed lines mark the league average on each axis; Chelsea highlighted. Hover any dot for the team.

The share of Chelsea's time on the ball spent in each phase of play, in the order the phases happen. A phase is where the ball is on the pitch (own third, middle third, final third); a transition or a set piece skips the build-up and is counted on its own.

Every pass linked to its likely receiver (the next same-team touch), aggregated into player-pair combinations at each player's average position. Line width = how often the pair combines; colour = threat generated per pass (blue low → red high). Hover a line for the pair, or click a player to isolate their connections.

Every touch across 80 matches. Brighter means more touches.

Passes on one map, carries (the ball travelling with a player) on the other. Each is binned into the 15 pitch zones (three thirds × five vertical lanes) and shaded by its share of that kind of action; the arrows are the real actions.

Which channel they attacked through this season: the share of every action made in the opponent’s half, down the left, through the middle and down the right. Chelsea leaned on their right (37%). The bars under the pitch put the league’s share beside theirs.

Every completed pass or carry that took the ball from outside the final third into it, by the channel it arrived in. Chelsea made 300 this season (37.5 a match); 34% came into the right wing. Longer, bolder arrows mean more entries. Entries by carrying the ball in count as well as completed passes; dribbles are not counted.

Build-up passes grouped into 5 real categories — not the old zone grid, not raw individual lines: CB (passer is a centre-back), Wide→FB, Central, Long Wide, Long Central. Each gets its own panel below, so nothing competes for attention. Arrow width = volume, labelled with the exact count; hover for the top passer.

Every pass of Chelsea’s last 8 matches as an arrow, theirs first and the opponents’ with other team, each side attacking left to right (solid = completed, dashed grey = failed). Eight families per team — click any map to open it wide, then filter by channel or third, from one to another (or click a cell of the flow grids), by outcome, half, minute and passer. Channels: right wing, right half-space, centre, left half-space, left wing.

Not just how many progressive passes, but where they start, who plays them, and whether they actually penetrate. 321 progressive passes across 8 matches (40.1/game).

Which lanes actually create the threat

Share of passes played from each lane against its share of the threat (xT) they add. Efficient = more threat than volume, sterile = lots of the ball, little damage.

Progression passes grouped into 5 real categories: CB, Wide→FB, Central, Long Wide, Long Central. Each gets its own panel below, so nothing competes for attention. Arrow width = volume, labelled with the exact count; hover for the top passer.

A 90% pass completion can hide a sideways habit. Grading every pass by the threat it adds separates value creation from safe circulation.

193 progressive carries this season across the whole squad (24% of 794 tracked carries total), grouped into up to 4 spatial patterns by where each carry starts and ends together — not just its origin zone. Median 15m, longest 44m. Most active carrier: Palmer (31). Filter to one player below.

Every shot sorted twice. First by the kind of attack: an attacking transition (the shot came within 15 seconds of winning the ball), a settled attack (open play after the team had the ball), or a set piece. Then by how it was set up: a cross, a cut-back, a long ball, a pass into the box, or no pass at all (the player created it himself). So a cut-back from an attacking transition is counted apart from a cut-back in a settled attack.

Every attempt to create — and what came of it

Every pass into the box, every cut-back and every pass before a shot. Red led to a shot, gold to a goal; grey ones were tried but no shot came: solid found a team-mate, dashed lost the ball. Filter by outcome, type or player; tap an arrow for the detail.

Key passes (the pass immediately before a shot) grouped into 5 real categories: CB, Wide→FB, Central, Long Wide, Long Central. Each gets its own panel below, so nothing competes for attention. Arrow width = volume, labelled with the exact count; hover for the top creator.

Danger Funnel — which service carries threat

Every key pass pooled by where it was played FROM, then drawn with its arrow width set by total xT rather than by count. A zone that produces a lot of cheap balls shrinks; one that produces a few decisive ones dominates. Where this disagrees with the volume map, the volume map is the misleading one.

110 crosses: 44 found a team-mate (40%), 6 led to a shot (5%). 36 cut-backs: 18 found a team-mate (50%), 11 led to a shot (31%). A cross starts wide and is aimed into the box; a cut-back starts near the byline and is squared back. Grey ones were tried but no shot came: solid found a team-mate, dashed lost the ball.

110 crosses this season (the same crosses as the map above), split into fifteen territories — five channels (both wings, both half-spaces and the centre) by three depths from the halfway line to the byline. 26% come from one territory, the advanced right wing; 6 come from the half-spaces and 0 from the centre.

107 shots for 9.4 xG (1.18/game, #12 in the league). Click an area of the box to see who shoots from there; filter by outcome, player or match.

The move behind each of Chelsea’s 17 goals this season, from the moment they got the ball to the finish. Pick a goal from the list, then step through it or press Play: numbered circles are passes and touches, dotted lines are carries (the ball travelling with a player), dashed grey is a pass that was lost, the star is the shot.

Every goal Chelsea scored across the loaded matches, broken down by how it was created and finished.

Every turnover has a location and a cause. This is a rate, not a raw count — the share of pass attempts from each zone that end in a turnover, so a zone they simply play through more often doesn't automatically look riskier. Grey cells are real zones with too few attempts to trust a rate.

A composite Pressing Index built from four event-derived dimensions, each ranked against the league. Not one metric — intensity, height, effectiveness and opponent disruption together. The pitch shows every defensive action (the only five events this data records); the white line is the average height that feeds the score. Toggle by action type below, hover any dot for the player.

Every pressing metric, ranked in the league

Counted over the 227 opposition possessions that Chelsea ended with a tackle, interception, recovery or challenge. Fewer passes and seconds means the ball is won quickly (a press); more means the opponent is allowed to play before Chelsea step in. Chelsea win it after 5.5 opponent passes in the first half-hour and after 4.8 after the hour mark (sooner).

Reconstructed from the full timestamped event stream: real possession spells (not zone-clustered), and shots that follow a regain within a fixed window. This phase of play needs ordered, per-event timestamps — which every match here carries.

The report flipped: where do opponents create their threat against Chelsea? 353 opponent progressive passes and 101 shots faced (10.6 xG against) across the season.

Opponent threat by lane (attacking into Chelsea)

Every goal Chelsea conceded across the loaded matches — the opponent's shots in Chelsea's own matches — plus every individual opponent key pass faced, each drawn as its own real arrow (not zone-clustered) with the % share of each zone behind it.

How much of the goal threat is manufactured from dead balls — a route to goal that bypasses their open-play mechanism entirely.

Team archetype: Direct / counter high-press. Their closest stylistic matches in the league (cosine similarity on style metrics) — sides that play a similar way.

Every match split into three tiers by how hard that specific opponent pressed (opponent PPDA — lower means a more aggressive press), matching Chelsea's own passing behaviour to the result.

Results

Behaviour — how Chelsea actually play against each

For each of ten match numbers, the one match this season that departed furthest from Chelsea’s own average, the eight biggest swings first. Bar right of centre means more than usual, left means less. It shows how much the way they play moves from game to game, and which opponent moved it.

Every pass, carry, take-on, shot and ball win is worth the change it made to the chance of scoring minus the chance of conceding in the next 10 seconds. This is Chelsea’s last 8 matches added up: where the value came from, who produced it, and the six biggest single moves. OBV only measures what a player does with the ball: it shows who added the most on-ball value, not who was the most important player (defending off the ball, positioning and pressing are not in it).

Who delivers Chelsea’s dead balls and who finishes them this season. 45 corners taken, most by Jordan Henderson (15); 21 shots from corners and free kicks, 2 goals, led by Roméo Lavia (1 goals, 0.77 xG). “Into the box” counts corners that landed inside the penalty area; free-kick takers are the players who shot from one, because this feed does not mark who delivers the others.

Ranked by Vision AI Score — the same tactical behavioural evaluation each player's own report page shows (min 216 minutes). Coloured chips are that player's strongest capability blocks.

Chance creation leaders — key passes /90 and box entries.

Crosses and cutbacks — wingers and full-backs only (by real average position, not the raw D/M position code), ranked by combined volume.

Who does the defensive work — tackles, interceptions, recoveries.

Every player with 100+ minutes (20 players).

The nine most-used players (450+ minutes), each with the role their event fingerprint points to — not just their listed position.

Each recruitment target is derived directly from the team's weakest areas vs the league — the gap, why it shows in the data, and the profile that closes it.