Goalkeeper distribution analysis: reading a keeper's feet

The centre-back drops into the corner of the penalty area with a hand up, the opposition striker curves his run to cut off the pass, and the goalkeeper has about a second and a half to decide whether to play into the space the striker has just left or clear it forty metres and start again. Serious goalkeeper distribution analysis is an attempt to grade that decision rather than the pass that follows it, and almost every public number attached to a keeper's feet grades the wrong one.

The short version is this. A goalkeeper's distribution should be judged on three things at once: what the situation allowed, what he chose, and what the team's possession was worth several passes later. A pass completion percentage answers none of those, and it is the figure that appears in nearly every player card and broadcast graphic. Getting from that figure to something useful means understanding what a keeper is actually being asked to do, and how the laws of the game arrived at asking him to do it.

What goalkeeper distribution analysis actually measures

Start with the raw material. Every time a goalkeeper releases the ball, an analyst can record a small set of facts: where the ball started, where it ended, how far it travelled and in which direction, whether it was struck with the hands or the feet, whether it came from a dead ball or from open play, how many opponents were within pressing range at the moment of release, and which team had the ball five and ten seconds later.

Those last two fields are the ones that separate a real assessment from a decorative one.

Pressure at release converts an outcome into a decision. A pass completed with nobody within fifteen metres tells you almost nothing about a goalkeeper. The same pass completed with a forward closing at speed and a second opponent cutting the return lane tells you a great deal. Without pressure context, distribution analysis is grading players on the difficulty of the situations their opponents chose to create.

Possession state after the delivery converts a pass into a consequence. Football does not reward completing a pass. It rewards keeping the ball in a position that leads somewhere, and a delivery that is technically completed to a full-back who is immediately dispossessed on the touchline has cost the team more than a long ball that was headed clear into midfield. Any framework that stops at the moment of first contact is measuring an event rather than an outcome.

• 1992Year handling a deliberate back pass became an offence
• 2019Year a goal kick became live as soon as it is kicked
• 8Seconds a keeper may control the ball from 2025/26
• 5Seconds the referee counts down visually

Provisions of the current Laws of the Game and the years they took effect. Structural facts, not performance figures.

The three law changes that turned a goalkeeper into a passer

Nothing about the modern goalkeeper's passing role is natural. Each part of it was created by a specific amendment, and each amendment was written to solve a problem that had nothing to do with build-up play.

The back-pass restriction, 1992. Before it, a defender under pressure could roll the ball to his keeper, who picked it up. That made the goalkeeper a permanent, untouchable escape hatch, and it made time-wasting trivially easy. The amendment removed the hands from any ball deliberately kicked to the keeper by a team-mate, and later extended the same restriction to a team-mate's throw-in. Overnight, every goalkeeper in the world needed a first touch. The clause left one route open that still shapes play today: a deliberate header, chest or knee back to the keeper can be handled, which is precisely why a defender in real trouble heads the ball to his own goalkeeper rather than passing it.

The goal kick amendment, 2019. Until then a goal kick was not in play until the ball left the penalty area, so a short pass to a centre-back standing inside the box was a retake. The amendment made the ball live as soon as it has been clearly kicked and has moved, with opponents required to stay outside the area until then. That single change created the modern short goal kick, because a defender can now receive inside his own six-yard box with the opposition legally excluded from the area at the moment of the pass. Every image of two centre-backs splitting to the corners of the penalty area for a goal kick is downstream of this clause.

The holding-time amendment, 2025/26. The old six-second limit on a keeper controlling the ball was almost never punished, because an indirect free kick inside the penalty area was a disproportionate sanction that took a long time to set up. The replacement gives the keeper eight seconds, has the referee raise a hand and visually count down the final five, and awards a corner kick if the time is exceeded. That is a lighter offence with a heavier and much faster consequence, and it removed the goalkeeper's ability to kill a game by holding the ball.

Read together, the three amendments describe a governing body steadily closing the routes by which a goalkeeper could stop the game, and accidentally inventing a new outfield position in the process.

Short, medium and long are three different jobs, not three distances

Coding distribution by length alone loses the point. The three bands correspond to three tactical intentions, and mixing them in one average produces a number that means nothing.

Short distribution is an invitation. Its purpose is not to advance the ball, it is to draw opponents forward so that space appears behind them. A goalkeeper playing five metres to a centre-back has gained no territory whatsoever and may have gained a great deal, because the opposing forward now has to decide whether to press. Judged on distance gained, this pass looks worthless. Judged on what it does to the shape in front of it, it is the whole plan. The mechanics of that plan are set out in the piece on building play from the goalkeeper outwards.

Medium distribution is the one nobody talks about and the one that separates keepers. It is the ball into a full-back who has pushed high, or the clipped pass into a midfielder who has dropped between the lines, played over the first line of the press rather than around it. It requires weight, disguise and the courage to release the ball later than is comfortable. It is also the band where the pressure context matters most, because the receiving player is usually marked.

Long distribution is a contest rather than a pass. Its success is decided by the aerial matchup at the far end and by whether the team has organised to win the second ball, neither of which is the goalkeeper's doing. The most useful long delivery is not the one that finds a head, it is the one that finds a head on the side of the pitch where the team has numbers.

47%29%24%

• Short, under 15 metres18deliveries
• Medium, 15 to 40 metres11deliveries
• Long, over 40 metres9deliveries

Invented composition on stated assumptions, chosen to show how the same total splits three ways. Not a measurement of any goalkeeper or competition. The point is that the three bands are separate jobs and should never be averaged together.

Show the numbers

Constructed example: one keeper's deliveries across a match

Item	Value
Short, under 15 metres	18deliveries
Medium, 15 to 40 metres	11deliveries
Long, over 40 metres	9deliveries

Why pass completion is the most misleading number on a keeper's card

A goalkeeper picks his own denominator, and that is the entire problem.

Suppose two keepers each make thirty-eight deliveries in a match. The first plays thirty-four of them short and sideways to a defender with no opponent within twenty metres, and completes thirty-three. The second plays fourteen short, sixteen into midfield through a press, and eight long, completing twenty-eight. On the published figure the first goalkeeper is the better passer by a distance. In football terms the second has broken the opposition press sixteen times and the first has broken it never.

Nothing about this is subtle, and yet the completion figure survives because it is easy to calculate and easy to display. The specific failure modes are worth naming.

It rewards passivity, because refusing difficult passes raises the average. It punishes system, because a keeper in a side that presses high and plays a compressed field faces a different distribution of difficulty from one in a side that defends deep. It conflates categories, because a long ball into an aerial duel has a natural success rate somewhere near a coin toss and a five-metre lay-off has a natural success rate near certainty, and averaging them produces a number governed by the mix rather than the skill. And it ignores what happened next entirely.

The corrective is not complicated. Split the deliveries by band, model the expected completion of each pass given its length, direction and the pressure on the passer, and report the difference between what the keeper completed and what a typical goalkeeper would have completed on the same set of attempts. That is the same logic that underpins comparing a shot against the average conversion of shots like it, applied to passing rather than shooting, and it removes the denominator problem at a stroke.

Launch rate, and the pressure that produces it

Launch rate is the share of deliveries above a distance threshold, commonly around forty metres. It is the fastest single description of how a team wants to start possession, and it is routinely misread as a competence measure.

A high launch rate can mean four completely different things. It can mean the coach wants it, because the side is built around a target forward and second-ball midfielders. It can mean the opposition forced it, because a well-organised press removed the short options. It can mean the goalkeeper is uncomfortable and is choosing the safe long ball when short options existed. Or it can mean the match state demanded it, because a team chasing a goal in the last ten minutes does not build patiently.

Distinguishing these requires the pressure field, and it is why the useful version of the metric is not launch rate but unforced launch rate: the share of long deliveries made when a short option was genuinely available and uncontested. That figure is much harder to compute and much more informative, and it is close to the only distribution number that survives contact with a coach.

The reverse error is equally common. A very low launch rate is treated as evidence of technical quality, when it can simply mean the opposition declined to press. A side that faces a deep defensive block every week will have a goalkeeper who never launches, because there is nobody within forty metres of him for the entire match. His feet have not been tested.

The goal kick is bait, and the trap is set before the ball is placed

The most instructive sequence in modern football starts with a goalkeeper putting a ball down on the six-yard line and looking bored.

1. 1The ball is placed and the shape spreadsThe two centre-backs move to the corners of the penalty area, the holding midfielder drops towards the D, and the full-backs push high and wide. Under the 2019 amendment the opposition must stay outside the area until the ball is kicked, so the receiving defender is momentarily unopposed.
2. 2The press commits or it does notThe opposing forwards decide whether to advance to the edge of the box. If they refuse, the defenders can carry the ball out unopposed and the attacking side has gained thirty metres for nothing. If they commit, they have vacated the space they were previously occupying.
3. 3The keeper plays the first short passThe pass goes to whichever centre-back the press has left later. This is not an attempt to advance play. It is an attempt to make one opponent run twenty metres, which pulls the whole pressing structure with him.
4. 4The press adjusts and the trigger appearsAs the forwards shift across, one passing lane opens behind their line. The midfielder who dropped is now free, or the far centre-back is, or the ball can go back to the goalkeeper with the press already committed and out of position.
5. 5The ball is returned to the goalkeeperThis is the move that looks like panic and is usually the plan. With the press committed high, the keeper receives with a clear view of the whole pitch and the opposition midfield line stretched away from its defence.
6. 6The line is broken, or the pitch isEither the goalkeeper plays through the gap into a midfielder facing forward, or he goes long into the space the press has vacated behind its own midfield. Both outcomes are wins, and which one is available was decided by the opposition four passes earlier.
7. 7The second ball decides the phaseIf the delivery was long, the team that organised for the knock-down keeps possession in the middle third. Sides that build this way rehearse the second-ball shape as carefully as the first pass, because that is where the sequence is actually won or lost.

The generic sequence, attacking and defending sides together. Individual teams vary the roles and the triggers; the underlying logic is common to all of them.

Two things follow from that sequence that are easy to miss on television. The goalkeeper's most valuable action in the whole passage may be a five-metre pass, because it was the pass that moved the press. And the risk everybody watches, the ball played short in his own six-yard box, is being taken deliberately in exchange for a positional advantage the camera angle does not show. Sides that press hardest against this are running the counter-argument, and the reasoning behind hunting the ball high is set out in the pressing philosophy built on immediate recovery.

Where the ball ends up matters more than whether it arrived

The single biggest upgrade to distribution analysis is to stop evaluating the pass and start evaluating the possession.

Define a window after the goalkeeper releases the ball, commonly around ten seconds or a fixed number of subsequent touches, and record what the team had at the end of it. Retained possession in the opposition half is the best outcome. Retained possession in your own half is neutral. Lost possession in the middle third is a modest cost. Lost possession inside your own defensive third is by some distance the worst thing that can happen, because it hands the opponent the ball with your defence facing the wrong way and your goalkeeper out of position.

Once outcomes are graded that way, several intuitions reverse. A long ball that is headed clear and recovered by the opposition in midfield stops looking like a failure, because the team is defensively organised and the ball is sixty metres from goal. A short pass completed to a full-back who is then trapped on the touchline starts looking like a failure, because the ball is lost in the one area where losing it is expensive.

Own penalty area6x

Own defensive third4x

Middle third1x

Opposition half0x

Invented weightings on stated assumptions, expressed relative to a loss in the middle third set at one. Not derived from any competition. The figure illustrates why zone of loss dominates completion rate in any serious distribution model.

Show the numbers

Constructed example: relative cost of losing the ball, by zone of loss

Item	Relative cost
Own penalty area	6x
Own defensive third	4x
Middle third	1x
Opposition half	0x

The asymmetry in that table is the reason coaches who demand short goal kicks are not being reckless. They are accepting a small number of high-cost errors in exchange for a large number of low-cost territorial gains, and provided the ratio holds, the trade is positive. It stops being positive the moment the players executing it are not good enough, which is why the same instruction produces triumph at one club and a catalogue of conceded goals at another.

Throws, rolls and the counter-attack that never gets coded

Hand distribution is systematically undervalued because it usually happens when the broadcast has cut to a replay.

A goalkeeper who has just caught a cross is holding the ball at the precise moment the opposition is at its most disorganised. Their full-backs are high, their midfielders have pushed up to support the cross, and there is a large area of unoccupied grass between their midfield and defence. A fast, accurate throw into that space, released within two or three seconds, is one of the most valuable actions available in the entire game, and it is recorded in most datasets as a completed pass of moderate length.

Three properties make hand distribution distinct. It is far more accurate than a kick over the same distance, because the release is controlled rather than struck. It can be released much faster, since there is no need to set the ball down or take a run-up. And it is almost always played to a team-mate's feet rather than into a contest, which means the possession is retained rather than gambled.

The eight-second limit interacts with this directly. A keeper who wants to slow the game down now has a hard ceiling on doing so, and the sanction is a corner rather than a whistle nobody hears. The amendment was written to stop time-wasting, and its side effect is to push goalkeepers towards releasing the ball while the counter-attack is still available, which is the outcome the analysts wanted anyway.

Open play distribution is a different skill from the dead ball

Any assessment that pools dead-ball and open-play deliveries into one figure is measuring two skills with one number.

At a goal kick, the goalkeeper has an unlimited amount of time, a stationary ball, a fixed and known geometry, and opponents legally excluded from the penalty area until he plays it. The only difficulty is the decision. Technically, it is the easiest pass in football.

In open play, he is receiving a back pass on the half-turn with a forward two metres away, often on his weaker foot, often with the ball arriving at an awkward height off a bobbling surface. The decision window is a fraction of a second and the technical demand is high. This is where goalkeepers are actually exposed, and it is where the difference between a keeper who is comfortable with the ball and one who has been drilled into looking comfortable becomes visible.

Any recruitment process that looks at goal-kick statistics and concludes a goalkeeper can play is going to sign somebody who cannot. The open-play sample is smaller, noisier and harder to collect, and it is the one that predicts.

Footedness, angles and the side the press attacks

Two structural facts about a goalkeeper's body shape his whole distribution profile, and both are exploitable.

The first is his weaker foot. A goalkeeper who is genuinely two-footed can receive the ball on either side and play out on either side, which means the press cannot cut off half the pitch by approaching from one angle. A goalkeeper who is not will consistently take his first touch back onto the stronger side, and an opposing forward who knows it presses from the weak side and effectively removes half the field. Watch which shoulder the pressing forward approaches over and you will usually learn what the opposition analyst found.

The second is the receiving angle. A goalkeeper standing on the goal line has almost no passing options, because every lane runs across his own six-yard box. A goalkeeper positioned twenty metres from goal has options on both sides and through the middle, and can also play the ball first time. Where a keeper takes up position when his team has the ball in the middle third is one of the clearest available signals about the role he has been given, and it is captured directly in positional feeds of the kind described in how tracking data is captured and joined.

How clubs actually test a goalkeeper's feet in recruitment

Recruitment departments know the public numbers are weak, so the internal process looks different.

The volume of ball-playing situations is checked first, because a keeper in a possession-heavy side may have faced very few genuine pressing situations across a whole season, and a large sample of easy passes is not evidence. Sample size, in this job, means the number of hard passes rather than the number of passes.

The situations are then filtered to those with a defined pressure level, and the video of each is watched rather than aggregated. What is being assessed is not whether the pass was completed but whether the correct option was chosen, whether the body shape before receiving allowed both options to stay open, and how late the keeper was willing to release the ball. Late release is the single most reliable indicator of confidence, because a nervous goalkeeper plays early and lets the defence reset.

Finally the outcome distribution is checked for its tail rather than its mean. A goalkeeper who never makes a catastrophic error and rarely breaks a press may be worth more or less than one who breaks presses regularly and gives away one goal a season, and which of the two is right depends entirely on the coach and the division. That judgement is not statistical, and clubs that pretend otherwise tend to sign the higher completion percentage and regret it.

Against a press and against a block are two different examinations

A goalkeeper's distribution profile changes completely depending on what the opposition does, and comparing two keepers without accounting for it compares two examinations rather than two candidates.

Against a committed high press, the question is escape. The opposition has put four or five players inside forty metres of the goal, which means it has left the same number of gaps further up the pitch. The goalkeeper's job is to survive the first ten seconds without conceding possession in a dangerous zone, and every one of his passes is a genuine test. The reward for succeeding is enormous, because a press that has been played through leaves the pressing team more exposed than any other defensive shape in football.

Against a deep block, the question is patience. Nobody is pressing, there is nothing to escape, and the goalkeeper has the ball at his feet for long stretches with all eleven opponents inside their own half. His passing is barely tested at all. What is tested instead is his willingness to step up the pitch and effectively become an extra outfield player, giving his side a spare man in the build-up that a compact block cannot mark without abandoning somebody in front of it.

Those two examinations reward different qualities. The first rewards composure and a clean first touch. The second rewards positioning and the nerve to stand fifty metres from your own goal, where a single interception ends in an empty net. A goalkeeper can be excellent at one and poor at the other, and a season's aggregate figures will average the two into a description of neither.

There is a third case that is often forgotten and is the most demanding of all: the mixed press, where opponents press some goal kicks and refuse others depending on the score, the phase of the game, and whether their forwards have anything left. Reading which one is coming is a recognition task performed in a second and a half, and it is what separates goalkeepers who look calm in build-up from goalkeepers who look calm in build-up until it matters.

The distribution premium, and why it changed what keepers cost

The commercial consequence of all this is the part that gets least attention and drives the most behaviour inside clubs.

For most of the sport's history, a goalkeeper was priced on one skill. He stopped shots or he did not, and everything else was a rounding error. The 1992 amendment created a second skill, the 2019 amendment made that second skill tactically central rather than merely useful, and the effect on the market was to split the position into two overlapping talent pools that do not fully coincide.

The awkwardness is that shot stopping and distribution are close to independent. Nothing about the ability to hold a low, hard shot to the near post predicts the ability to receive on the half-turn under a press. So a club shopping for a goalkeeper is not selecting on one axis with some noise, it is selecting on two axes, and the supply of players who are strongly above average on both is small. That scarcity is the entire source of the distribution premium.

It also explains a pattern that recurs across divisions. A side that plays a possession-heavy system will pay a lot for average shot stopping plus excellent feet, and will accept the goals that decision costs, because the alternative is a build-up structure that does not function. A side that defends deep will do the reverse, because a keeper who never has to pass under pressure gives up nothing by being ordinary with the ball. Neither club is wrong. They are buying different jobs with the same job title, and the transfer fees they pay are not comparable for the same reason the pass completion figures are not.

The uncomfortable part, and the reason the premium is contested inside analytics departments, is that nobody has established that distribution moves the scoreline as much as shot stopping does. The measurement problems in this piece are the reason why. A skill that cannot be cleanly measured cannot be cleanly priced, and a premium is being paid anyway.

What distribution models still cannot see

Three blind spots are worth stating plainly.

Instruction is invisible. A goalkeeper who goes long forty times because he was told to is indistinguishable in the data from one who went long forty times because he was frightened. Only the coach knows, and the coach is not going to say.

Deception is invisible. The most valuable skill in short distribution is convincing an opponent that a pass is going one way and playing it the other, and the record contains only the pass that was made. A keeper who consistently freezes the first presser buys his defenders half a second each time, and no dataset contains half a second.

Second-ball organisation is credited to the wrong player. A long delivery that leads to a recovered second ball is recorded as the goalkeeper's success, when the work was done by the midfielders who arranged themselves around the drop. Long distribution numbers therefore say as much about the team in front of the keeper as about the keeper, which is the same context problem that afflicts pressing metrics built on opposition passes.

Watching a keeper's distribution live

Six habits will tell you more in one match than any player card.

Watch his starting position when his team has the ball. On the line means he is a shot stopper. Twenty metres out and drifting means he is part of the build-up. Everything else follows from that.

Count the times he plays the ball back to himself. A goalkeeper who receives a return pass with the press committed is running a plan, not panicking, and the pass that follows is usually the important one.

Watch which foot he takes his first touch with. Then watch which shoulder the pressing forward approaches over. The second one answers whether the opposition has noticed the first.

Time the throws. A caught cross released inside three seconds is a counter-attack. The same catch released after eight is a reset, and now also a corner kick against him.

Judge the long balls by where the second ball lands. Not by whether a head met it. The organisation around the drop is the whole outcome.

Note when the short goal kicks stop. Sides abandon short build-up when they are ahead late, when the pitch is poor, and when the pressing side has worked them out. The moment it stops is more informative than the fact that it was happening.

Distribution is the part of goalkeeping that has changed most in thirty years and is still measured with the crudest tools of any position on the pitch. More on tactics, the laws and the money behind them is collected in the football archive, and the wider set of sports explainers sits in the blog index.