Analysis
Football possession statistics explained: what 62% means
Possession is calculated three different ways and the methods disagree. What pass share, stopwatch time and sequence counting each measure.
By CricketTaken EditorialPublished Analysis18 min read
The half-time graphic reads 62 to 38, the studio agrees that one side has dominated, and nobody asks the only question that would make the number mean anything: dominated by which count? Three different sums are in use across the industry, they were designed for different reasons, and applied to the same forty-five minutes they can be seven or eight points apart.
Football possession statistics explained honestly start there, with the admission that the figure on screen is a calculation rather than an observation. Nobody is timing the ball with a stopwatch and putting the result on air. Something is being counted, a share is being taken, and which thing gets counted decides which team looks dominant. That is before you reach the harder problem, which is that even a perfectly measured possession share tells you almost nothing on its own about who is winning the match.
The graphic says 62 to 38 and three different sums could have produced it
Consider a passage that lasts ninety seconds. One side plays twenty-eight passes between its defenders and its deepest midfielder, none of them travelling more than fifteen yards. The other side, in the same ninety seconds, wins the ball twice, launches it forward twice, and loses it twice, playing four passes in total.
Under a pass-share calculation the first team has taken 28 of 32 passes, which is 87 per cent of that ninety seconds. Under a stopwatch the first team has had the ball for perhaps seventy of those ninety seconds, which is 78 per cent, because both long clearances spent several seconds in the air belonging to nobody and two duels ate up the rest. Under a possession-count method the first team has had one possession and the second team has had two, and if you weight each possession equally the second team is ahead.
The last of those is a caricature, since nobody weights possessions equally without accounting for length, but it makes the point. Three defensible ways of asking "who had the ball" produce three answers, and the gap between them widens exactly as the football gets scrappier.
An orderly match between two technical sides gives you three numbers within a couple of points of each other. A wet Tuesday of second balls and long throws gives you three numbers that disagree enough to change the story.
Football possession statistics explained: the three sums in general use
Share of passes. Add up each side's passes, divide one by the total. For a long stretch this was the industry standard, defended on the reasonable ground that pass counts correlate very tightly with time on the ball and are far cheaper to collect. Its weakness is that it can only see the ball when somebody kicks it to somebody else. A player who runs with the ball for eleven seconds contributes one pass, or none.
Time in possession. Run a clock on which team is in control. This is the intuitive definition and the one every viewer assumes is being used. It requires somebody or something to decide, continuously, who is in control, which is easy for two thirds of a match and genuinely contested for the rest.
Share of possessions. Define a possession as beginning when a player takes a controlled touch and ending when control is lost through an interception, a tackle, a shot or a stoppage. Count them, or measure their combined duration, and take a share. This is the direction the major suppliers moved in, and it fixes the dribbling blind spot in the pass method because a carry is part of a possession whether or not it ends in a pass.
A fourth quantity sits behind all of them and is worth knowing about: the sequence. A sequence is a passage of play belonging to one team, ended by a defensive action, a stoppage or a shot. Sequences are the unit that most modern team-level analysis is built on, and they are not the same as possessions, because a sequence ends at a shot and a possession may not.
- 64%Share of passes, team A
- 57%Stopwatch time on the ball, team A
- 53%Share of possession sequences, team A
- 11%Gap between highest and lowest
An entirely invented forty-five minutes, constructed so that the three calculation methods separate visibly. The figures are chosen to show the direction and rough scale of the disagreement, not its true size in any real fixture. No actual match is described.
Why the pass-share method rewards the team that passes short
The pass method contains a hidden assumption: that every pass occupies roughly the same amount of match time. It does not.
A five-yard pass between two centre-backs standing twelve yards apart takes about a second from the receiver's first touch to the next player's first touch. A fifty-yard diagonal takes three or four seconds of flight and often a further second of control. Count them as one each, and the team playing short is credited with three or four times as much possession per second of actual play.
This is not a small effect. A side built to circulate the ball in tight triangles will show a pass-share possession figure several points above its stopwatch figure, consistently, across a season. A direct side shows the reverse. Two teams could hold the ball for identical lengths of time and be separated by ten points on a pass count.
The second distortion is the dribble. A winger who receives, drives forty yards and is fouled has held possession for six seconds and produced zero passes. In a pass-share world he did not have the ball. Teams whose progression runs through carrying rather than passing are systematically under-credited, which is a real problem given how many modern sides progress that way. Whether the ball moves through a player's feet or through the air is the distinction covered in the piece on how progressive passes are counted.
The third is the goalkeeper. A keeper who holds the ball for six seconds before releasing it produces one pass. A keeper who rolls it out immediately produces one pass. The stopwatch sees a large difference and the pass count sees none.
What each method does with the time nobody has the ball
A football match contains a great deal of time in which no team is in possession by any sensible definition, and the three methods handle it in three ways.
The pass method handles it by not noticing. Passes are events, events have no duration, and time in which no pass occurs simply is not part of the denominator. Dead time is invisible rather than allocated.
The stopwatch method has to make a decision, and the decision is usually to stop the clock. Ball out of play, foul, injury, substitution, goal celebration: the clock pauses and the share is taken over ball-in-play time only. That is the right choice, and it has the consequence that a match with sixty minutes of ball-in-play time and a match with fifty produce shares over different denominators, so a possession percentage does not translate into minutes without knowing the ball-in-play total.
Where the stopwatch approach gets genuinely difficult is contested time. A ball is in the air between a centre-back's clearance and a striker's chest. It is nobody's for two seconds. It can be assigned to the clearing team, assigned to nobody, or split, and different implementations do different things. Aerial duels, ricochets, blocked shots and the two or three seconds after a tackle when the ball is loose all fall into the same bucket. In a match of two hundred aerial duels that bucket is large.
The possession-count method resolves contested time by definition rather than by judgement. Nobody owns the ball until somebody takes a controlled touch, so the loose phase belongs to neither side and ends the previous possession cleanly. That is tidier, and it means the two teams' shares no longer sum to the whole match.
Where a possession starts, and the moment it is judged to end
The definition of a controlled touch is doing an enormous amount of work here, and it is the least examined part of the whole apparatus.
A defender who stretches out a leg and diverts the ball twenty yards to a teammate has not controlled anything, but the ball changed team. A midfielder who takes a deliberate touch out of his feet under no pressure clearly has. Between those two lies an entire category of half-controlled touches: the header that loops to a teammate, the block that rebounds usefully, the shin that redirects a pass.
Where the line falls determines the possession count, and the count determines the share. Set the bar for control low and both teams' possession totals rise, individual possessions get shorter, and a scrappy midfield battle looks like end-to-end football. Set it high and the same twenty minutes reads as a small number of long possessions with a lot of unowned time in between.
The ending is a second decision with the same character. A shot ends a sequence. Does it end a possession? If the shot is blocked back to an attacker, is that the same possession continuing or a new one beginning? If a cross is headed clear only as far as the edge of the box and returned, has possession changed twice or not at all?
None of this is arbitrary in a given system, because each supplier writes it down and applies it consistently. It is arbitrary across systems, and that is the source of the gap between two published figures for one fixture. Consistency within a feed is what makes season-long comparison valid. It is also why mixing feeds inside one comparison quietly ruins it.
Set pieces, throw-ins and the ball in the goalkeeper's hands
The dead-ball edge cases deserve their own treatment because they are where the biggest single distortions hide.
A team defending a corner does not have the ball. A team taking one, on most definitions, does, from the moment the taker is judged to be in control. A match with eighteen corners for one side therefore hands that side a quantity of possession that had nothing to do with anybody keeping the ball.
Throw-ins do something similar and more often, because there are far more of them. A side that wins repeated throw-ins in the opposition half accumulates possession credit each time, even where the throw is contested and lost immediately.
The goalkeeper holding the ball is the case that most upsets viewers. A keeper who catches a cross and stands with it for eight seconds is not doing anything a supporter would call possession, and a stopwatch method that counts it inflates the defending side's share during precisely the passage where it is under most pressure. Implementations vary: some stop the clock while the ball is held, some do not, and the choice materially changes the numbers for a side that defends deep. The reasons a side chooses to sit in and concede the ball are set out in the explainer on defending in a low block.
Substitutions, injuries and goal celebrations are handled more consistently, since almost everybody excludes them, but they still change the denominator. A match with three long stoppages has a smaller ball-in-play total, so every remaining second carries more weight in the final share.
Worked example: six invented minutes, three running totals
Watching the methods diverge in real time makes the abstraction concrete.
- Team A restarts with a goal kick, played shortThe keeper rolls it to a centre-back who plays two sideways balls. Pass method: three passes for A, none for B. Stopwatch: fourteen seconds. Sequence method: one possession opened.
- Centre-back to holding midfielder to full-backTwo more passes, ten more seconds, still the same possession. Pass method has A at five. The three methods are still in rough agreement and will not stay that way.
- Full-back drives forward with the ball for eleven secondsNobody touches it but him. Pass method: nothing added, because a carry is not a pass. Stopwatch: eleven more seconds for A. Sequence method: the possession continues. The first real divergence has appeared, and it favoured the stopwatch.
- He is tackled, the ball runs loose for three secondsNeither side controls it. The stopwatch either pauses or splits the three seconds. The sequence method opens nothing until somebody takes a controlled touch. The pass method has no concept of these three seconds at all.
- Team B wins it and hits a fifty-yard clearanceOne pass for B against A's five. Four seconds of flight. Pass method now reads A 5, B 1, or 83 per cent. Stopwatch reads roughly A 35 seconds, B 4 seconds, or 90 per cent, though only if the flight time is credited to B.
- The clearance is headed back by A, contested twice moreTwo headers each, all four inside eight seconds. Pass method: headers logged as passes only if they find a teammate, so perhaps one each. Stopwatch: eight seconds of genuinely contested ball that has to be assigned somehow. Sequence method: possibly four possessions, all under three seconds.
- Team B settles on the ball and keeps it for ninety secondsTwenty-two short passes, no carries of note. Pass method: B closes the gap fast, twenty-two against A's six. Stopwatch: ninety seconds against A's thirty-five. Sequence: one long possession for B against A's one long and a scattering of fragments.
- The ball goes out for a throw-in, forty seconds passNobody has possession. The stopwatch pauses. The pass method is unaffected because no passes occurred. The denominators have now permanently separated.
- Six minutes closedPass share puts B ahead on volume of short passes. The stopwatch puts B ahead by more, because A's carry is worth less than B's ninety-second spell. The sequence count puts them closer together, because A's fragments each count. Same six minutes, three defensible answers.
A wholly invented passage of play, constructed to show which method credits which event. No real match, team or figure is described. Running totals are kept for team A under each of the three methods.
The four ways a possession figure is manufactured
Nothing that follows involves anybody breaking a rule. These are ordinary coaching instructions that happen to move the number.
Circulating behind the first line. Two centre-backs, a goalkeeper and a dropping midfielder can exchange passes indefinitely against a side that has no interest in pressing. Each exchange is a pass, each pass is possession, and the ball has gone nowhere. A team can build a twenty-point possession advantage this way without threatening once.
Taking the ball when the opponent gives it away. A side defending a lead in a low block is not trying to have possession. It concedes the ball on purpose to keep eleven players between the ball and its goal. The team in front of it accumulates a large share by default, and the number describes the opponent's plan rather than its own quality.
The goalkeeper as an outfield player. Every backward pass to a keeper who then plays it back out is two passes and several seconds. Sides that build patiently from the back generate a great deal of possession volume through this loop, which is a legitimate tactic and also a large contributor to a headline figure.
Managing the clock in front. A leading side that keeps the ball in the corner, takes its time over restarts and recycles rather than attacking is accumulating possession as a defensive act. In a pass-share world it looks like dominance. It is the opposite.
The unifying point is that possession is a resource, not an achievement, and every one of the four behaviours above spends it differently. What the resource is spent on is the thing worth measuring, and the possession figure by construction cannot see it.
- Own third41.0%
- Middle third44.0%
- Final third15.0%
An entirely invented distribution of one invented team's passes across the thirds of the pitch during one invented match, constructed to show what a high possession figure can conceal. Nothing here is measured and no real team is described.
Show the numbers
| Item | Value |
|---|---|
| Own third | 41% |
| Middle third | 44% |
| Final third | 15% |
Sterile possession, and the shape that produces it
The phrase gets used loosely. It has a specific shape on the pitch and it is recognisable within about four minutes of watching.
The ball travels from a centre-back to a full-back, forward to a winger who has no route inside, back to the full-back, back to the centre-back, across to the other centre-back, and out to the opposite full-back. It traces a U around the outside of a compact defensive block, never entering it. Ten or twelve passes per circuit, three or four circuits per spell, and the possession column climbs steadily.
What makes it sterile is not the passing itself but the absence of anybody occupying the space inside the block. A circulation pattern only becomes penetration when a player receives between the lines and turns, and that requires somebody to make a movement that risks losing the ball. Sides that are anxious about counter-attacks stop making those movements, and the circulation becomes an end in itself.
The tell is the ratio of final-third entries to possession share. A side with two thirds of the ball and a handful of entries into the last third has spent ninety minutes doing something that resembles control and produces nothing. A side with less of the ball and more entries has been more dangerous with less.
There is also a defensive version of the same trap, where a pressing team allows the circulation deliberately because it is happy for the opponent to hold the ball in front of it. What triggers a press to break out of that arrangement is set out in the piece on counter-pressing after a turnover, and the counting rule for pressing intensity is covered in the PPDA explainer.
Field tilt, and why that has two definitions as well
The obvious fix for sterile possession is to stop counting all the ball and start counting only the ball that matters. Field tilt does that.
It takes each team's share of the play that happens in the attacking third: your final-third figure divided by the sum of yours and your opponent's. Above fifty per cent means more of the match's dangerous-area play belonged to you.
The catch is exactly the one this whole subject keeps producing. The metric can be built from final-third passes or from final-third touches, and different publishers use different quantities. Passes and touches are correlated and not identical, since dribbling and shooting generate touches without generating passes. A side that runs at defenders in the final third will read higher on a touch-based tilt than a pass-based one.
Even so, tilt is a much better description of territorial control than raw possession. A common and useful pattern is the side with sixty per cent of the ball and under forty per cent field tilt: dominant in its own half, absent in the opponent's. That combination has a name in every coaching room and it is not a compliment.
Tilt still cannot say whether the final-third play was any good. Twenty sideways passes twenty-five yards from goal count identically to twenty passes that break a line. Chance quality remains a separate question, and the model that answers it is described in the piece on what expected goals actually measures.
Possession-adjusted defensive numbers, and the assumption inside them
Possession also gets used as a divisor, and that deserves a warning of its own.
A defender in a team with seventy per cent of the ball has fewer opportunities to tackle and intercept than one in a team with thirty. Comparing their raw defensive counts is meaningless, so analysts adjust: divide the defensive actions by the share of the match the player's team spent without the ball, and scale to a common baseline.
The adjustment is necessary. The assumption underneath it is that defensive opportunities scale linearly with time out of possession, and that is not quite true. A team that spends thirty per cent of the match defending in a compact shape near its own box faces a different kind of defending from one that spends thirty per cent of the match defending in transition across a stretched pitch. The opportunities are not merely fewer, they are different in type.
The result is that possession adjustment corrects a large bias and introduces a small one. A defender in a high-possession side who is adjusted upwards is being credited on the assumption that he would have made proportionally more tackles given proportionally more chances, which the best defenders in such sides often would not, because their job is to prevent the situation arising at all.
Use adjusted numbers for cross-team comparison and unadjusted ones for describing what actually happened. Quoting an adjusted figure as though it were an observation is a category error that appears in recruitment presentations more often than anyone would like.
Why possession and winning are correlated, and why the arrow points the wrong way
Teams that have more of the ball win more often. This is true and it is one of the most misused facts in the sport.
Part of the relationship is straightforward: better teams tend to keep the ball better, and better teams win more. Possession is a symptom of quality rather than a cause of it. Give a weak side more of the ball and it will not become stronger.
The larger part is game state, and it runs backwards. A team that scores first very often sees its possession share fall, because the opponent now has to attack and it does not. A team that concedes first sees its share rise for the same reason. Full-time possession figures are therefore contaminated by the scoreline they were accumulated under, and a side that leads for seventy minutes will post a possession figure that describes its opponent's desperation.
This is why the interesting number is possession before the first goal, or possession while the match is level. It removes most of the game-state effect and leaves something closer to a description of how the two sides wanted to play. Very few published figures do this, and almost no broadcast graphic does.
The practical rule: never interpret a possession figure without the scoreline timeline next to it. A 65 per cent share in a match won 1-0 with the goal in the ninth minute means something entirely different from the same share in a match won 1-0 with the goal in the eighty-ninth.
Sequence length and directness: the two numbers that describe how possession is spent
A share tells you how much. Two derived figures tell you what kind, and they are far more diagnostic than the headline.
Passes per sequence is the average number of passes in a team's uninterrupted passages of play. A high figure describes a side that builds in long chains. A low figure describes a side that moves the ball forward in three or four actions and either creates something or loses it. Neither is better in the abstract, and the same season-long possession share can sit on top of either.
Direct speed measures how quickly a team's sequences travel up the pitch, usually expressed as metres of progress towards goal per second of possession. It separates two teams with identical pass counts per sequence into the one that passes sideways and the one that passes forward.
Put the two together and the four combinations describe most of professional football. Long sequences with low speed is patient positional play. Long sequences with high speed is a side that builds carefully and then attacks the moment a door opens. Short sequences with high speed is direct or counter-attacking football. Short sequences with low speed is usually a team in trouble, playing hurried square balls under pressure and giving the ball away.
None of that is visible in a possession percentage. Two sides can both post 55 per cent and belong in opposite quadrants.
There is a third figure worth knowing, though it is published less often. Sequences that begin in the attacking third, sometimes labelled high turnovers, count the passages that started because the opponent lost the ball near his own goal. A side that generates a lot of these is manufacturing possession in the most valuable place on the pitch, and that possession is worth several times an equivalent spell begun with a goal kick. The figure belongs next to the possession share every time, because it distinguishes territory that was earned from territory that was conceded.
The reason these breakdowns matter more than the headline is that they survive game state better. A leading team's possession share is dragged around violently by the scoreline. Its passes per sequence and its direct speed move much less, because they describe the mechanics of how the side plays rather than how much of the ball the opponent chose to give it. For season-level judgement of a team's identity, the derived figures are simply the better instrument.
They also make it possible to ask sensible questions about a manager's effect. A new appointment who lifts possession from 48 to 57 per cent may have changed everything or may have inherited a fixture list of weaker opponents. The same appointment doubling the number of sequences starting in the attacking third has changed something structural, and that change is much harder to explain away.
Reading a possession figure without being fooled
Six checks, and they take about twenty seconds each.
Which method produced it? If the source does not say, treat it as unusable for comparison against a figure from anywhere else. A five-point gap between two publishers is normal and neither is wrong.
What was the ball-in-play time? A percentage of fifty minutes and a percentage of sixty-five minutes are different quantities wearing the same symbol.
What was the score, and when? Game state moves possession more than tactics do in most matches.
Where was the ball? Possession share plus field tilt tells you far more than either alone. High share with low tilt is the signature of circulation without penetration.
How did it end? Final-third entries, shot volume and chance quality say what the possession was converted into. Possession is an input and not one of the things that wins matches by itself.
Whose plan was it? A low share is sometimes a deliberate design rather than a failure, and reading it as failure is the single most common mistake made about defensive sides.
One further habit is worth building, because it costs nothing. Whenever a possession figure surprises you, look for the second half separately. Matches change character after the interval more reliably than they change anything else, and a side that held 70 per cent before the break and 45 per cent after it has usually made a substitution, conceded a goal or decided to defend a lead. A single full-time number averages those two matches into one and hides the transition that explains the result.
The figure is not useless. Tracked across a season, from one consistent source, and read against territory and chance creation, it describes a team's method well. Quoted on its own at half-time, it is a number that sounds like an argument and contains none.
The rest of our work on how the modern game is measured sits in the football section, and every explainer we have published is listed on the blog.
Common questions
How is possession calculated in football?
There are three methods in general use and they do not agree. One divides a team's completed passes by the two teams' combined passes. One runs a stopwatch on which side has controlled the ball. One counts discrete possessions, where a possession begins with a controlled touch and ends when control is lost, and expresses each team's share of the total. Broadcast graphics rarely say which they are using.
Does possession percentage measure time on the ball?
Often it does not. For many years the headline figure was a share of passes rather than a share of time, on the grounds that the two track each other closely. They track closely in an ordinary match and come apart badly in a scrappy one, which is exactly the match where the number gets quoted.
Why do two websites show different possession figures for the same match?
Because they use different methods, different treatments of dead-ball time, and different rules about who owns a loose ball. A five to eight point gap between two published figures for one fixture is unremarkable and does not mean either is wrong.
What is sterile possession?
Possession that does not advance or threaten. A side circulating between its centre-backs and holding midfielder can accumulate a large share while making no attempt to enter the final third. The possession figure cannot see the difference, which is why territorial and progression measures are read alongside it.
What is field tilt in football?
It is a team's share of the play that happens in the attacking third, calculated from either final-third passes or final-third touches depending on who is publishing it. It answers where the ball spent its time rather than which team had it, and it separates territorial dominance from mere circulation.
Does having more possession make a team more likely to win?
Possession and results are correlated, but much of the correlation runs the other way. A team that goes ahead often concedes the ball deliberately, and a team that falls behind takes it because the opponent hands it over. Reading a full-time possession figure without checking the scoreline it was accumulated under produces the wrong conclusion most of the time.
Filed under Football·football · analytics · possession · tactics · data · broadcasting