How possession is counted, and why the figure disagrees with itself
The possession figure flashing on screen during a football match is not the universal constant it appears to be. Different data providers use fundamentally different counting conventions, and the same ninety minutes can produce starkly different percentages depending on whether the graphic reflects time in control or share of completed passes.
In this piece
Possession as a Measurement Convention, Not a Definition of Dominance
Broadcasters treat possession as self-evident: whoever has the ball more, wins the statistic. The reality is more mechanical. Possession is a counting rule, and the rule being applied determines the number you see.
One method tracks time. As a sports analytics site explained in 2023, this approach records how long each team maintains control of the ball during the match, with control defined broadly as the team that last touched the ball. The clock runs until possession changes.
The other method counts passes. A 2013 post quoted Opta’s then-public explanation that possession was calculated by totaling each team’s passes and dividing by the match total to produce a percentage. NBC Sports reported in 2012 that Opta’s possession figure appeared to be based on pass attempts divided by total pass attempts, and not on a running clock. NBC Sports also reported that Opta’s own published post said its approach "came up with exactly the same figures" as time-based methods "on almost every occasion."
The Provider Definitions: Stats Perform and the Opta Legacy
Stats Perform, which now operates the Opta data brand, has published a detailed "Possessions Framework" that clarifies how its system structures the count. A sequence is defined as a passage of play belonging to one team, ended by defensive actions, stoppages in play, or a shot. A possession is one or more sequences in a row belonging to the same team.
The framework specifies that a possession ends only when the opposition gains control of the ball. This matters for how sequences and possessions relate: a series of passes ending in a shot saved for a corner counts as one possession but more than one sequence, because the ball has gone out of play. The same team retains control, so the possession continues.
Stats Perform also notes that the number of possessions belonging to each team can differ by only one in a given match. However, the time of possession by each team and the number of sequences within each team's possessions need not be equal. The framework allows for asymmetry: one team can have longer possessions with fewer sequences, or shorter bursts with more rapid exchanges.
What Counts as a Possession: Starts, Stops, and Continuities
The Stats Perform framework illuminates how granular these definitions become. A sequence starts with a player making a controlled action on the ball, including passes but not defensive events such as tackles and interceptions—unless those defensive events are followed by a controlled action such as a pass or dribble.
This creates edge cases that matter for the count. A tackle that immediately clears the ball ends a sequence. A tackle followed by a dribble begins a new one. A shot saved by the goalkeeper, a ball played out for a throw-in, a corner kick—all of these end sequences without necessarily ending possessions, provided the same team maintains control.
The distinction between sequence and possession is not academic. It determines how analysts and broadcasters present the flow of a match, and it explains why possession statistics can feel disconnected from the watching experience. A team that dominates sequences without dominating time may generate misleadingly high pass-share possession while appearing to chase the game.
Why the Same Match Can Disagree With Itself
The divergence between time-based and pass-share methods produces practical contradictions that viewers encounter regularly. One broadcast shows 62% possession; a rival graphic on the same match displays 48%. Both cannot be correct under the same definition. Both can be correct under different definitions.
The time-based method rewards teams that hold the ball without acting. A goalkeeper cradling the ball for thirty seconds, a defender shielding at the corner flag, a midfield patient in buildup—all accumulate time without generating events. The pass-share method rewards activity. A team that moves the ball quickly, that completes passes in tight spaces, that maintains rhythm through rapid exchange, will score highly on pass-share even if the ball changes hands frequently.
Neither method captures "control" in any meaningful tactical sense. A team can lead on possession by either metric while being systematically dismantled on the counter. The statistic measures what it measures: time or passes. It does not measure danger, pressure, territorial dominance, or probability of winning.
The Reporting Rule: Name the Method Before the Number
For anyone citing possession statistics—analysts, commentators, journalists—the obligation is straightforward. The number is meaningless without its denominator. "Possession" is not a single thing. It is at least two things, and possibly more depending on the provider's specific implementation.
The honest formulation is: "Team A led on time-based possession, 62% to 38%" or "Team B led on pass-share possession, 55% to 45%." The dishonest formulation, and the common one, treats the percentage as interchangeable across contexts, as if Opta and a manual time-tracker are counting the same events.
They are not. Stats Perform's framework makes explicit what many broadcasts obscure: possession is a constructed statistic, built from definitional choices about what constitutes control, when a sequence begins and ends, and whether continuity or activity is the relevant measure. The same match can disagree with itself because the disagreement is built into the method.
Statistical definitions differ between providers. Figures quoted here are attributed to the provider that published them.