Possession Value Explained: How Football Actions Create Attacking Value

Possession value is a family of football analytics models that estimates how much an action changes a team’s likelihood of scoring or conceding. Instead of counting every pass or carry equally, a possession value model asks whether the action improved the situation.

A sideways pass between centre-backs may retain possession without adding much attacking value. A line-breaking pass into space between the opposition midfield and defence may increase the probability of creating a goal, even if the move does not immediately produce a shot.

This allows analysts to evaluate the attacking process before the final chance. Expected Threat, commonly known as xT, is one form of possession value. Other frameworks use richer information to assess passes, carries, dribbles, turnovers, defensive actions and the changing risk of conceding.

What Does Possession Value Mean in Football?

Traditional possession statistics measure how long each team controls the ball. Possession value tries to measure what a team does with that control.

The central principle is that not all football situations are equally valuable. Having the ball near the opposition penalty area is usually more promising than having it beside your own corner flag. Central possession may offer different passing and shooting options from possession near the touchline. Space, pressure, teammates, opponents and the current match situation can alter the value further.

A possession value model represents the game before and after an action. It then estimates how the action changed the team’s expected outcome.

In simplified form:

Action value = value of the situation after the action − value of the situation before the action

If a pass moves the ball from a state valued at 0.02 expected goals to one valued at 0.07, the pass has added 0.05 of attacking value under that model.

The figure is an estimate, not a statement that the pass literally created five per cent of a goal. Its meaning depends on the model’s target, data and assumptions.

Why Possession Percentage Is Not Enough

Possession percentage records control of the ball but does not distinguish between harmless circulation and productive progression.

Imagine two teams each have 50% possession:

• Team A repeatedly moves the ball across its defensive line but struggles to enter the final third.
• Team B attacks quickly, breaks midfield lines and repeatedly reaches the edge of the penalty area.

The possession split treats their control as equal. Their attacking impact may be very different.

Even completed passes can be misleading. A player completing 70 safe passes may have less influence on attacking danger than a midfielder completing 30 passes that regularly eliminate defenders or find teammates between the lines.

This is why useful analysis must move beyond raw counts. GoalIQAI’s guide to the football statistics that actually matter explains that a metric becomes valuable when it captures meaningful performance and survives contextual scrutiny.

How Possession Value Models Work

There is no universal possession value calculation. Different models define the game state, outcome and time horizon differently. Most nevertheless follow the same broad process.

1. Describe the situation before an action

A basic model may use only the ball’s location. More advanced models can include:

• The type and location of recent actions.
• The score and time remaining.
• Whether the team is playing at home or away.
• The positions of teammates and opponents.
• Defensive pressure on the player in possession.
• The speed and direction of the attack.
• Whether losing the ball would expose the team to a counterattack.

The more detailed description may capture football more accurately, but it also requires richer data and a larger sample.

2. Estimate the value of the game state

The model assigns an expected value to the situation. Depending on its design, this could represent:

• The probability of scoring during the current possession.
• The probability of scoring within the next few actions.
• The probability of scoring within a defined number of seconds.
• The combined probability of scoring and conceding.
• The expected value of the next goal.

These targets are related but not identical. Two possession value models can therefore assign different values to the same action without either calculation necessarily being wrong.

3. Recalculate after the action

Once a pass, carry, dribble or other event occurs, the model estimates the value of the new state. The difference between the two estimates becomes the action’s value.

A successful pass into the penalty area will normally add attacking value. A turnover near the team’s own goal is likely to produce a negative value. A backwards pass may reduce immediate attacking threat but still be tactically sensible if it helps the team escape pressure or reorganise the attack.

Expected Threat as a Possession Value Model

Expected Threat is one of the best-known possession value approaches. A typical xT model divides the pitch into zones and estimates the probability that possession in each zone will eventually lead to a goal.

A pass or carry receives positive xT when it moves the ball into a zone with greater scoring potential:

xT added = destination xT − starting xT

This makes xT useful for measuring ball progression before a shot occurs. It can identify line-breaking passes, carries into the final third and other actions that increase attacking danger without appearing in goals, assists or Expected Goals.

However, a basic location-based xT model does not see the complete football situation. Two players may receive the ball in the same zone under very different conditions. One may have time, space and several forward options; the other may be surrounded by defenders and facing away from goal.

xT is therefore better understood as one practical implementation of possession value, not as a complete measure of every action or decision.

Other Possession Value Frameworks

More complex models attempt to describe the context and consequences of actions more fully.

VAEP

VAEP stands for Valuing Actions by Estimating Probabilities. The framework, developed by researchers at KU Leuven, values an action according to how it changes the team’s probability of scoring and conceding in the near future.

This gives the model an offensive and defensive component. A risky pass can increase the chance of scoring when successful while also carrying a cost if it leads to a dangerous turnover.

The researchers describe VAEP as a way to assign value to on-ball events using the difference between the game states before and after each action. Their public VAEP explanation also makes clear that the original implementation represents the current state using the previous three actions. That is a modelling choice rather than a complete representation of everything happening on the pitch.

Expected Possession Value

Expected Possession Value, or EPV, can use tracking data to model the positions and movements of the ball and all 22 players.

The research framework developed by Javier Fernández, Luke Bornn and Daniel Cervone estimates the expected outcome of a possession at a particular moment. Its richer spatial information can help evaluate not only what happened, but also potential passing and carrying options that were available.

This matters because the recorded action is not always the best available decision. A completed sideways pass may retain possession, but tracking data could reveal that a more valuable forward pass was open. Event data records the chosen pass; richer models can begin to examine the decision itself.

Commercial on-ball value models

Data providers have also developed their own possession-state models. Their definitions, training targets and inputs can differ.

For example, Hudl StatsBomb describes its On-Ball Value model as estimating how events affect a team’s likelihood of scoring and conceding. Its public explanation of possession value models distinguishes OBV from simpler xT approaches and from models trained only on goals scored.

These provider-specific metrics may be useful, but identical labels do not guarantee identical calculations. Analysts should establish what a model includes before comparing numbers across platforms.

A Practical Possession Value Example

Consider an attack that begins with a centre-back:

1. The centre-back passes sideways to a full-back.
2. The full-back carries forward and attracts an opposing midfielder.
3. A central midfielder receives behind that player.
4. The midfielder passes between two defenders to release a winger.
5. The winger’s attempted cutback is intercepted.

The possession creates no shot and therefore no xG. Conventional statistics record several completed passes, one carry and an unsuccessful cross or pass.

A possession value model can describe how the attacking potential changed throughout the sequence.

• The opening sideways pass may add almost no value.
• The carry may add value by advancing possession and drawing pressure.
• The line-breaking pass may produce the largest increase.
• The final interception may remove much of the accumulated value.

This produces a richer account than simply declaring that the attack failed because it generated no shot. It identifies the actions that helped create danger and the point at which that danger disappeared.

What Possession Value Can Reveal

Ball progression

Possession value can distinguish progression that genuinely increases danger from movement that merely gains distance.

A 20-metre pass down the touchline may advance the ball without creating a strong attacking position. A shorter pass into a central pocket may eliminate several defenders and increase the range of available options.

Players involved before the assist

Goals and assists concentrate credit near the end of an attack. Possession value can identify the defender who broke the first line, the midfielder who carried through pressure or the player who made the pass before the assist.

This is one reason analysts examine metrics beyond xG. Shot quality remains important, but it does not explain every contribution that made the chance possible.

Risk and reward

Players who attempt ambitious actions may add substantial value when they succeed but also surrender dangerous possession when they fail.

A useful model should account for both sides. Looking only at positive contributions can reward players for volume and aggression while ignoring the cost of unsuccessful decisions.

Team attacking patterns

Aggregated possession value can show where and how a team creates danger. Analysts can examine whether value is generated through central combinations, wide progression, counterattacks, carries or long passing.

This complements field tilt. Field tilt measures the share of advanced territorial possession, whereas possession value tries to estimate the attacking quality created by individual actions and situations.

How Clubs Can Use Possession Value

Football clubs can apply possession value in several areas.

• Recruitment: identifying players who progress the ball, create useful situations or combine attacking output with controlled risk.
• Performance analysis: finding where a team’s build-up adds or loses value.
• Opposition analysis: identifying the zones, players and actions through which an opponent creates danger.
• Player development: assessing whether a player recognises and executes valuable options.
• Tactical evaluation: comparing how different structures move the ball into dangerous situations.

These applications still require video and tactical interpretation. A model may reveal that a full-back adds little possession value, but the team’s structure may deliberately require that player to circulate the ball safely while others take creative risks.

The number describes an outcome under the model. It does not explain the player’s instructions by itself.

Why Context Changes Possession Value

The same action can mean different things depending on the match situation.

A team protecting a one-goal lead late in the match may reasonably prefer a low-risk backwards pass to an ambitious progression. A trailing team may need to accept greater turnover risk. A red card can change the available space and the cost of losing possession.

This connects possession value with game state in football analytics. Score, time, player numbers and competition context influence team behaviour. If a model does not include those variables, its output must be interpreted alongside them.

Team strength matters as well. A player in a dominant side may accumulate more possession value because the team has more of the ball, occupies better territory and surrounds the player with stronger options. Raw totals should not be treated as context-free rankings of ability.

The Main Limitations of Possession Value

The model values what it can observe

Event data records actions such as passes, carries, shots and tackles. It may not capture off-ball movement, defensive positioning, communication or the run that created space for somebody else.

Tracking data provides more spatial detail but still requires modelling choices about pressure, options and intent.

Outputs depend on the target

A model predicting a goal within ten seconds is answering a different question from one estimating the probability of scoring before the possession ends. Results from the two models should not be assumed to be interchangeable.

Team and role effects remain

Possession value depends partly on opportunity. Dominant teams generate more actions in valuable areas. Set-piece takers and primary creators may accumulate more value because of their roles.

Per-90 figures, value per touch and possession-adjusted measures can help, but no normalisation removes every contextual difference.

A higher value does not always mean a better decision

Models estimate expected consequences from historical patterns. They do not fully know the manager’s tactical instruction, the player’s physical condition or every option perceived on the pitch.

An action that reduces immediate attacking value may improve control, preserve energy or prevent a dangerous transition. Football decisions have objectives that a single metric may not contain.

Possession value is not automatically predictive

A useful descriptive metric does not necessarily improve forecasts. Analysts must test whether possession value adds information beyond established variables such as shot quality, team strength and market expectations.

As with every advanced statistic, the question is not whether the metric sounds sophisticated. It is whether it provides reliable information for the decision being made.

How to Interpret Possession Value Properly

Before using any possession value figure, ask:

• What outcome is the model estimating?
• What time horizon does it use?
• Does it measure both scoring and conceding risk?
• Which actions can receive value?
• Are unsuccessful actions and turnovers included?
• Does the model use event data, tracking data or both?
• Does it account for game state and defensive pressure?
• Is the player’s role or team possession share considered?
• Has the metric been validated for the intended use?

This is a better approach than treating possession value as one universal statistic. The label describes a modelling family. Its numbers only become meaningful when the underlying definition is understood.

Key Takeaways

• Possession value estimates how an action changes a team’s likelihood of scoring or conceding.
• It measures the quality of possession rather than simply the amount of possession.
• Expected Threat is a possession value model focused mainly on how passes and carries move the ball between areas.
• VAEP, EPV and commercial models can incorporate wider actions, scoring risk, conceding risk and richer context.
• Different models can produce different values because they use different data, targets and time horizons.
• Possession value can reveal progression, build-up contribution, risk and team attacking patterns before a shot occurs.
• The metric must still be interpreted alongside tactics, player roles, team strength and game state.
• Possession value is a tool for asking better questions, not a complete measure of football performance.

Related Guides

• Expected Threat (xT) Explained
• Field Tilt Explained
• Game State in Football Analytics Explained

Continue Building Your Football Analytics Knowledge

Possession value is one part of a broader analytical framework. Explore the GoalIQAI Football Betting & Analytics Knowledge Base for evidence-led guides to football metrics, probability, markets and decision-making.

Subscribe to GoalIQAI for new football analytics guides and evidence-based analysis. The aim is not to promise certainty, but to help you interpret football information more clearly.