Expected Pass Completion (xPass) Explained

Expected pass completion, usually shortened to xPass, estimates how likely a football pass was to reach its intended teammate given the pass’s difficulty and context. A routine five-metre recycle may receive a probability close to 1.00, while a pressured diagonal through a crowded lane may be far lower. Adding those probabilities gives the number of completed passes a model would expect; comparing that estimate with actual completions can show whether a player completed more or fewer passes than expected.

xPass improves on raw pass completion by recognising that not all attempts are equally difficult. It does not, however, tell us whether the pass was progressive, valuable or the best decision available. Those questions require separate measures and tactical context.

What Is Expected Pass Completion?

Expected pass completion is a modelled probability that a pass will be completed. It is normally estimated from historical passes with similar characteristics and expressed between zero and one, or as a percentage.

A pass assigned 0.80 xPass has an estimated 80% completion probability under that model. This does not mean that the individual pass is partly complete or that exactly four of the next five similar attempts must succeed. It means that a well-calibrated model would expect roughly 80% of a large group of comparable passes to be completed.

The terminology is not completely standardised. Depending on the provider or analyst, “xPass” may describe:

• the completion probability assigned to one pass;
• the sum of expected completions across several passes;
• an expected completion percentage; or
• a related inverse estimate of turnover probability.

The underlying definition should therefore be checked before comparing figures. The peer-reviewed Expected passes study, for example, defines xPass as the predicted probability that a pass will be completed to a teammate using contextual event and positional information. Related research has framed the inverse question through Expected Pass Turnovers.

How Is xPass Calculated?

An xPass model is trained as a probability model. Each historical attempt has an observed result—completed or incomplete—and a set of features describing the situation. The model learns how those features have been associated with completion in its training data.

For pass i, let:

• pᵢ be the estimated probability of completion;
• cᵢ be the observed result, where one means completed and zero means incomplete; and
• N be the number of passes assessed.

The main aggregate calculations are:

Expected completed passes = Σpᵢ

Expected completion percentage = (Σpᵢ ÷ N) × 100

Actual completion percentage = (Σcᵢ ÷ N) × 100

Completion above expected = actual completion percentage − expected completion percentage

The difference can also be reported as actual completed passes minus expected completed passes. Analysts should state whether they are using pass counts or percentage points because the two outputs are not interchangeable.

Typical Model Inputs

The available features depend on the data source. An event-data model may use only recorded action details, while a tracking model can describe the positions and movement of the ball, passer, intended receiver and defenders.

• Pass geometry: starting point, intended destination, distance and angle.
• Pass type: ground pass, aerial ball, cross, through ball, set piece or other provider-defined class.
• Defensive pressure: opponents near the passer or receiver and their ability to close the passing lane.
• Receiver context: available space, movement, body orientation and distance from defenders.
• Ball and player movement: speed, direction and estimated interception possibilities when tracking data are available.
• Match context: pitch area, phase of play, game state or competition, where the model design includes them.

More inputs do not automatically produce a better analytical answer. The features must match the question. A retrospective model may use information observed shortly after the ball is kicked, whereas a model assessing the options available before execution should use only information that was knowable at that moment.

A 2026 study on temporal leakage showed why this distinction matters for pass-turnover models: post-execution descriptors can improve retrospective prediction while making a model less suitable for real-time tactical decisions.

Worked xPass Example

Consider five illustrative passes by a midfielder. These probabilities are fictional and demonstrate the calculation rather than a universal xPass scale.

The player completed three of five passes, producing an actual completion rate of 60%. The model expected 2.89 completions, or 57.8%. The player therefore finished 0.11 completed passes above expectation, equivalent to 2.2 percentage points above expected.

That is a calculation, not evidence of exceptional passing skill. One changed outcome would move the actual percentage by 20 points in this tiny sample. Meaningful evaluation requires many more attempts, an appropriate comparison group and stable model performance.

What Completion Above Expected Can Show

Raw completion percentage treats every pass equally. A player completing 90% of mostly uncontested short passes and another completing 90% while attempting more pressured or ambitious passes receive the same basic statistic. xPass provides a difficulty-adjusted baseline.

Across a suitable sample, actual completion above expected can help analysts investigate:

• whether a player consistently completes harder passes than the model predicts;
• how much risk a player’s passing role contains;
• whether a team retains possession effectively under pressure;
• which opponents or pitch areas create repeated passing problems;
• whether a player’s raw completion rate is supported by safe selection or strong execution; and
• how passing performance changes after a tactical, positional or team change.

The word “investigate” matters. Completion above expected is affected by more than the passer’s technique. The receiver’s movement and control, team spacing, tactical instructions, opponents and the provider’s definition of a successful pass can all influence the result.

xPass vs Progressive Passes, xT, Packing and xA

xPass measures estimated completion difficulty. Other passing metrics answer different questions.

A pass can have low xPass because it is difficult but create little value if completed. Another can be relatively easy while moving possession into a much more dangerous area. Threshold-based progressive passes and carries describe meaningful territorial advancement, while Expected Threat (xT) estimates the change in possession value.

Similarly, packing rate and line-breaking passes describe the opposition structure penetrated. They may help explain why an attempted pass was ambitious, but they do not replace a completion-probability model.

How Professionals Can Use xPass

Separating Execution from Pass Selection

An observed pass model can estimate how difficult the selected passes were and whether they were completed. Assessing decision quality is harder because the analyst must also consider the alternatives that were available.

A player may complete a difficult pass when a simpler, more valuable option was open. Conversely, a failed low-probability pass may have been a defensible choice late in a match when the team needed to attack. xPass describes execution risk; it does not settle the tactical decision.

Evaluating Roles and Team Style

Centre-backs, deep midfielders, full-backs and creative attackers face different passing distributions. Comparisons should therefore consider role, receiving options, possession share, tactical responsibility and the areas in which attempts begin and end.

Team analysis can also separate controlled circulation from higher-risk progression. A side may post a high raw completion rate because it recycles possession safely, while another accepts more passing risk to attack space quickly.

Scouting and Recruitment

xPass can help identify players whose raw accuracy is shaped by pass difficulty. Recruitment analysts can examine whether strong completion survives pressure, longer distances or more aggressive destinations rather than relying on one headline percentage.

Transfer projections still require caution. A player moving into a different league or tactical system may receive less time, different passing options or teammates with different movement and control qualities.

Pressing Analysis

At team level, expected completion or turnover models can highlight areas and opponents associated with vulnerable distribution. This may support pressing preparation, but it should be combined with video and tactical evidence. A low expected-completion area may reflect deliberate long clearances rather than a reliable pressing target.

Limitations and Common xPass Errors

Treating the Model Estimate as a Fact

xPass is learned from data and model assumptions. It is not the true hidden probability of a pass. Different samples, features and algorithms can assign different probabilities to the same action.

Comparing Providers Without Checking Definitions

Providers may disagree over the intended receiver, the point at which a pass is considered complete, aerial duels, deflections, goalkeeper distribution and whether the receiver must establish control. Their xPass values should not be combined as though they came from one measurement system.

Ignoring Calibration

A useful probability model should be calibrated as well as accurate. Among passes assigned approximately 0.70, close to 70% should be completed over a sufficiently large and representative sample. A model can classify many passes correctly while still producing unreliable probabilities.

Using Information from After the Decision

Ball-flight or post-pass features may help explain what happened retrospectively. They can also leak information that was unavailable when the passer chose and executed the action. Analysts must align the feature timing with the intended use.

Assuming Above Expected Equals Individual Skill

The passer matters, but so do the receiver, teammate movement, opposition pressure and data labelling. Completion above expected is evidence for investigation rather than a pure, isolated measure of technique.

Rewarding Difficulty Without Considering Value

Attempting difficult passes is not automatically positive. A player can reduce expected completion by forcing low-value balls. Risk should be assessed alongside progression, possession value, game state and the available alternatives.

Overreacting to Small Samples

A few completed long passes can create a large short-term difference from expectation. Aggregates need enough attempts and should be broken down by role, pass type and tactical context before conclusions are drawn.

GoalIQAI Analytical Interpretation

xPass is most useful as one part of a wider passing framework. GoalIQAI separates three questions:

1. Difficulty: how likely was the selected pass to be completed?
2. Outcome and execution: was it completed, and how does actual performance compare with expectation over time?
3. Purpose and reward: what did the pass achieve, and was its risk justified by the tactical situation?

This prevents a common analytical mistake: treating the hardest completed pass as automatically the best pass. A low-probability action can display excellent technique but still be a poor decision. An easy pass can be tactically valuable if it finds a free teammate in a dangerous position.

The strongest analysis combines xPass with progression, possession value, pressure, receiver context, video and the alternatives available. No single passing statistic can fully separate technique, decision-making and team structure.

Key Takeaways

• xPass estimates the probability that an attempted pass will be completed.
• Expected completions are calculated by adding the probabilities assigned to individual passes.
• Completion above expected compares actual completion with the modelled difficulty of the attempts.
• xPass measures risk, not attacking value, progression or decision quality by itself.
• Model inputs, feature timing and completion definitions differ between providers.
• Receiver movement, tactics, opponents and sample size affect interpretation.
• The best passing analysis combines difficulty, execution and reward rather than relying on one headline metric.

Related Guides

• Expected assists explains how models value the probability that creative passes or resulting chances become assists.
• Browse the Football Betting & Analytics Knowledge Base for more guides to football metrics, modelling and evidence-based decision-making.

Stay Ahead of the Market

Subscribe to GoalIQAI for evidence-based football analysis, practical analytics guides and clear explanations of what performance metrics can—and cannot—tell us.