Non-Penalty Expected Goals Explained

Non-penalty expected goals, usually shortened to npxG, is a player’s or team’s total expected goals after the xG assigned to penalty kicks has been removed. If a team records 1.78 xG including a penalty valued at 0.78, its npxG is 1.00.

The metric helps analysts compare chance creation and goalscoring involvement without allowing penalty-taking duties to dominate the result. It is narrower than total xG but should not be confused with open-play xG: shots from corners, free kicks and other non-penalty set pieces can still count. Readers new to the underlying model should begin with GoalIQAI’s general expected-goals guide.

What Is Non-Penalty Expected Goals?

Expected goals assigns each shot an estimated probability of becoming a goal. Non-penalty expected goals adds those modelled probabilities together while excluding penalty kicks.

The basic formula is:

npxG = total xG − penalty-kick xG

The calculation changes the set of shots being measured, not the underlying meaning of xG. A non-penalty shot valued at 0.20 xG still represents an estimated 20% scoring probability under the assumptions of that particular model.

npxG normally retains:

• shots from open play;
• direct free-kick attempts;
• shots following corners and indirect free kicks;
• other non-penalty attempts classified by the provider; and
• separate rebound attempts where the provider records them as eligible non-penalty shots.

It removes the penalty kick itself. The exact treatment of later phases and unusual events depends on the data provider, so definitions should be checked before figures are compared.

npxG vs xG: What Is the Difference?

The distinction matters because npxG is not another name for open-play xG. A header from a corner is not a penalty, so it can remain in npxG even though it did not come from open play. GoalIQAI’s guide to set-piece expected goals explains why provider classifications and second-phase rules can change those totals.

How to Calculate npxG

A reliable calculation requires three inputs:

1. Total xG: the summed model values of every eligible shot in the selected match, period or competition.
2. Penalty xG: the model value assigned to each penalty kick in the same dataset.
3. A consistent scope: the same provider, competitions, dates and model version for both figures.

Consider this illustrative team shot set:

Total xG is 1.78. Removing the illustrative penalty value gives:

1.78 − 0.78 = 1.00 npxG

The shot events are observed data, but their xG values are model estimates. Another provider could assign different probabilities to the same non-penalty attempts or use a different value for the penalty.

Why There Is No Universal Penalty xG Value

Penalties are unusually consistent shots, so providers often assign them a fixed pre-shot value based on historical conversion. That value should not be treated as a law of football.

Official Opta material illustrates the issue. One Opta Analyst methodology explainer uses 0.79 xG for a penalty, while another Stats Perform overview presents 0.76. Model populations, definitions and versions can differ.

The safe approach is to subtract the penalty value recorded by the dataset being analysed. Replacing it with a familiar number from another provider can create a false level of precision.

Labels also require care. In its radar documentation, StatsBomb describes the xG shown on certain radars as non-penalty expected goals. An analyst should therefore confirm whether a dashboard’s “xG” column already excludes penalties before subtracting anything.

A Worked npxG vs xG Player Comparison

Suppose two forwards produce the following rounded figures over the same number of minutes:

Player A leads on total xG because penalty-taking duties added approximately 3.1 expected goals. Player B leads on npxG and therefore generated the stronger non-penalty shot profile during this sample.

That does not automatically make Player B the better goalscorer. Player A’s penalty role has genuine match value, and taking penalties successfully is a relevant skill. The comparison answers a narrower question: who accumulated more expected goals from non-penalty attempts?

Why Analysts Use Non-Penalty xG

Comparing Players with Different Penalty Roles

A designated penalty taker receives opportunities that are unavailable to most teammates. npxG removes that role advantage when the objective is to compare movement, shot selection and involvement in chances generated during normal play and non-penalty set pieces.

Separating Attacking Process from Penalty Events

Penalties can materially change a match or season total despite occurring infrequently. Removing them makes it easier to inspect whether a team has repeatedly reached valuable shooting positions.

This does not mean penalties are pure luck. Teams that enter the penalty area frequently, create defensive emergencies and field difficult dribblers may be more likely to win them. npxG deliberately excludes the resulting kick; it does not prove that everything leading to the award was random.

Building Clearer Forecast Components

A model can estimate non-penalty scoring and penalty contribution separately. This helps prevent a small historical run of awards from being projected forward as though every past penalty will recur at the same rate.

How to Interpret npxG in Practice

A headline npxG total becomes more useful when it is decomposed into volume, average chance quality, playing time and tactical context.

• npxG total: the accumulated quality of non-penalty shots.
• npxG per 90: a playing-time rate calculated as npxG ÷ minutes × 90.
• npxG per shot: average modelled quality of each non-penalty attempt.
• Non-penalty shots: the volume of attempts contributing to the total.
• Share of team npxG: the proportion of a team’s non-penalty threat attributed to a player.
• npxGA: non-penalty expected goals conceded.
• npxG difference: non-penalty xG for minus non-penalty xG against.

Two forwards can record the same npxG in different ways. One may take many difficult shots, while another receives fewer but much clearer chances. Combining the metric with reading shot maps helps reveal where those attempts originated and whether one major chance dominates the total.

Role and opportunity also matter. A winger, target forward and substitute striker should not be compared as though they have identical responsibilities. Per-90 figures can improve comparability, but they do not remove selection effects, substitute game states or uncertainty around limited minutes.

What Non-Penalty xG Cannot Tell You

• It does not measure finishing by itself. npxG describes the estimated quality of chances, not how well the player converted them.
• It does not fully measure shot execution. Conventional pre-shot xG may not incorporate where the ball travelled after it was struck.
• It begins at the shot. Runs, passes, carries and decisions that create an opportunity may receive no npxG credit.
• It does not value penalty-winning contribution. A player can create the action that produces a penalty without receiving shot-based npxG.
• It does not forecast the next match automatically. Opposition, expected minutes, tactics, team news and game state can change future opportunities.
• It is not provider-independent. Shot values and event classifications can differ across models.

Common npxG Mistakes

Treating npxG as Open-Play xG

npxG removes penalties only. Non-penalty set pieces can remain unless the provider or analyst applies another filter.

Subtracting One Universal Penalty Value

Do not assume every dataset uses 0.76, 0.78, 0.79 or another familiar figure. Use the value and definition supplied by the relevant provider.

Mixing Total and Non-Penalty Measures

Comparing non-penalty goals with total xG, or total goals with npxG, combines different populations. Use non-penalty goals minus npxG when evaluating non-penalty finishing, then assess penalty outcomes separately.

Comparing Different Providers as Though the Models Are Identical

Two npxG figures can differ because of model inputs, data coverage, shot classifications or historical training samples. Small decimal differences do not establish that one source is wrong.

Trusting Per-90 Rates from Limited Minutes

A substitute can produce an impressive npxG-per-90 rate from a handful of late appearances. The rate may describe the sample accurately while remaining a weak estimate of future output. GoalIQAI’s sample-size guide explains why minutes, event counts and context must be considered together.

Removing Penalties from Every Analytical Question

npxG is useful for comparison, but total scoring probability still includes the possibility of a penalty. Ignoring a secure penalty taker’s role can understate that player’s real goalscoring outlook.

How GoalIQAI Uses npxG

GoalIQAI treats non-penalty xG as a decomposition tool rather than a superior replacement for total xG. A structured interpretation is:

1. Define whether the question concerns total scoring, non-penalty chance generation, open play or a specific set-piece phase.
2. Confirm the provider’s labels, penalty value and shot classifications.
3. Compare npxG with non-penalty shot volume, npxG per shot and the underlying shot locations.
4. Adjust for minutes, role, opposition, game state and material personnel changes.
5. Evaluate finishing and post-shot execution separately rather than expecting npxG to answer every question.
6. For forecasts, estimate penalty contribution as a separate probability-weighted component.

A simplified forecast might express that final component as:

Expected penalty goals = probability of a penalty being awarded × probability the player takes it × probability of scoring

Each input is uncertain. A model should not add the full xG value of one penalty to every match merely because a player is the normal taker.

The central interpretation is simple: total xG describes all eligible shot quality, while npxG isolates the part that did not come from penalty kicks. The most useful choice depends on the question being asked.

Key Takeaways

• Non-penalty expected goals is total xG with penalty-kick xG removed.
• npxG is not the same as open-play xG because non-penalty set-piece shots can remain.
• Penalty xG values and metric labels can differ between providers and model versions.
• npxG improves comparisons between players with different penalty-taking duties.
• Total npxG should be assessed alongside shots, npxG per shot, minutes, role and shot locations.
• Non-penalty goals should be compared with npxG when analysing non-penalty finishing.
• Removing penalties helps isolate one process but does not make penalty-winning or penalty-taking value irrelevant.
• npxG remains a model estimate and should not be used as a complete forecast on its own.

Related Guides

• Finishing Overperformance Explained
• Expected Goals on Target and Post-Shot xG Explained
• Football Betting & Analytics Knowledge Base

Stay Ahead of the Market

Subscribe to the GoalIQAI newsletter for evidence-based football analysis and practical guides to performance metrics, probability and better decision-making.