Expected Goals on Target and Post-Shot xG Explained

Expected Goals on Target and post-shot expected goals estimate how likely a shot is to become a goal after information about its execution is available. Unlike conventional expected goals, which evaluates the chance at the moment before the shot is struck, a post-shot model can consider where the ball travels within the goalmouth and, depending on the provider, its trajectory, speed and the goalkeeper’s position.

This makes the metric useful for analysing shot placement and goalkeeper shot stopping. It does not provide a pure measure of finishing talent or goalkeeper ability. Model definitions vary, off-target shots require careful treatment and short-term differences can be driven by variance.

What Is Expected Goals on Target?

Expected Goals on Target, commonly abbreviated to xGOT, estimates the probability that an on-target shot becomes a goal given the information available after the ball has been struck.

Opta’s Expected Goals on Target model combines the underlying quality of the original chance with information about the shot’s goalmouth location. A shot directed towards a difficult corner normally receives more xGOT than one travelling centrally from the same initial position.

Post-shot expected goals, commonly abbreviated to PSxG or post-shot xG, describes the same broad family of models. The terminology is not completely standardised. Providers can differ in:

• the shots included;
• how off-target and blocked attempts are treated;
• whether shot speed or trajectory is available;
• whether goalkeeper position is included;
• the historical competitions used to train the model;
• whether penalties have a separate model;
• the precise moment at which the shot is evaluated.

Values from different providers should not be combined or compared as if they came from one common model.

What Is the Difference Between xG and Post-Shot xG?

Conventional expected goals measures chance quality before the result of the shot is known. Typical inputs include shot location, angle, body part, assist type and the surrounding defensive situation.

Post-shot xG begins later in the sequence. It asks how dangerous the attempt became once the player had struck the ball.

A high-quality chance can therefore produce a poor post-shot outcome if the attempt misses the target or is directed at the goalkeeper. A low-xG chance can produce a high post-shot value if the player sends a powerful or well-placed attempt towards a difficult part of the goal.

Shot-Sequence Example: xG Compared With Post-Shot xG

Consider three illustrative shots taken from comparable positions. Each chance receives a pre-shot xG value of 0.15 because the chance location, angle and surrounding context are assumed to be similar.

The post-shot values below are hypothetical model outputs used to explain the calculation. They are not taken from a particular match or data provider.

Across the sequence, the player generated 0.45 xG before shot execution and 0.62 post-shot xG afterwards. On this illustrative all-shot basis:

Post-shot improvement = total PSxG − total xG

Post-shot improvement = 0.62 − 0.45 = +0.17

The positive difference indicates that the final placement of the three attempts increased their aggregate probability of becoming goals relative to the original chances.

This does not mean the player “should” have scored exactly 0.62 goals in that sequence. The value is a model estimate across comparable historical attempts. One realised goal is compatible with many underlying probability combinations.

Why the off-target shot matters

The first shot retains its 0.15 xG because a meaningful chance was created. Its post-shot value is zero in the illustrative presentation because the attempt missed the target.

If a provider publishes xGOT only for on-target shots, the off-target attempt may instead be absent from the xGOT dataset. Analysts must then ensure that they compare compatible totals. Comparing xGOT from two on-target shots with xG from only those same two shots answers a different question from comparing all-shot xG with an all-shot post-shot representation.

How Post-Shot xG Can Analyse Finishing

Pre-shot xG helps separate chance volume and quality from the goals scored. Post-shot xG adds another stage to the sequence:

1. Chance creation: What was the probability before the shot was executed?
2. Shot execution: Did the player increase or reduce the danger through placement and other measurable shot characteristics?
3. Final outcome: Did the shot become a goal after interaction with the goalkeeper and normal variance?

A player whose post-shot xG persistently exceeds pre-shot xG may be directing attempts towards more difficult goalmouth locations than the average shooter facing similar chances.

A player whose post-shot xG is below pre-shot xG may be missing the target frequently or producing relatively saveable on-target attempts. This could reflect weak execution, difficult shot selection, defensive pressure, injury, a temporary loss of form or ordinary variance.

The comparison should not be reduced to a universal rule that:

• positive PSxG minus xG proves elite finishing;
• negative PSxG minus xG proves poor finishing;
• goals above PSxG belong entirely to the shooter;
• one season establishes a permanent player skill.

The GoalIQAI guide to finishing overperformance explains why goals above expected can contain both repeatable skill and non-repeatable variance.

Separating placement from scoring outcomes

Several comparisons can be useful, but each answers a different question:

Using the three comparisons together can provide a more informative diagnosis than goals minus xG alone.

How Post-Shot xG Evaluates Goalkeepers

Post-shot models can estimate how many goals an average goalkeeper would be expected to concede from the on-target shots faced.

A common calculation is:

Goals prevented = post-shot xG faced − goals conceded

If a goalkeeper faces on-target attempts with a combined post-shot value of 14.0 and concedes 11 goals:

Goals prevented = 14.0 − 11 = +3.0

The goalkeeper conceded three fewer goals than the model expected from that set of shots. If the goalkeeper conceded 16, the value would be −2.0.

This is more informative than save percentage because it gives more weight to difficult shots. Saving five weak central attempts is different from saving five attempts directed towards difficult areas of the goal.

It still does not isolate goalkeeper ability perfectly. Results can be affected by:

• deflections after the recorded shot point;
• blocked or restricted sightlines;
• goalkeeper starting position;
• shot speed or dip omitted by a provider;
• defenders interfering with movement;
• rebound situations;
• penalties and own goals;
• data-collection and model error.

A complete goalkeeper assessment should also consider crosses, sweeping, distribution and tactical role. Post-shot xG evaluates shot stopping, not the entire position.

Goals prevented totals versus rates

Total goals prevented measures estimated contribution across the complete workload. A goalkeeper who faces more shots has more opportunities to accumulate a large total.

A rate can standardise performance by shots faced, post-shot xG faced, minutes or another denominator. This may improve comparisons between different workloads, but rates become unstable in small samples.

Before comparing two published goalkeeper figures, check:

• whether penalties are included;
• whether own goals are removed;
• the definition of shots on target;
• the denominator used for the rate;
• the competitions and seasons covered;
• whether the figures come from the same model version.

How xGOT Appears on Shot Maps

Traditional shot maps normally display where attempts were taken and the pre-shot xG assigned to each chance. A post-shot visual may instead show the goalmouth destination, allowing the reader to compare placement as well as initial shot location.

The two visualisations answer different questions:

• Pitch-based shot map: Where were chances generated and how difficult were they before execution?
• Goalmouth map: Where did on-target attempts travel and how difficult were they to save?

A team can create a strong pitch-based shot map but produce little post-shot danger by missing the target or directing attempts centrally. Another team can create fewer chances but execute its on-target shots unusually well.

Neither visual should be treated as complete without match context. Game state, defensive pressure and the number of observations can materially change interpretation.

Why Provider Models Produce Different Values

xGOT and PSxG are model outputs rather than directly observed facts. Two providers can assign different values to the same attempt while both apply reasonable methods.

Differences can arise from:

• Training data: competitions, seasons and sample composition;
• Feature availability: placement, height, velocity, trajectory and spin;
• Goalkeeper information: position, movement and visible goal area;
• Defender information: pressure, screens and possible blocks;
• Shot categories: headers, volleys, penalties and one-on-ones;
• Outcome definitions: goals, saves, blocks, own goals and off-target attempts;
• Model architecture: statistical method, calibration and model updates.

Opta’s enhanced xGOT model states that it uses shot trajectory, end location and other contextual factors, with separate models for men’s football, women’s football and penalties. StatsBomb has described post-shot models that use shot trajectory, speed and other characteristics available after the strike.

The name of a metric is therefore not enough. Analysts should consult the provider methodology before interpreting small differences.

What Post-Shot xG Cannot Tell You

It cannot prove finishing talent from a short run

Players take relatively few shots compared with the number of observations available in many other sports. A small number of exceptional or poor attempts can dominate short-term totals.

The GoalIQAI guide to sample size in football analytics explains why the number, stability and comparability of observations all matter.

It cannot completely separate shooter and goalkeeper

The realised goal outcome is shared between shot execution, goalkeeper response and variables around the event. Even a detailed post-shot model may not record everything affecting the save probability.

It does not measure chance creation

A player cannot receive post-shot credit for a shot that was never created. Pre-shot xG remains necessary for evaluating the quality and volume of opportunities.

It does not evaluate the goalkeeper’s complete role

Shot stopping is only one part of goalkeeping. Cross claiming, sweeping, passing, starting position and communication can change team performance without appearing in a goals-prevented figure.

It does not make every difference predictive

A metric can describe past execution accurately without forecasting that the same player or goalkeeper will repeat the result. Predictive use requires out-of-sample evidence and appropriate regression towards a longer-term baseline.

Common xGOT and Post-Shot xG Mistakes

Calling xG and xGOT competing versions of the same answer

They describe different stages. xG measures the original chance; xGOT or PSxG measures the danger created after the strike.

Ignoring off-target attempts

Looking only at on-target shots can make a player appear efficient while hiding frequent misses. Pre-shot and post-shot totals must use clearly defined shot populations.

Comparing different providers directly

Provider values are not interchangeable. A difference may reflect methodology rather than disagreement about the player.

Treating goals prevented as saves made

Goals prevented is quality-adjusted. A high save count can accompany a modest goals-prevented result if most attempts were easy to stop.

Using one match to judge a goalkeeper

An exceptional post-shot total in one fixture describes the difficulty of that specific workload. It does not establish a stable ability level.

Attributing team overperformance to one component

When a team concedes fewer goals than expected, finishing faced, goalkeeper performance, defensive context and variance can all contribute. The GoalIQAI framework for identifying team overperformance considers these factors alongside expected points, schedule strength and close-match outcomes.

How GoalIQAI Interprets Post-Shot Data

GoalIQAI uses a staged interpretation:

1. Start with pre-shot xG. Establish the volume and quality of chances created or conceded.
2. Check shot execution. Compare the post-shot danger with the original chance quality using compatible shot populations.
3. Separate the final outcome. Compare goals with post-shot expectations without assigning the entire difference to one player.
4. Verify definitions. Identify the provider, model version, included shots and treatment of penalties.
5. Examine the sample. Use multiple periods and comparable situations where possible.
6. Review the video. Check placement, power, goalkeeper positioning, sightlines and deflections.
7. Express uncertainty. Treat the metric as evidence rather than proof of permanent skill.

Using Post-Shot xG in Predictions

Post-shot xG can help explain why recent goals differ from pre-shot chance quality. Its forecasting use should remain cautious.

A short run of strong post-shot execution should not automatically cause a large increase in a player’s future scoring probability. Similarly, a goalkeeper’s recent goals-prevented figure should not be projected forward unchanged.

Before using the metric in a prediction, ask:

• Is the difference supported across a meaningful sample?
• Does it persist under the same provider and methodology?
• Have shot selection, tactical role or opposition quality changed?
• Is the goalkeeper facing a comparable type of workload?
• Does video support the statistical interpretation?
• Has the market already incorporated the relevant performance?

Post-shot data is most useful as supporting evidence within a broader forecast. It should not independently determine goal, player or clean-sheet probabilities.

Key Takeaways

• Pre-shot xG measures the quality of a chance before the shot’s execution is known.
• xGOT and post-shot xG estimate the danger of an attempt after measurable execution information becomes available.
• Two shots with the same xG can receive very different post-shot values because of placement, trajectory or other provider inputs.
• Post-shot xG minus pre-shot xG can describe how execution changed aggregate shot danger, but it does not prove repeatable finishing skill.
• Post-shot xG faced minus goals conceded produces a common goals-prevented measure for goalkeepers.
• Provider definitions, shot populations and model versions must be checked before comparing values.
• Post-shot data should support, rather than replace, wider finishing, goalkeeper and team analysis.

Related Guides

• What Is Expected Goals?
• Shot Maps Explained
• Finishing Overperformance Explained
• Football Betting and Analytics Knowledge Base

Stay Ahead of the Market

Subscribe to GoalIQAI for evidence-based football predictions, betting-market analysis and practical football analytics guides.