What Is Post-Shot xG, and What Does It Reveal?

Andy

Expected Goals

August 26, 2026

26 Views

After Contact

The chance creates the possibility; the finish decides what the goalkeeper must stop.

A low cutback reaches a striker near the penalty spot. One attempt is struck firmly but straight at the goalkeeper; another is guided inside the far post. Before contact, the opportunities are almost identical, so ordinary expected goals (xG) may value them equally. The scoreline records only that one went in.

Top Football Betting Bonuses for September 2026

BetOnline

$250 + 100 Free Spins

18+. Full T&C's apply. View full terms

Claim Offer

BetUS

200% up to $1,000

18+. Full T&C's apply. View full terms

Claim Offer

MyBookie

$500 + $25 Casino Chip

18+. Full terms apply. View full terms

Claim Offer

BetNow

$2,500

18+. Full terms apply. View full terms

Claim Offer

Post-shot xG adds the missing layer. For an on-target effort, it estimates the chance of scoring after placement and trajectory are known. The central shot may receive a low value despite coming from a strong position, while the corner-bound finish may rate highly. Comparing pre-shot xG with post-shot xG therefore helps separate the quality of the opportunity from the quality of the execution—and can also show when a goalkeeper prevented an especially difficult shot.

Expected goals (xG) asks how likely a chance was to become a goal before the ball was struck. It describes the quality of the opportunity: a central shot from six yards will usually carry more xG than an attempt from a tight angle outside the box.

Typical xG inputs include:

• Shot location and angle
• Distance from goal
• Body part used
• Type of assist or preceding action
• Open play, set piece, or penalty
• Defensive pressure, where the data provider records it

Post-shot expected goals (PSxG) asks how likely the resulting on-target shot was to beat the goalkeeper. Its inputs commonly include where the ball was heading within the goal, along with shot speed, height, trajectory, and sometimes the goalkeeper’s position. Exact features vary between providers.

That last distinction matters: PSxG is generally calculated only for shots on target, while xG can cover shots that miss or are blocked. Provider definitions can differ, so comparisons are safest within the same dataset.

In short, xG helps judge the chance created; PSxG helps judge the danger produced after the finish. The gap between them can offer clues about shot execution, but a single attempt says little about lasting finishing ability.

Consider three attempts from the same team. The figures below are illustrative, but they show how each layer adds a different piece of information.

The first shot begins as an excellent opportunity. Its central placement makes it less threatening after contact, yet it still goes in—perhaps through the goalkeeper’s legs or after a mistake. The goal does not prove that the finish was outstanding; it records only the final result.

The second attempt comes from a less promising position, so its xG is modest. Once struck accurately toward the corner, however, it becomes a difficult save and receives a much higher PSxG value. A goalkeeper can still stop it: high PSxG means dangerous, not inevitable.

The third effort contributes to xG because a shot was attempted, but it receives no PSxG in models restricted to shots on target. That distinction is central to understanding why xG and the scoreline can disagree.

Across these attempts, the team scores once, records 0.80 xG, and produces 0.95 PSxG from its two on-target shots. The scoreline says one chance was converted; xG says the team created one major opening plus two lesser ones; PSxG says the on-target execution was dangerous overall. None is the single “correct” story—they describe different stages of the same attacks.

Subtracting xG from PSxG offers a rough view of what happened at contact. When PSxG exceeds xG, the shot was made more dangerous through placement, trajectory, or clean ball striking. A lower PSxG suggests the opportunity lost value on its way toward goal—perhaps because the effort was too central, weak, or otherwise comfortable to save.

That difference can make a finish informative without making it definitive. Turning a 0.10 xG opening into a 0.45 PSxG shot indicates excellent execution on that attempt; it does not prove the player is an elite finisher. One spectacular strike can dominate a small sample, while several mishits can distort it in the opposite direction.

Context before conclusions

Longer-term gaps deserve attention, but several checks are needed when evaluating sustained xG overperformance:

• Sample size: hundreds of shots carry more weight than a short scoring run.
• Role and selection: poachers, set-piece takers, and frequent long-range shooters face different shot profiles.
• Repeatable technique: composure, preferred-foot use, and an ability to find corners may persist more reliably than occasional wonder goals.
• Model design: providers use different inputs and may value placement, power, headers, or deflections differently.

The gap is therefore best treated as evidence of execution, not a standalone finishing grade. Video, shot maps, and multiple seasons can help distinguish a stable skill from favorable variance.

For a goalkeeper, PSxG faced is the sum of the post-shot values of all eligible shots on target faced. Comparing that total with goals conceded gives a simple estimate of shot-stopping performance:

Goals prevented = PSxG faced − goals conceded

If a keeper faces 12.4 PSxG and concedes 10 goals, the result is +2.4 goals prevented. Conceding fewer goals than the model expects suggests above-average saving; conceding more produces a negative figure. Some sources reverse the subtraction or report the number per 90 minutes, so the label and sign convention should always be checked.

This comparison improves on save percentage because it recognizes that shots are not equally difficult. Stopping a close-range effort headed for the corner receives more credit than collecting a weak attempt near the center of goal.

Still, goals prevented is an estimate, not a complete verdict. Results can be influenced by small samples, model differences, defensive pressure, screens, deflections, rebound sequences, and whether the goalkeeper’s starting position made the eventual save easier or harder. A few unusually difficult shots can move the total sharply.

PSxG also evaluates only the shots that reach its model. It does not adequately capture:

• claiming crosses and controlling the penalty area;
• sweeping behind the defensive line;
• distribution and decision-making in possession;
• communication and defensive organization;
• actions that prevent a shot from occurring;
• rebound control that changes the next phase.

It is therefore strongest as a shot-stopping indicator, especially over substantial samples, rather than an all-purpose goalkeeper ranking.