Post-shot expected goals
PSxG°An estimate from a model we built. Useful, but not a fact — every modeled stat's page says where it's weak.
Definition
Where xG asks how good the chance was, PSxG asks how good the shot was — it conditions on the shot's final trajectory and placement. Only shots on target have a PSxG. Compared against goals allowed, it is the standard shot-stopping measure.
How it is computed
A second model, fitted on on-target shots, reading placement, pace, and the pre-shot context.
How much is enough
Shot-stopping is the slowest-stabilizing metric on this site. A single season is not enough; treat differences under about 4 goals as fog.
Where it is weak
League-wide PSxG−GA should sum near zero. In 2025 it does not (−41.95), which is a calibration signal we are still chasing — read individual keeper figures with that in mind.
What it refuses to measure
PSxG says nothing about distribution, claiming crosses, or sweeping. It is shot-stopping only.
What would prove this wrong
League-wide PSxG minus goals allowed should sum near zero each season. It currently does not for 2025 (-41.95, a known open calibration issue); if the gap persists at that size after the next model revision, per-keeper figures should not be published.