All methods
Forensic xG
Validated modelTested on data the model never fit, with the accuracy published.
What it means
The chance a shot becomes a goal, judged on where and how it was taken.
How it's computed
Logistic regression over distance, angle, rebound, power play and playoff state.
What's validated - and what's not
Scored on a true 20% holdout the model never trained on; accuracy and calibration published on the Methods page. It knows location and state, not shooter talent or pre-shot movement.
Coverage
2010-11 to present (reliable shot coordinates).
Provenance
Foundational: Alan Ryder, “Shot Quality” (Hockey Analytics, 2004).
Reproduced on our 2010-11→present data (1.26M shots) and extended with angle, rush and rebound features; validated on a 20% holdout.
Our research ledger - every source we tested, adopted, or rejected →Tiers are a promise, not a style. Reconciled is checked against the official NHL record; Validated is tested on data the model never fit; Experimental is a derived construct without out-of-sample validation - read it as informed opinion. When in doubt we take the lower tier.