Apparently it's systemic through the hockey world.
Travis Yost’s
recent article on the difficulty of evaluating goaltenders sings that familiar refrain. As usual, his research and analysis is illuminating. General Managers don’t know how to evaluate and pay goalies. Save percentage, in all its forms, is poor at predicting future performance. Average and replacement-level goaltenders are almost statistically indistinguishable, and aside from a tiny elite and small batch of failures, everyone is in this range.
Nonetheless, the evolving narrative of goaltending madness influences, and is influenced by, the perception that goaltending is an unknowable mystery. The position is inherently complex, to be sure, but the hockey community has traditionally preferred to dismiss that complexity with clichéd folk-psychology instead of learning it.
Yost’s article points out the current limitations of predicting goaltenders through save percentage. Strides are being made with descriptive and comparative measures, like my own work using
danger zones and
adjusted even strength save percentage, as well as
Nick Mercadante’s voodoo-evoking version of the latter. These post-hoc approaches provide usefully concrete descriptions, but do little to statistically predict a given goaltender’s future performance.
ingoalmag.com