Piggy-backing off a thread from @stephen_burtch this morning, one thing we know about goaltending is that the only really repeatable portion of SV% is high danger SV%. Low and medium danger shots are largely just noise. https://puckplusplus.com/2015/03/23/predicting-save-percentage-dangers-zones-and-shot-volumes/
Frederik Andersen's overall SV% during his time in Toronto has probably been misleading. Outside of his first season, he's consistently ranked in the bottom 30-40% of goalies in HDSV% (minimum 500 minutes TOI).
Basically, Andersen has gotten good results in the portion of SV% that's largely random, but done poorly on the repeatable portion. His overall SV% last year may be closer to his true talent level than what Leafs fans saw in his first few seasons in Toronto.
I hope I'm wrong about this because the Leafs have a lot riding on Andersen providing good goaltending this season (or Campbell supplanting him), but if he struggles again this year I don't think it should be surprising.