In case you are interested, here is an article that looks at the impact of the quality of defense on goalie metrics.
By: Albert Carreno
www.bruinsportsanalytics.com
The article shows that there is no definitive way to quantify this.
Thanks for the discussion... to the above, I'm not at all surprised. It's teeny tiny little details that make all the difference and it relates to why xGA still isn't perfect.
Example: whenever I go to a new team, or we bring on a new D, there is always a get-to-know-you phase. But invariably, there is one lesson I nearly always have to teach. And I almost never have to teach it twice because as soon as they hear it, guys get it.
There could be much more going on, but the sake of simplicity, imagine a one-on-one fwd vs D. Smart forwards will always try to use the screen to surprise the goalie. But even if I am "screened" as xGA might capture, there is a difference between "screened with forward having many outs" and "screened where options are taken away".
I used "outs" because it's exactly like poker. At the moment the shot comes my defender will either (A) keep his legs apart and lunge to get a stick on the puck if he can, or (B) close his legs together and do the same thing.
If he does (A) the shooter has at least 6 "outs" from left to right that I need to cover: (i) clean shot to left of left leg, (ii) deflected shot off inside of left leg, (iii) clean shot through D's five hole, (iv) deflected shot off inside of right leg, (v) deflected shot off outside of right leg and (vi) clean shot to right of right leg. All hell can break loose and usually does.
And that's not including stick deflections, it's a lot for a goalie to cover AND it is harder to see around a guy with his hips wide and legs spread, so to get a look at the puck I'm REALLY cheating to look through or around his five-hole.
However, if (B) he puts his legs together at that moment the number of outs drops to, at most, three (i), (v) and (vi). And if he has his legs together and jams his breaks or even coasts toward the shooter, then I can probably completely disregard (i) and square up for the other two since the angle to hit that shot (i) is disappearing.
What do you think that does for my save percentage? And both of these would very likely be characterized as a screened shot from a high danger location. xGA would be no different but you can bet ACTUAL GA would differ significantly.
THIS is why they say goaltending is a mental game. You are trying to calculate those odds in real time on every play.
(And a sidebar here: when (wise) announcers talk about a goalie having an active head or active tracking of the puck, they mean that up until the moment scenario A or B is happening, the goalie is looking on all sides of that hulking defenseman to decide which side has the best odds)