I don't think you even understand the concept of small sample size and one random stat and as it applies here with actual context.
First the sample size is so small it doesn't even allow Murray to face every team even a single time. Everybody understands the smaller the sample size the less trustworthy the results.
Murray played only 20 games last year (< 1/4 a season) in those 20 games he let in 60 goals on the 640 shots he faced...
Based on someone subjectively carving up Ozone and Dzone and assigning a fictions % based on where a shot is coming from decides if it should be a save or goal.
So it was determined that Murray should have allowed 63.16 goals on the season based on tracking shot location instead of 60 actual which gives him a whooping magnificent 3.2 saves above expected.
So based on a mere 20 game sample that means Murray on average made 1 magnificent unexpected marvelous save every 6-7 games from a HDSC area.
Its really those 3 total Saves above expected you're hanging your hat on. Murray wins and everything else doesn't matter?
If Murray plays in only 20 games next year and makes only 3 saves above expected the Leafs are in a world of trouble, if he repeats his 5 wins, GAA and SV%, and everyone else needs to play the other 62 games.
Campbell played in 49 games (> 1/2 a season) faced 1430 shots allowing 123 GA and based on shot location should have given up only 120.66. So he gave up a mere +2-3 more goals than expected on a volume of over double the shots Murray faced, in over double the games Murray played.
Based on Murray's .906 sv% had he faced the same volume of shots 1430 (instead of 640) he would have made 1296 saves and allowed 135 GA compared to Campbell at 123 GA.. His +3 goals saved above expected every 20 games would not have saved him.
The NHL does track actual verifiable stats that are not subjective like Shots. Saves, SV% and GAA and WIns and SO. There is no subjectivity involved if the goalie either makes a save or lets a goal in. In all the real stats Campbell >> Murray. In the pretend stats based on shot location and shoulda woulda coulda saved it, on limited sample size Murray wins, but its so negligible its almost irrelevant.
I'm not the one having trouble with samples, or utilizing this stat properly, or applying proper context.
First off, it's the smaller the sample size, the more
variable the results. That doesn't mean the results you have in that sample are wrong. You have to be careful about extrapolating outward too much from that, but that's a concern for goaltending metrics in general. It's certainly not okay to look at last season's results for these players, and then randomly add on a sample from 2 years ago for only one player because Murray having the better results didn't fit your argument.
Second, GSAx is a "real stat". It is not "fictions" (I assume you meant fictitious) or "pretend". It is in fact much more accurate than what you call "actual verifiable stats", and you seemed to have no problem selectively using it yourself in your post and throughout the summer. Some of those "actual verifiable stats" you named, like shots and wins, aren't even goalie stats. In fact, SV% is the only one you named that's really all that relevant to goaltender performance at all, and that doesn't factor in shot quality or the massive contrast in defensive performance in front that we have in this situation. Expected and shot quality metrics are based on extensively supported and verified methodology, and are often much more complex than what you're suggesting.
Third, these calculations you tried to do randomly mixing together SV% and GSAx to argue that Murray would be at a negative with the same sample is just... so wrong. The difference between Murray's results and Campbell's results last year were about 5 goals, not 3 saves, and while we probably shouldn't be extrapolating results to bigger samples, if we were extrapolating Murray's performance to Campbell's sample like you're trying to do, Murray would have let in over 10 goals less.
Does all of this mean that Murray is definitely the better goalie and will do better than Campbell next year? No, of course not, but it also doesn't mean that Campbell is the better goalie and will do better than Murray next year, and if we're going to discuss their recent sample, we should do so accurately and appropriately. In truth, both Murray and Campbell are in a similar tier of goalies where, like most goalies in the league, they could put up positive or negative results next year.
At this point, we can only really wait and see, but this idea that Campbell (5 x 5m, signed until 35, limited history, coming off negative result sample) is somehow a much better option than Murray (2 x 4.7m, signed until 30, history of high end play/two cups/big moment performance, coming off positive result sample) does not hold up.