Quoting: BeterChiarelli
If we look to hockey-reference, over the span of January 13th to February 6th while Smith was hurt, Koskinen was an .889sv% goaltender. This was bad yes but also saw him play 12 games in 24 days, a workload he's not known to handle. Once Smith returned, over the span of February 9th to May 15th (13 games), Koskinen sported a 0.929sv%. Had it not been for Holland's inability to keep a goaltender to play with Koskinen, it's increasingly likely that both of Edmonton's goaltenders would have been in the Vezina conversation.
If we turn to naturalstattrick, over the past three seasons Koskinen has maintained a 0.916sv% (re: league average) at even-strength play. For goaltenders over this span that have played 1900min or more (such that there are 62 goalies to look at, two per team over this window) Koskinen ranks 43rd with that same save percentage ahead of names like Demko, Georgiev, Korpisalo, and Hart. If we assume that there are three tiers of goaltender - starter, tandem, and backup - with natural jenks every 20 some-odd goaltenders, Koskinen is either a poor man's tandem goaltender or the peak of backup netminders. This is the exact value I want to extract out of him.
I recognize that he's paid like a starter and cannot perform in that role. That's the point of the retention. He has ONE year left on his deal and I've shown that the Jets can make the cap work in a trade for Koskinen. What hasn't happened in this conversation is any rational sort of reply from any of Koskinen's detractors about how or why he "sucks". Koskinen is at minimum a league average goaltender who's only genuine downside is his cap hit.
It's not bias. It's not trolling. You simply don't have a point to make and resort to name-calling.
Save percentage is a common indicator used to evaluate a goaltender, but it is based on shot volume and quite heavily influenced by the quality of shots allowed by the team in front of the goaltender. I know it seems ridiculous to suggest that Edmonton's defense may bolster their goaltender's stats, but oddly enough the numbers seem to suggest the Oiler d-men did okay. Goals saved above expected (GSAx), while not perfect, is probably a better measure of a goaltender's performance because it factors quality of scoring chances and not simply shot volume.
FWIW last season Mike Smith had the fourth best GSAx, at 14.4, and that was born out by his outstanding play. On the other hand, among the 28 goaltenders with between 20 and 30 starts, Koskinen ranked 18th. He posted a GSAx of -10.2, which is right between Braden Holtby (-11.4) and Pekka Rinne (-9.7). That's not great company to be keeping when you are paid a starter's wage, but the number is roughly middle of the pack. Koskinen's GSAx would be much better if not for his struggles at the beginning of the season, his penchant for allowing first-shot goals from time to time, and his miserable performance in his final couple of starts. However, most of those poor performances are on him, and all goaltenders would have better stats if we simply eliminated those games in which they struggle.
Of interest to the the discussion, in 2020/21 Koskinen and Khuldobin have remarkably similar stats. Khudobin played six more games than Koskinen and allowed the same number of goals. The difference between the rest of their numbers is minimal, and could easily have ended in Koskinen's favour if he didn't struggle in his first dozen starts. Then again, they might have been better despite that start, had Koskinen not suffered the performance gaffs mentioned earlier.
The numbers suggest that in comparison to his peers, Koskinen is a league average backup goaltender, capable of starting between 25 and 30 games, and likely to cost his team 3-4 of those games because of his inconsistency. The difficulty in moving his contract stems from a his salary, and the relative availability of comparable goaltenders. On that note, the advanced statistics also suggest that the Oilers might not improve their lot by pursuing Khudobin. While he might be able to start a few more games than Koskinen, and he may not suffer the same wild swings in performance, his overall contribution is likely to be very similar.