Observation 1: Why someone might think Oleksiak over Johns might be a good idea. Quoting Sean Shapiro, "Now it appears Oleksiak may have passed him on that list."
Hey look, Oleksiak has been a plus player much of this year, and generally better than Johns in that regard, especially since the beginning of February, when Johns has been on the ice for more than 4 times as many goals against as goals for.
Observation 2: You should be careful relying on the eye test only.
Much of the year Oleksiak has had a higher PDO than Johns, whose PDO is very, very low in the past 6 weeks or so. The idea is that PDO is a proxy for luck, and will even out over time. Johns has been very unlucky, while Oleksiak’s luck seems to have worn out. Notice how closely their goal for percentage follows their PDO. You don’t want to throw away a good player because of a streak of bad luck.
Observation 3: Since goals are so rare, we need to look at other metrics that have far more sampling.
There are lots of shot attempts in a game, and because shot blocking (and getting shots through) is a skill, Fenwick is a good measurement. First, see why Johns was demoted in November? The thing is, he was right around Oleksiak. For the season, he has generally been better than Oleksiak except for a few short stretches (including a time when Oleksiak was injured). But notice how Johns’ period since Feb 1 has not been awful possession wise, and actually better than Oleksiak’s. Just a reminder not to dump a guy because of bad luck.
Observation 4: Did I mention that Johns has had some fairly bad luck lately?
xG is your stat that takes into account shot quality, and more, to answer "how many goals should have been scored, allowed, based on the shots that were taken/given up?" It takes into account shot location (shots in the crease are more likely to go in than shots form the blue line), the type of shot, the shooter (Tarasenko should score more than Ryan Reaves), etc… You can actually see Oleksiak’s improvement… to Johns’ level of play. The xGF% has been higher with Johns on the ice for just about the entire season. Notice that Johns has had a positive rel xGF% since Feb, despite being a minus 7. That’s more evidence of his "bad luck" lately.
Observation 5: Intangibles — handedness and penalty killing.
Next year, Honka and Klingberg will both be RHD on the roster. Klingberg doesn’t kill penalties. Honka does in the AHL, but allowing Johns to be taken will open a hole on the right side. We already have a hole in the left side with regards to a player who can make Klingberg great again, and that will absolutely have to be addressed in the offseason if they want significant improvement.
Oleksiak has allowed the same number of PK goals against in about 1/3 of the ice time as Johns. The stark difference is mostly due to luck (good for Johns, bad for Oleksiak), however, the worst PK in the league is worse with Oleksiak on the ice in just about every metric, and often quite a bit better with Johns on the ice.
My final thoughts: Johns, when struggling, is similar to Oleksiak when he has "figured it out." Johns is a far better penalty killer and his loss would result in the loss of yet another effective penalty killer. Given Oleksiak's history, I would suggest protecting Johns over Oleksiak.