Statistically speaking this is as misleading as reversion to mean. Because extraneous variables can exist and explain data aberration. Just that they are not presently accounted for by the metrics, so that they call some outlier data as luck, or PDO. Its probably something that should be stated and emphasized less. Data tails are just that. Just data that don't occur around the mean. Nothing like luck needs to be described to that except if one wants to engage in the falsehood of reversion to mean.
Karlsson above is an interesting casestudy. The stats would have him being lucky. What was seen on ice was a player making elite plays like crazy for a period of time and really much of the season and playoffs. Such spans are not even limited sample, they are considerable sample. he wasn't lucky, he was redhot for an extended time. Will he ever replicate this season? no. But the reason for his season was not luck, it was being really in the zone. This happens, and it happens in all sports and in life. There are times for all of us in which we are at our absolute best. Its not a completely understood phenomenon. But it exists. as does "in the zone".
So that an argument could be made that Darnell Nurse was simply in an outlier zone the first half of last season. I don't think so, I think with him, a young D stud, we will see some replication of his stellar play. But Darnell was not lucky to have a great first half last season. He was simply very good.