Interesting stats posted by Travis Yost over at TSN regarding secondary assists. Puljujarvi was the second most unlucky player in the league when it came to secondary assists last year. Probably doesn't help when your most common linemate was stone cold Milan Lucic.
https://www.tsn.ca/the-noise-surrounding-secondary-assists-1.1134501
Sorry, read the article, what a brutal site, advertising right in the print columns resulting in the print expanding and contracting continually and literally impossible to stay reading on the right line. My lord who webmasters this kind of junk site.
Anyway this is largely garbage in garbage out. Simply postulating average numbers of pts awarded while on ice or secondary assists while on ice and saying that differentiation from that mean is simply " bad Luck" is an inherent misunderstanding of statistical methodology, science, statistical analysis or extraneous 3rd variables.
My head starts to hurt at all these bloggers that don't even remotely comprehend statistical methodology.
Well here goes. If a player substantially deviates from means it could be do to a number of things. Maybe even that they sucked. Lets draw this to an extent. Lets say X player was a complete passenger (I'm not saying Pulju was, just a hypothetical) by this metric he would be standing in a corner sucking his thumb all the time but on the basis of this metric substantially deviating from mean points accrual on goal events. This metric would determine this player just had "horrendous bad luck"
Sorry, but what a garbage stat.
Heres the real deal, some players, like a Sedin are constantly involved in plays and the probability that a play goes on without them being involved for 15 secs is remote. Puck movement flows through those players. They QB every play, they make plays on every shift. Other players are not very involved in plays typically, and thus the stats. its really not a lot more complicated than that. No, it isn't bad luck. Conversely if a player happens to play with multiple other puck movers in a unit and including a quality pmd their pts accrual on goal events will be diminished by QOT factor on the ice.
Finally statistical noise, i.e. unexpected data is indicative of data not represented by a theorem, not being explained by a hypothesis, not being nailed down. Too much noise in data through repeated experiments and one deduction could be that the parameters of the study are not close to detecting signal. It would often be a stage where research study is abandoned and wherein the data failed to support a hypothesis. How about that..?