It absolutely does change the argument. Are you reading what I'm writing because I've talked about all of these things several times. Luck implies the player is not at fault. Variance can be anything, including the way the player is playing and so forth. If we don't know what it is then we don't know what it is. Calling it one thing when we have no real proof isn't scientific or statistically sound.
I've mentioned several times that I'm not interested in what it's called, I'm interested in what it means and the implications regarding Kuznetsov.
He had a very poor run in terms of PDO. This is a fact. Historically, a player's PDO does not stay that low. This is a fact. And assuming shot attempt differential stays the same (or increases, in Kuznetsov's case), a rebound in PDO means an increase in goal differential. This is a fact.
So in order for me to worry about Kuznetsov, you have to make the argument that either Kuznetsov's PDO will remain historically low, or you have to argue that Kuznetsov will suddenly transform into a poor possession player despite his possession metrics actually increasing as the season went on (and into the playoffs, in fact). Even if the REASONS revolve around intangibles or unquantifiable things, these would as a consequence be reflected in the statistics.
There's a reason PDO stays around 100...it's crap. It's generally a statistical washout of basic O and D stats that only really shows blips that again you ASSUME are due to "luck". The entire premise is flawed.
PDO isn't crap, it's a number. Indeed, it has very little predictive power. The PDO statistic in one game, in general, has nothing to do with the next game. It's not flawed, only its interpretation can be flawed.
I interpret it as describing puck-luck. "Puck-luck" doesn't mean it is out of a player's control, but it does mean that there is expected variance with any play and sometimes the result is good and sometimes it is bad. You can pick your spot with a shot, do everything exactly the same as you always do in terms of mechanics, etc., and have two different results. Slight variations in the ice surface could cause the puck to go into the corner of the net or off the post and wide. Is this under a player's control? Well, technically yes. But in practice there are far too many variables to account for as a player and sometimes these variables work in your favor (Kuznetsov's goal in the playoffs) and sometimes they don't (hitting posts, pucks bouncing on an uneven ice surface).
These variables that are unreasonable to control even out over time. For every time something like that hurts you (or helps), statistically speaking these unreasonable to control factors affect the opposition at extremely similar rates over time.
The value of PDO (and its individual components) is using it to help explain past results. Nick Foligno had a career year in 2014-15, and also an enormous PDO. A smart front office would have seen that his stats were fueled by high amounts of circumstances that were unreasonable for him to control. Instead they gave him a long term contract that they immediately regretted.
Similarly, Alex Ovechkin's -35 season in 2013-14 was widely panned by mainstream media members, but he also had a 96.5 PDO for the season (further proof that poor PDOs can last an entire season, but they still don't last indefinitely). Was he really costing the Capitals about half a goal a game at even strength? Hell no. There were many things that were unreasonable for him to control that worked against him for most of the season.
I'm just a random guy on the Internet. But by saying PDO is worthless basically ***** on the opinion of well respected members of the analytics community who find it (and other measures of "puck-luck") valuable.
Where did I say the "eye test" cannot fail? Don't put words in my mouth based on your own biases about anyone who disagrees with your approach to statistics. You're just making things up and cherrypicking what you want to believe based on your biases toward the infallibility of stats (despite disclaimers) and uselessness of "intangibles" and/or defensive play that can't be measuresd. This is what you say here, not something I need to make up.
I value stats but not stat abuse. Even the professionals and creators of these stats warn against the kind of over-reliance you're falling into, so I'm not just dumping on your contortions for no reason.
In the past, your arguments have boiled down to almost exclusively subjective eye test arguments (specifically in defense of Wilson and Orpik). When these eye test arguments are directly contradicted by an analytic argument, it's always the analytic argument that is wrong. This inane PDO argument is a perfect example.
And when you argue that PDO is "crap" despite a long list of prominent analytics members (and almost certainly many NHL front office members) showing and applying its usefulness, it's really difficult to take you seriously when you say you "value stats".
