Do those stats explain why Scheifele and Laine shoot high percentages without saying "luck"
That's a genuine question
As an aside conversation, since the answer yes or no does not refute the previous question:
Sort of...
The reason for "sort of" is because of you asking "why" and also saying "high" and "luck".
"High" is an undefined and relative term.
Mark Scheifele over his four seasons shot 13%, 9%, 15%, and 20%. Last year could be considered "high" relative to league average (9ish), but also "high" relative to his career shooting percentage of 14%.
"Luck" is a loose term too.
Over 100 coin tosses, you will likely be fairly close to 50/50. If you broke down each set of 10 tosses, though, many would significantly differ from 50/50.
In a game where "heads" is a positive result, getting more or fewer than 50/50 could be seen as good or bad luck.
Hockey differs from coin tossing in three ways.
1) The odds are not perfectly 50/50. It is weighted, and you do not know perfectly by how much (but stats and eye test can help you estimate it to some level of confidence).
2) There are many large variables at play causing each coin flip to not be perfectly in the the same environment to some significant degree (coin toss has the same but it is not very significant in pulling from 50/50).
3) The human element is a variable that will cause things to fluctuate slightly from that mean. A player is not equally good one game to the next, so even the underlying weighting outside from the environment and outside variables will fluctuate.
When we hockey statisticians use the term, it's typically to represent results being different from what one should expect moving forward given what the player controls.
Now, if you want an answer to "why" Scheifele shot higher than 14% outside of variance, stats can offer you some solutions.
As I noted earlier, we have a model called expected goals. It looks at variables like:
1) Shot distance - closer => more goals per shot
2) Shot angle - tighter => more goals per shot
3) Shot type - slap, snap, wrist, wraparound, etc. have different sh%
4) Rebound shots - sh% of shots differ given time between last shot
5) Rush shots - sh% of shots differ given time and location from last non-shot event
6) Handedness - sh% from each side of ice differs given a righty or lefty shooting
Now, getting into more of the fun stuff, there are factors in shot quality we currently do not measure: shooter's accuracy, shooter's velocity, etc.
We do not account for those things. However, we (well some of us) do use regressed shooter history to adjust expected goals.
For example: Stamkos historically scores about 41% more goals than expected after accounting for those variables in expected goals. In all likelihood, given his large sample, we can safely say he's better at the things we do not measure which gives him this impact.
We can adjust by his regressed shooter history, severely reducing the impact of his non-accounted for factors.
Now there are two issues with this adjustment:
1) Some people change. Most do not, at least not significantly so, but some do.
2) Not everyone has enough history to account for. One example of this is Laine, as he was a rookie with no past history.
#2 is an issue with Laine, but you can still look at some stuff and make some assumptions.
One assumption I would make is that I expect Laine, given my eye test, to be an above average finisher, so I'm not surprised he has higher goals than xgoals.
One other assumption I would make is that last year's shooting percentage was still somewhat lucky. As I pointed out earlier, Stamkos finishes about 41% more goals than expected after accounting for the shot quality variables we can account for. That's the highest conversion rate for those with sufficient sample. Laine last year out performed xgoals almost with 140% more goals than expected... I do not expect Laine to be 3x better than average than the next best finisher in modern history.
