Quote:
Originally Posted by DeluxeMoustache
It’s a good attempt, a work in progress, and a basis for discussion.
I do not say it’s exclusively team not goalie.
Generally, there are shots you should stop, those you could stop, and those you can’t stop
I believe that roughly (focus on the concept, not the exact numbers) that, say, somewhere over 80 but less than 90 percent of shots *should* be stopped. Team play (reference the Nashville 21 shot 3rd period) can affect these shots that can pad goalies’ stats. Team play can certainly impact quantity of shots that can’t be stopped. Goalies can make the difference on those *could be stopped* shots
So from the area of league average of sv.%, team play can contribute +/ say 10 or so points, and a goalie another 10 or so points (a poor goalie can also, of course, really affect his own downside)
There is no personal bias, by the way, in noting the shortcomings of the models. They simply don’t measure shot placement. A shot that ends up bar down can be classified the same as a shot put right in to the goalie’s logo.
And there’s not enough shots on a single game sample size for different quality of shots to come out in the wash.
I don’t simply read a number and say “look at me, I’m unbiased”. I look at a number, think about how it is calculated, and think about how all of the things I see, that are not measured, could impact that number if they were measured
|
And you'd be right. But the inconsistency is applied across every team and every goaltender.
So to ignore the number you'd have to have a case where team X or goalie Y have specifically wrong numbers because they thwart the model more than their counterparts.
You have a huge bias towards goaltenders, you walk it out 12 times a year.
And I don't say "look at me" about anything.
But I certainly love that you can look something up that principally unbiased and negate someone who walks into every situation with an eye test bias and an inability to see anything but what they've told themselves to see (not pointing at you directly on that one)