By The NumbersHockey Analytics... the Final Frontier. Explore strange new worlds, to seek out new algorithms, to boldly go where no one has gone before.
On the whole save percentage and workload are not correlated but if you break it down to goalie by goalie you will find some goalies do better with lots of shots and other goalies perform better with less shots.
I'd love to see that analysis (particularly if the results are distinct and truly not the result of statistical variation at smaller sample sizes).
Intuitively, I agree - there are likely goalies who do better with high workloads (I'd guess Craig Anderson would fall into that category), and there are likely goalies who do better with low workloads (Ken Dryden seems to be the archetype here).
However, I've tried to tease out this in the data on multiple occasions now, and I've never been able to show anything that couldn't be explained by sample size.
there are likely goalies who do better with low workloads (Ken Dryden seems to be the archetype here).
Something I've never noticed until now, but looking at the above, Ken Dryden's shots per game decreased from year to year in every single one of his seasons (from 38.7 in his short rookie season to 25.4 in his last year).
It's a losing battle, I'm afraid. A lot of people will simply accept that save percentage is worthless, and that shot quality varies wildly (but is apparently undetectable) from team to team. It's because it allows them to claim their favourite goalie is better while downplaying the success of another team's goalie.
It's a losing battle, I'm afraid. A lot of people will simply accept that save percentage is worthless, and that shot quality varies wildly (but is apparently undetectable) from team to team. It's because it allows them to claim their favourite goalie is better while downplaying the success of another team's goalie.
Oh stop. Shot quality is somewhat detectable. You have to understand what's happening on the ice and realize that the number is a by-product of a series of events. Understand what creates the number, don't try to re-create what the number appears to represent.
It's a losing battle, I'm afraid. A lot of people will simply accept that save percentage is worthless, and that shot quality varies wildly (but is apparently undetectable) from team to team. It's because it allows them to claim their favourite goalie is better while downplaying the success of another team's goalie.
I don't think that that's the case (maybe it's the professor in me).
I agree that there will always be naysayers, but I also think that there are people who genuinely want to learn about this stuff (plus, once you're educated on this stuff, it's easier to selectively apply it to your own team anyhow ).
And you can risk-adjust this stuff (with the modern level of data available); even-strength save percentage is one of those flavors (and the easiest available), but you can do so much more. The problem is that you can't then apply it backwards to the older time periods.
Shot quality does vary, but we don't really fully understand yet *how* (or to what extent different factors wash out or don't wash out).
Ultimately, statistics reflect events on the ice, and to the extent that they can be done better or more accurately, that's good, but they're always going to be a one-dimensional representation of a four-dimensional process.
The best ones around atm are the digr based ratings but they are hard to get up to date data for. It's not perfect, they don't take into account high/low workloads, no idea of whats happening in the ice with defense placement etc. It's just shot distance/angle, but they are better than sure sv%.
It actually kinda suggests that Sv% is the meaningless stat here..
Not really.
It basically suggests the basic goalie stats need context to be relevant.
Halak's shutouts are potentially inflated by playing behind a great defensive team. Halak's SV% may be deflated because he doesn't face enough shots to keep him sharp.
__________________ CanadianHockey________ __ __________Sens, Oilers, and Team Canada
Generally, a team's defense has two primary goals (in one order or the other): limit the number of shots a goaltender faced, and limit the quality of those shots.
A team's goaltender has one primary goal: of the shots he faces, stop as many of them as possible.
Elliot allows 4 goals on 11 shots tonight, gets pulled, Jake Allen replaces him and allows a goal on his first shot.
All St Louis goalies are doing badly sv% wise
i may be wrong, but it seems to me that one of the main problem of measurements of shot quality is that they do not account for passes before the shot that force the goalie to move.
wrist shot from the faceoff dot on the rush and wrist shot from the faceoff dot off a cross-ice pass are very different, but i think are recorded the same in shot quality.
i may be wrong, but it seems to me that one of the main problem of measurements of shot quality is that they do not account for passes before the shot that force the goalie to move.
wrist shot from the faceoff dot on the rush and wrist shot from the faceoff dot off a cross-ice pass are very different, but i think are recorded the same in shot quality.
Exactly, and I believe you're right about how they're recorded.
It basically suggests the basic goalie stats need context to be relevant.
Halak's shutouts are potentially inflated by playing behind a great defensive team. Halak's SV% may be deflated because he doesn't face enough shots to keep him sharp.
Oh definitely.. i was just countering his extreme statement with another.
Context is everything when it comes to stats... you cant take one in a vacuum hoping it tells you the whole story.
I have to say, I'm interested in the mathematical principle that dictates that facing a lower of number shots inexorably leads to a lower save percentage.
Perhaps all of the mathematicians in this thread who have asserted such a relationship can enlighten me.
Like through adducing some kind of supporting evidence.
I'd say the obvious thing is that when you face less shots, your save percentage is more variable than a larger sample size. If a few shots are lucky, perfect shots, or you have a small lapse, it affects you far more than the same occurring with more shots.
Halak is still a very good goalie. However I think Elliott is just a good backup. When you're playing for a team that gives up few shots a night you can't expect to just give up only one goal just because you only surrender 15 shots.
Also facing fewer shots can cause a goalie to get cold. I'm pretty sure Brodeur has a much better save percentage career facing a lot of shots, than he does facing fewer. I think most any goalie does.
Halak is still a very good goalie. However I think Elliott is just a good backup. When you're playing for a team that gives up few shots a night you can't expect to just give up only one goal just because you only surrender 15 shots.
This.
Elliot has proven he can't handle a starter's role, IMO. Halak needs to take the bull by the horns and to be the clear cut #1 Goalie for the Blues, and stay healthy!
).
