MBH
Players Play
Who will have the better save percentage this year?
Greiss or Ned?
For reference.
Last year:
Ned .932
Greiss .912
Greiss or Ned?
For reference.
Last year:
Ned .932
Greiss .912
IMO completely unpredictable
It also won't even tell us who played better. Unless it's markedly different.
Goalie stats require a ton of context.
Over the course of the season, the save percentage of two goalies on the same team can say quite a bit.
If they played the same number of games, if they both played a good mix of opponents, if they got a fairly equal share of second game of back to backs, if one doesn't get more starts when there are significant injuries on the team, then yes I think it could say something but I still wouldn't call it a lot.Over the course of the season, the save percentage of two goalies on the same team can say quite a bit.
If they played the same number of games, if they both played a good mix of opponents, if they got a fairly equal share of second game of back to backs, if one doesn't get more starts when there are significant injuries on the team, then yes I think it could say something but I still wouldn't call it a lot.
I don't watch the Kings as much these days but when they were a defensive powerhouse Quick's stats would often swing wildly from game to game because he faced so few shots. Sounds great but they give up a doozy of a turnover, the opponent scores then his save percentage takes a big hit in spite of playing a great game overall.Assuming 40 starts each and 30 shots a game, the difference between a .920 and .930 is 12 goals.
Now factor in random bounces, difference in shot quality, and quality of competition... And it's not hard to imagine that those two goalies could have performed nearly the same give or take a handful of goals.
Mike Smith and Vasi had nearly identical save percentages last year. I'm not going to read too much into it, you know?
Now if we're talking .930 vs .910 then it becomes much harder to find enough random variance to explain that.
If they played the same number of games, if they both played a good mix of opponents, if they got a fairly equal share of second game of back to backs, if one doesn't get more starts when there are significant injuries on the team, then yes I think it could say something but I still wouldn't call it a lot.
That depends entirely if they are different enough.
In the past you and I have had a very different definition for statistical significance.
.,899 vs .912 is pretty f***ing signficant.
It’s a profession where the difference between leading the league and being a fringe player is four shots in a hundred (.940 save percentage versus .900). Humans make
mistakes, and they will appear in random patterns. And while there approximately 650
skaters in the NHL at any given moment, there are only 62 goalies. That is simply a small group of athletes, and the distribution of their performance may not normalize.
Here and now, we see the improvement in statistical quality by using All Attempts or
Clean shots, utilizing expected goal models, understanding how much the team factors
into an individual goalie’s save percentage, and beginning to visualize specific aspects
of goalie performance with hot zones.
It's "very similar deployment" that's the key. And doesn't usually happen with the two goalies on a team. I think it's where backup goalie love often comes from. Backups get fewer starts, tend to get easier opponents, so if they play well some people think they should be starting. But it's not the same as being the starter.So two goalie splaying in front of the same team, with very similar deployment - and that wouldn't trust save percentage as saying "a lot."
And people wonder why nobody gives a f*** about goalie stats when it comes to Hall of Fame stats.
It's "very similar deployment" that's the key. And doesn't usually happen with the two goalies on a team. I think it's where backup goalie love often comes from. Backups get fewer starts, tend to get easier opponents, so if they play well some people think they should be starting. But it's not the same as being the starter.
Goalie stats over a career is a different thing, but still needs context.
But it actually might not be significant. That's the point.
You'd need to run a more comprehensive look at multiple data sets to actually draw meaningful conclusions that exhibit wisdom instead of drawing conclusions from base level averages.
Someone actually wrote a scholarly paper on it.
Link to PDF:
Goalie Analytics: Statistical Evaluation of Context-
Specific Goalie Performance Measures in the
National Hockey League
https://www.google.com/url?sa=t&sou...MQFnoECEMQAQ&usg=AOvVaw0aWLd2i3nX0VwhgRnyJpuS
Assuming 40 starts each and 30 shots a game, the difference between a .920 and .930 is 12 goals.
Now factor in random bounces, difference in shot quality, and quality of competition... And it's not hard to imagine that those two goalies could have performed nearly the same give or take a handful of goals.
Mike Smith and Vasi had nearly identical save percentages last year. I'm not going to read too much into it, you know?
Now if we're talking .930 vs .910 then it becomes much harder to find enough random variance to explain that.
Strength of Opponent alone is enough alone to skew stats.