[mkobit]

I think you are greatly understating the difference between amateurs and top players in the NHL. Watching the players coming from the feeder leagues to the NHL, you can tell that there can be a _huge_ difference in skill.

The Blackhawks this season are a good example to look at. If you just compare the goalies this year you will see a large difference. Corey Crawford has been their goaltender for 2 Stanley Cup championships, and was injured in the middle of this year. He has not played since then, and it has forced the team to try and find a replacement. I think looking at the statistics shows the difference even between the most elite goalies and those that are close to that level.

W is wins, L is losses, GAA is "goals against average" per game, SV% is save percentage

- Corey Crawford: 16 W, 9 L, 2.27 GAA, .929 SV%

- Anton Forsberg: 10 W, 16 L, 2.97 GAA, .908 SV%

- Jeff Glass: 3 W, 6 L, 3.31 GAA, .898 SV%

- J-F Berube: 2 W, 5 L, 3.93 GAA, .891 SV%

Statistics taken from ESPN [1]

[1]: http://www.espn.com/nhl/team/stats/_/name/chi/chicago-blackh...

[gringoDan]

Which of these metrics would be most important in evaluating a goalie?

I know nothing about hockey, but my intuition would be that you can throw W/L out the window - if you have an amazing offensive team you could let in 5 goals a game and still win. GAA is a reflection of both the team's defensive ability and the goalie's performance. It seems like SV% is the "purest" metric to compare apples to apples here. Unless I'm only looking at first-order effects here and the team's entire strategy changes based on who is in goal...

[dbt00]

SV% does have team and score effects, but it's the best single metric. It's often best to compare 1 - SV%, as that tells you roughly how many goals they're letting in.

So on 1000 shots, crawford allowed 71 goals, forsberg 92, Glass 102, Berube 109.

Goalie W-L is typically bullshit, but comparing goalies on the same team it can be useful. With crawford they were winning 64% of their games, without him around 35%.

[kd5bjo]

All of the stats are heavily affected by what the rest of the team is doing. If the team knows it's playing in front of a weak goalie, they'll play more defensively so that more shots are blocked in the field and only lower-percentage shots ever make it to the goal in the first place, raising the SV%.

[DrScump]

  you can tell that there can be a _huge_ difference in skill

I would say that the huge difference is in speed and size rather than "skill" (puckhandling, etc.).

If you want to see scary-fast hockey, go to an NHL training camp where everybody is fresh and rested, and half the players are fighting for jobs.