Here's a chart that you should always keep in mind when evaluating goaltenders:
The average even-strength save percentage across the league is approximately .919. (It's actually slightly lower, but that includes sub-replacement goaltenders used for emergency purposes only.) If you call up a goaltender and he plays 40 games, sees 1000 even-strength shots and posts a .925 even-strength save percentage, what are the odds that his true talent level is actually league average or worse? Reading off the chart, it's almost 30%. Would you sign that guy to a big contract given those odds? Even-strength save percentage happens to be the most reliable measure we have - if you used overall save percentage or even - god forbid - goals-against-average, you'd be taking a much bigger risk.
I built this chart using a coin-tossing model, so it doesn't capture some of the deterministic drivers of save percentage. Over a small number of games, it is entirely possible that a goaltender faced particularly strong competition or played behind a defense that was missing its star players, which would change our expectation. But what are the odds that the GM signing a goaltender paid attention to subtleties like that?
This is what makes it very difficult to find a good goaltender - you need four years of data to figure out how good he is, but you can usually only get one or two. Some people would say that lets GMs off the hook when they sign a guy to a big contract and he doesn't work out. I would say that level of risk means you should bet a smaller percentage of your portfolio on goaltending.