Who's the better goaltender? Ryan Miller vs Roberto Luongo

I was talking to a friend of mine last night about the gold medal game, and he said he'd rather have Ryan Miller in goal than Roberto Luongo.  Why?  Miller has a 930 save percentage this season; Luongo is just 919.  Not only that, Miller has "looked better" so far in the Olympics.  I tried to convince him that 50 or so starts (let alone 4) were too small a sample size to establish a goaltender's true talent, and that we should look at a number of seasons (various people have previously found that four seasons is about right.)  He was unconvinced - so I thought we should look a little deeper. 

First, let's pick some arbitrary lengths of time to compare Luongo and Miller over - first, overall save percentage:

So Luongo and Miller have put up the same numbers over the last two months, and Luongo has been better over the last five seasons (and four and three...)  But Miller was better during the first two and a half months of this season.  Does cutting off our performance measure at the beginning of the year give us a better estimate of their relative abilities than longer or shorter periods?  Well, let's say you're still not convinced - let's eliminate bias caused by having to kill penalties and look at their even-strength save percentage over the last five seasons:

Over the course of a full season, Miller has never bested Luongo at even-strength.  Luongo was better in the playoffs too:

Over the last five seasons, Luongo was also more likely to limit his opponent to between 0 and 2 goals:

I don't think there's a whole lot of evidence to suggest that Ryan Miller is better than Roberto Luongo.  If a high save percentage in two and a half months out of one season was all you needed to be the best goaltender in the world, then Scott Clemmensen would be someone's starting goaltender; if all you needed was a handful of good games, then the Ottawa Senators wouldn't have sent Mike Brodeur to the minors.