During the summer, like many of us, I'm an occasional fan of baseball (occasional because the area's teams only occasionally win). Baseball is renowned for it's statistical analysis of the game (see: Moneyball, SABRmetrics), affecting everything from draft strategy to stolen bases to intentional walks. One of the more recent developments in this statistical analysis are the various Defensive Rating systems.
In the past, attempting to quantify a certain player's defensive ability with respect to his peers was difficult at best: it required that you analyze every defensive chance for every player in every game played. You have to analyze starting position, distance traveled, whether the ball was caught for or turned into an out. In the end you could theoretically say that on a given play, a certain player should or shouldn't have gotten to a ball based on the performance of every one else at his position.
Could something like this work for NHL goalies? I say yes.
You'd have to anaylze every shot on goal on every goalie in the NHL for distance, angle, whether the goalie had to move laterally to react to the puck, whether the goalie was screened, whether the puck was deflected or any subsequent shots or goals resulted from rebounds. Then you'd have to group all shots of similar types together and determine the save percentage for each catergory. Then you could determine which goalies make saves they shouldn't, and which ones let in goals that most of their peers would stop.
This type of analysis would be more beneficial to goalies than a generic Save % number (which doesn't take into account the difficulty of certain saves or a goalie's performance in spite of his defense). It would identify areas for a particular goalie to work on to bring his ability up that that of his peers.
Now all I need is video of every shot on goal in the league...