When Derek wrote a piece on Nikolai Khabibulin's true cost to the Oilers over the last several seasons, he decided to use the league's combined save percentage as a guideline to determine that value. To me, that didn't make a lot of sense because freely available talent (i.e. replacement level talent) usually isn't able to deliver average performance. In order to give us a better benchmark, I've taken a stab at establishing replacement level for NHL goaltenders and will take a look at the results after the jump.
In order to establish replacement level, I thought it would be wise to look at those goaltenders that have served as replacements at the NHL level, and so I began by looking at the combined save percentage of those goaltenders who appeared in 15 or fewer games during a given season. One of the problems with that system was sample size. These goaltenders generally faced between 2,000 and 3,000 shots at even strength and 500 to 1,000 shots on the penalty kill. As such, I decided to resolve that problem by using the past three seasons to serve as replacement level for any given year. So, for example, the 2007-08 season would use the results from 2005-06, 2006-07, and 2007-08. Here are the results:
That's a pretty sharp difference in quality from the combined save percentage of all NHL goalies during these seasons:
The big takeaway for me isn't the performance of Khabibulin (unsurprisingly, he's below replacement level and still terrible value), but with how underpaid goaltenders are generally. Gabriel Desjardins has calculated the UFA cost for players over the last several seasons, and suggested that the price per goal above replacement for the 2011-12 season was about $405,000. If we had a goaltender who faced 1,200 shots at even strength with a .921 save percentage and 300 shots on the PK with an .875 save percentage, that would have been worth 13.8 goals above replacement or $5,589,000. Add that to the minimum salary of $525,000 and we get a total salary of $6,114,000. And that goalie is the definition of mediocre!
I'll admit up front that the analysis here is pretty crude (assigns all credit in save percentage variation to the goaltender and all credit for variation in shot totals to other players), and it's obvious that NHL teams don't use the same scale when they pay goalies as when they pay skaters. Furthermore, it's possible that I've set replacement level a little bit too low. Still, it seems to me that if you can be confident that a goaltender will deliver superior performance, you're likely to get better value paying him than you will paying players at other positions.