Somehow, Schneider

I was going to write a pretty long post today about Cory Schneider’s regression to the mean. I think he’s been pretty lax on a couple of straight starts and hasn’t looked like he’s been pulling his weight as much as he did during that stretch at the end of November and his numbers are finally returning to his career average.

So I looked into it, under the assumption that Schneider hadn’t put up a quality start in three starts. For those who are unawares still, a quality start is a Hockey Prospectus statistic that counts whether or not a goaltender played well enough to allow his team to win: to earn a quality start, a goalie must stop 91.3% of shots in a single game, or stop 88.5% of shots while allowing two or fewer.

Turns out, despite my observations, that Schneider put up quality starts against both Anaheim and Boston, and although he definitely didn’t get one last night in Tampa Bay, he also didn’t get “blown up” as was my initial concern. A goalie gets blown up when he fails to stop 85% of shots or surrenders 5 goals in a game on fewer than 40 shots.

Here’s a rundown of all of Schneider’s appearances this season:

Game Result EV SV%
Columbus Quality Start 0.882
Detroit Quality Start 0.905
Minnesota Quality Start 0.898
Edmonton (Relief) N/A 0.909
St. Louis   0.903
Minnesota   0.891
Anaheim (Relief) N/A 0.891
Chicago Blown Up 0.888
Ottawa Quality Start 0.898
Colorado Quality Start 0.908
Phoenix Quality Start 0.916
San Jose Quality Start 0.919
Columbus Quality Start 0.930
Nashville (Pulled) Blown Up 0.927
Colorado (Relief) N/A 0.929
Carolina   0.926
Anaheim Quality Start 0.925
Boston Quality Start 0.925
Tampa Bay   0.920

The EV SV% column is a cumulative look at Schneider’s even strength save percentage. His career average coming into this season (including playoffs) was .923.

However, what’s the deal with this, here?

  EV SV% QS Rate
Before ’11/12 0.923 62.0%
11/12 Season 0.920 62.5%

Schneider’s quality start rate has slightly increased, but his even strength save percentage has slightly decreased. On a game-to-game basis this season, his EV SV% has not been as strong as his quality start numbers would reflect. Schneider’s save percentage while a man down is .958, which is higher than his rate at 5-on-5 or 4-on-4. If that seems weird to you, it’s because it should. Our friend Jonathan Willis has already investigated this phenomenon—goalies that have really high save percentages when down a man can rarely repeat the feat. It’s been theorized that a goalie’s PK SV%, in fact, is reflective of how the team plays in front of him, rather than the goaltender being the best penalty killer as has been waxed in many a hockey broadcast.

So Cory is having a good season, but by checking the game-by-game EV SV% statistics, you can tell his success is somewhat illusory. Not much, but somewhat. This isn’t to disparage the guy in any way. Considering what you expect out of a backup goaltender, he’s having a strong season and is a big reason why the Canucks are where they are in the standings.

Luongo, on the other hand, has just a 53.8% quality start rate this season, but he’s at 73.3% since he returned from injury, and, in his last 16 starts, like we’re counting for Schneider, that number increases to 75%. The difference is that Luongo’s EV SV% on the season is .928 and his PK SV% has turned around after a pretty ugly start, but at .871 it sits well below the overall estimated mean of between .880 and .890.

On a game-by-game basis, sure, I guess we can credit a goaltender for stopping 18-of-18 PK shots like Schneider has in his last three games (in chunks of 7, 6 and 5) but I think we need to recognize that those sort of performances aren’t entirely repeatable.

  • Given the sample size when calculating PK SV% in a given season (especially for a backup), it seems likely to me that you can draw absolutely no conclusions from Schneider’s .958, i.e., not only is it not indicative of his ability to make saves in a PK situation, it also says nothing about the quality of the team’s PK in front of him.