Todd McLellan defended his team's PK performance before Monday night's game against the Kings, noting that the Sharks had "done a lot of analysis" and that winning 2 or 3 faceoffs virtually guaranteed that you wouldn't get scored on on the penalty-kill.
Let's see how this works from first principles - I noted earlier that winning 100 additional faceoffs on the power-play was worth +3.66 goals, or 0.0366 goals per faceoff. Let's look instead at the 120 seconds following faceoffs to start a penalty-kill:
|Faceoffs||Goals Vs||Goals For||PP%||SH%|
If I go to two decimal places, the difference between winning and losing that initial faceoff is +3.64 goals (per 100 faceoffs) by this method. So the two analyses agree as we would expect them to given the massive sample size.
Now, what about the case of two faceoff wins on a penalty-kill? It's important to remember how a power-play operates. On average, a team gets one shot per minute on the PP, and if there's a second faceoff during a PP, that means that the PK team saved a shot already, so on average it should only see one more shot over the remainder of the man-advantage. Here's how the numbers work out:
|PP%||1 FO||Lose 2nd||Win 2nd|
Given that a second PK faceoff generally takes place after the penalized team stops a shot, it's no surprise that the initial faceoff has no impact on what happens after the 2nd faceoff. And, as you can see, once you win that second faceoff, it's virtually guaranteed that you won't get scored on over the course of the entire two minutes because you've now dumped the puck down the ice twice and stopped the attacking team's shot after allowing them to gain the offensive zone and get into position.
So what McLellan says is true, but it's also a tough thing to execute - even if you had the best faceoff taker and the best goaltender in the league, you wouldn't even be twice as likely to have one of these two-and-done PKs as an average team - which is to say, not very likely.