Analytics with Alison: Understanding Icings

If you've been around the game of hockey, it's been an adage for years: "don't ice the puck, it can give your opponent an unfair advantage!"

Why is that? In its purest sense (there are some nuances, we'll talk hybrid icing another day), icing is when a team playing in their own end of the ice (the defensive half) sends the puck out of their own zone and it crosses the goal line at the other end of the ice without being touched.
When this happens, icing is called on the defensive team and the puck comes right back to the defensive zone, giving your opponent an offensive zone faceoff. That surely sounds opportunistic for your opponent. Add on the fact that if your team ices the puck, you are not allowed to make any changes of which players are on the ice, AND the other team gets to pick which faceoff dot is used (and can put out a fresh center) and icing the puck sure seems like a not great thing to do.
This was something Kraken radio color commentator Dave Tomlinson had engrained in him through his years in the game. But, it was also something he'd started to question. And so, at a Kraken practice in December we started to break it down.
"From a young age we're told 'don't ice the puck if you don't have to...gain center ice to do so,'" Tomlinson said. "I wanted to see if data would prove that right or wrong. My original thought was that icing the puck to get a reprieve when tired was less dangerous than trying to make passes out of the zone when you're at the end of a shift and hemmed in."
We were digging into the data already with our friends at Sportlogiq, and in the meantime, the Kraken went to Dallas to play the Stars. Twice in the first period, the Kraken were called for icing, and within seconds, the goal was in the back of their net.

Tomlinson and I got to talking. "I've never seen such blatant examples of (a goal coming off an icing)," he said.
So, is committing an icing really that dangerous or that much of a threat to your team?
Let's dig in.
Sportlogiq (with a special stick tap to Nick Czuzoj-Shulman) ran data on eight seasons of games and here's what they found (all data represents 5-on-5 play).
FACEOFF WINS:
GOALS AFTER AN OFFENSIVE ZONE FACEOFF WIN (Immediate possession following):
So, it doesn't look like there is that much impact on goal scoring after an icing. As an example, there were 66 faceoffs in the Seattle-Dallas game. If we apply the average win rate of 47.11%, the difference in goal scoring if all of these faceoffs came off an icing versus in normal play is .06 goals.
But, (our gut says) icing feels like a desperation move when players are tired or not connecting on a play, does it matter how long skaters were on the ice before the icing occurred? After all, if players were already tired, surely they're going to be fatigued as they must stay on the ice for the ensuing faceoff.
According to Sportlogiq, the average shift length for defensive players prior to a faceoff in a team's own zone is 13.6 seconds. This does go up to 31.8 seconds prior to an icing, but, there's not much differential when it comes to any resulting goal scoring. The only real area where shift length starts to give attacking teams a real advantage is when the defending team has shift lengths of 60 seconds or more, and, yeah, that makes sense.
Ultimately, understanding what makes icings more dangerous is understanding what truly makes an impact. It's not necessarily about the fatigue of the defending team, it's more about setting up your opponent to pick who and on which side they want to take a faceoff (which also often puts defensive teams on a sub-optimal side for faceoffs).
This is something Tomlinson had sniffed out already as part of his original thought process.
"The new rule about picking which faceoff dot after an icing is excellent," Tomlinson said. "Traditionally great faceoff teams are probably licking their chops (at an icing opportunity) … but my gut tells me you should be able to get the puck out of the zone (without worrying too much about icing)."
Knowing what we now know, let's circle back to the Dallas game to feel what the impact really is of an icing (at least in this game).
Yes, of course those two goals sting. No negating that. But there were 16 icings in the game across both teams. Five by Seattle, and 11 by Dallas. Of those 16, two gave way to goals, even though both teams lost more faceoffs than they won after icing the puck (Seattle won 1, lost 4; Dallas won five lost six).
At the end of the day, as Sportlogiq's Kenny Matheson said, icing the puck may give your opponent a small advantage, and specific circumstances (faceoff specialists; total volume of faceoffs/icings, etc.) may make icings more costly, but their impact is not truly significant to the overall outcome of the game.
It's those special circumstances that invite the impact of coaching, pre-scouting and deployment to take advantage of any situational play.
"Stuff gets drilled into our head by coaches and we don't question if it's actually what will or will not happen," Tomlinson said. "When I think about the game, I try to think 'if I was a coach, what would I want my team to do?'"