15 Comments
User's avatar
Stephen Meserve's avatar

This is a great example of using data to not overreact to an anecdote. Sean saw an icing play come back to bite a team that he was watching with his own two eyes. Then, digging into the data, it's one of only three times that's happened all year in the entire 32-team league? Yea, that's the challenge of anecdotal fallacy.

Expand full comment
Ralph Strangis's avatar

The “more than half the time it’s icing” argument isn’t really an accurate reflection of the data and needs to be played out. You need to factor “neutral” outcomes - which is to say - defending team possession on a post icing draw isn’t an adverse outcome. Basically what the data tells me is - 1 in 3 you end the game. About half the time it’s an icing. And less than 1 % of those wind up as a goal by the trailing team. It’s more than Mac and those guys saying we got better shooters - although there’s probably some truth to that. It’s numbers.

Expand full comment
SebA's avatar

Brilliant stuff, I subscribed off the back of this!

Expand full comment
Sean Shapiro's avatar

Thank you!

Expand full comment
Aaron King's avatar

To accurately compare outcomes, you also need to know the likelihood that the teams score if you don’t ice the puck and try to keep possession.. the difference in likelihood of scoring trying to shoot vs holding on to puck is the change in risk of allowing a goal. That’s much harder to measure though as I am sure the NHL doesn’t keep stats on “could have shot from over 100 feet but didn’t”

Expand full comment
Sean Shapiro's avatar

Fair point, kind of why this is base level stuff and what I’m able to do with public facing data. Would love to be able to ask this question with what teams actually have access to.

Expand full comment
Aaron King's avatar

By the way, should have started this comment with love the post and the question and the data you were able to get.. hopefully it didn’t come across as critical - just thinking through the analysis

Expand full comment
Frank Dantonio's avatar

This begs another empty net question, which players get the biggest stat boost from empty net and overtime goals and assists? ANd are these not partially a factor of ballooning point totals in recent years?

Expand full comment
Sean Shapiro's avatar

I just don’t think there are enough empty net goals for it to be a reason for overall ballooning offense. But I haven’t dove into that question further.

Expand full comment
Aaron Knodell's avatar

Shooting on empty nets is one of the clearest ways stats have improved decision-making/strategy. The only thing I don't buy is skill being the reason for the uptick. You'd have to be as bad as me to make shooting for the empty net not worth it.

Expand full comment
Angie's avatar

Based on the numbers, shooting is 28 times more likely to result in a goal for than a goal against? Why wouldn't you green light that?

Expand full comment
Carl's avatar

Do literally anything that gives you a 35.5% chance at a goal.

Expand full comment
Ronnie Hughes's avatar

This information has a statistics major somewhere giving himself the spins in his dormitory with 3 energy drinks down the hatch. IMO I don't have anything bad to say about it. These guys are NHL players for a reason. I think it boils down to how much time they have. Decision making, do you have the time to actually get off a shot? Go for it. Is it rushed and a bit of a shot in the dark? Take some off of it and hope it's short of icing. Long winded way of saying, I think it's situational lol

Expand full comment
Michael Strawn's avatar

Let 'er rip!

Expand full comment
Dave Shapiro's avatar

Great analysis

Expand full comment