Serious Debate - Goaltending

[Boom Boom Apathy]

One suggestion: include shots against and minutes played.

At the end of the season the accumulated totals would allow for rate-based metrics.

Good idea. Soon we'll have enough sample size where I won't list individual games and can start adding metrics that are more meaningful.

 

[Boom Boom Apathy]

This is an interesting metric, but seems wildly subjective. The idea that one Penguin goal was “stoppable” but none of the Blues goals were is pretty crazy to me.

Sure it's subjective. One issue with advanced stats is that they also lack that. For instance, Jordan Staal taking a shot from 6 feet out with no traffic would likely be treated the same in terms of High Danger as Kucherov taking a shot from 6 feet out with traffic.

With the chips in the puck and shoulder pads, as well as sensors around the rink, the advanced data will start to try to account for that sort of stuff more and more and will improve over time.

I still think advance stats are still usable in large sample sizes as some of those things average out, but you can't completely ignore subjective measures.

 

[AhosDatsyukian]

I think it's way too early to have a "serious" debate about any aspect of this team.

 

[Boom Boom Apathy]

You probably weren't around when BlueDevil58 posted here. It was NEVER a wrong time for him to have a Serious Debate. DISCUSS!

 

[Svechhammer]

Here me out, I think a debate about this team can be about trends over the last few years and not just a sample size of 3 games this year. Like, how we all know Kooch plays better when allowed to be a workhorse for a long stretch, so its stupid to give him spot starts right now if we have any desire for him to take the #1 spot this year. DISCUSS

 

[Joe McGrath]

Sure it's subjective. One issue with advanced stats is that they also lack that. For instance, Jordan Staal taking a shot from 6 feet out with no traffic would likely be treated the same in terms of High Danger as Kucherov taking a shot from 6 feet out with traffic.

With the chips in the puck and shoulder pads, as well as sensors around the rink, the advanced data will start to try to account for that sort of stuff more and more and will improve over time.

I still think advance stats are still usable in large sample sizes as some of those things average out, but you can't completely ignore subjective measures.

I don’t think your example has any subjectivity. Any shot from XYZ distance is considered “high danger”. No matter who takes it.

This is literally one person judging if the goalie should or could have made the save or not and I don’t know the criteria per se but I already disagree with it.

Does he give explanations like tarheel did way back when or is it just updating numbers?

 

[Boom Boom Apathy]

I don’t think your example has any subjectivity. Any shot from XYZ distance is considered “high danger”. No matter who takes it.

I think you missed the point. Every shot from xyz distance is considered “high danger”, thus no subjectivity. I’m reality, we know not every shot from that distance is truly has the same level of danger, so it lacks subjectivity that is needed to truly assess the quality of a scoring chance. Not saying this is the best way, I’m just acknowledging shortcomings of the metrics.

This is literally one person judging if the goalie should or could have made the save or not and I don’t know the criteria per se but I already disagree with it.

Ok. I have no problem if you disagree with it. Just ignore that piece of it.

Does he give explanations like tarheel did way back when or is it just updating numbers?

He’s been doing it for many years and used to give explanations why, and still does occasionally when asked about a on a given goal, but doesn’t explain every one anymore.

I’m not trying to convince you, or anyone on the merits of it. I know it’s very subjective. I like reading the thread. If you don’t, that’s fine.

 

[A Star is Burns]

Here me out, I think a debate about this team can be about trends over the last few years and not just a sample size of 3 games this year. Like, how we all know Kooch plays better when allowed to be a workhorse for a long stretch, so its stupid to give him spot starts right now if we have any desire for him to take the #1 spot this year. DISCUSS

I know you had your inside info, or whatever, but I have been pretty much of the opinion that they don't necessarily intend for Kochetkov to take the #1 role unless Freddie is bad, injured, or possibly when he's given the job in the playoffs. I think they largely continue to roll both guys out there.

Pyotr has to be able to overcome that hurdle with respect to playing time. He looked great against the Devils when he definitely hadn't played recently. He looked somewhere between meh to fine depending on if you think he could have gotten any of the goals against the Blues (I happen to think one or maybe two were "stoppable"). And for his playing time this year, at least, we can argue he should have started the opener, but we did have the weird weather situation that took away a back to back where he's the obvious starter and you'd obviously split guys. I don't think they've really done anything wrong with how they've played them this year.

 

[cptjeff]

Sure it's subjective. One issue with advanced stats is that they also lack that. For instance, Jordan Staal taking a shot from 6 feet out with no traffic would likely be treated the same in terms of High Danger as Kucherov taking a shot from 6 feet out with traffic.

With the chips in the puck and shoulder pads, as well as sensors around the rink, the advanced data will start to try to account for that sort of stuff more and more and will improve over time.

I still think advance stats are still usable in large sample sizes as some of those things average out, but you can't completely ignore subjective measures.

FWIW, NHL teams use this sort of expert subjective analysis themselves, and have done so for years now. There are commercial providers and I'm sure some teams do similar work internally. They watch video and track puck movement immediately prior to every shot, accounting for distance laterally traveled, speed the receiving player gets the shot off, screens, shot speed, and other factors, and grade each shot by difficulty. Doing that right requires full time skilled staff and is thus not a free or cheap product easily available to fans, so it doesn't make its way to forums like this much, but it's out there. The shot distance alone stuff is better than raw shot totals, but I don't think it's taken all that seriously by teams that are truly serious about analytics.

And of course, with the puck and player tracking, you can now begin to automate what has traditionally been experienced hockey analysts sitting in a room and watching video. I would be far from shocked if teams weren't also throwing that data into AI models now as well.