Devils 2020-21 team discussion (news, notes and speculation) - part XIX

[MartyOwns]

Hockey analytics are about to make an incredible jump, the tech is there or getting there now. I think a lot of the eye test people are going to get lit up hard this next decade.

Doesn't mean the organization / character-building stuff isn't timeless, though.

if you're right, a lot of scouts will be out of a job soon. i don't think that will happen. advanced stats should be used to supplement the eye test, not the other way around.

 

[Blackjack]

Someone here (elsewhere on hf) posted a video of a Forsburg goal that was a one timer from right next to the goal mouth where the xG value was something like 0.15 (it was almost a tap-in goal) and Micah casually mentioned the other day that the Devils had zero shots from beneath the faceoff dots when Palmieri had scored a goal from there. So one problem with advanced stats is that it seems like there may be some problems with the underlying data.

 

[Oneiro]

Is this in reference to player tracking? I always felt like that was the next leap forward.

For sure. There are a lot of game states in hockey but far fewer than what is being worked on in other domains.

For me, one of the bigger things I'm interested in is seeing more players improve in an atypical fashion. A lot of the debates we have here seem to be centered around whether a guy is as fixed a property as the stats assume he is and in most cases, the stat sheet is right. But maybe that's just because we don't yet know have information that creates a path to actionable improvement. In that sense, it's kind of pathetic to me that firing or hiring a coach unlocks so much value on a regular basis.

 

[My3Sons]

I saw him in "A Midsummer Night's Dream" one summer in the Berkshires in Massachusetts. I kid you not. It was pretty much what you'd expect. They did let him throw in an "excellent" at one point.

 

[Oneiro]

if you're right, a lot of scouts will be out of a job soon. i don't think that will happen. advanced stats should be used to supplement the eye test, not the other way around.

Yeah, many will be, but the best ones will be well-compensated. But that's every field right now. Some suggest that a mix of AI/human judgment is the way forward and I don't see why not, though I have no basis for saying that. If you can line up a bunch of players that check both analytical and intangible boxes out of a small pool, one, I think any advantages get competed away fast and two) you're still putting probabilities on your side. There's an argument that human judgment might be even more important in that context.

Ultimately, we're talking about people. A guy becomes a parent and his game dips or improves. Not much can be done ahead of time for those things so there will always be room for perspectives.

 

[Richer's Ghost]

if you're right, a lot of scouts will be out of a job soon. i don't think that will happen. advanced stats should be used to supplement the eye test, not the other way around.

You just spoiled the movie Trouble With the Curve.

 

[Derps]

Hockey analytics are about to make an incredible jump, the tech is there or getting there now. I think a lot of the eye test people are going to get lit up hard this next decade.

Doesn't mean the organization / character-building stuff isn't timeless, though.

Maybe, but if there were numbers that accurately represent what is going on during the game to the degree that people think, I'd be more than happy to jump on board. There are many people in the same boat. The goal should be to get the best information possible.

There is an underlying assumption that the models these websites use now to come up with stuff like xGF are accurate to the point where 3-5% differences are enough to determine whether a player had a good game or bad game. As someone that writes mathematical models for a living, it's absurd. And the application and misuse of these numbers on a regular basis is infuriating.

 

[TheUnseenHand]

Maybe, but if there were numbers that accurately represent what is going on during the game to the degree that people think, I'd be more than happy to jump on board. There are many people in the same boat. The goal should be to get the best information possible.

There is an underlying assumption that the models these websites use now to come up with stuff like xGF are accurate to the point were 3-5% differences are enough to determine whether a player had a good game or bad game. As someone that writes mathematical models for a living, it's absurd. And the application and misuse of these numbers on a regular basis is infuriating.

Preach.

 

[My3Sons]

Maybe, but if there were numbers that accurately represent what is going on during the game to the degree that people think, I'd be more than happy to jump on board. There are many people in the same boat. The goal should be to get the best information possible.

There is an underlying assumption that the models these websites use now to come up with stuff like xGF are accurate to the point were 3-5% differences are enough to determine whether a player had a good game or bad game. As someone that writes mathematical models for a living, it's absurd. And the application and misuse of these numbers on a regular basis is infuriating.

If the future of analytics is anything like those NFL commercials that show a long pass with some percentage of catching it with various alleged measurable and arrive at some sort of number they leave a lot to be desired. What they don't include is who is catching the ball. I'm sure that the percentage chance on the famous Monday night catch by Odell was less than 1%. Nonetheless he made it look easy. The human element will always matter and can't always be measured. At least, as you've noted, not yet.

 

[billingtons ghost]

that is absurd

how can you compare players across teams or leagues? You can't possibly watch every minute of every player of every game.

That is why stats are important in context.

Don't change the topic.
We are not talking about comparing every team in every league. Inherently, when discussing the 'eye test' we are talking about games that are viewed, and the opinions that are formed by watching them.

I trust the opinions of a fanbase of informed hockey fans all watching all of the same games more than I trust some loosely correlated collection of events that are at the discretion of some bean counter in their home arena. Lord knows that the interpretation of a Corsi event or shot block or hit or whatever is so variant from counter to counter that it is hard to take seriously as it is. The official scoresheet can barely manage to get Shots on Goal correct.

Yes - stats are important - IN CONTEXT. And absolutely no one here with their gUy I LIke hAvE Good CF%!!!!!!! mEANs EH GOOD!!! posts run fast and loose with context, and sample sizes? - and they are building on sand to begin with when it comes to the data.

 

[billingtons ghost]

Hockey analytics are about to make an incredible jump, the tech is there or getting there now. I think a lot of the eye test people are going to get lit up hard this next decade.

Doesn't mean the organization / character-building stuff isn't timeless, though.

No they won't get lit up hard - they'll get on board.

I don't object to stats. I object to TODAY'S stats and those who hold them sacrosanct.

 

[Oneiro]

Maybe, but if there were numbers that accurately represent what is going on during the game to the degree that people think, I'd be more than happy to jump on board. There are many people in the same boat. The goal should be to get the best information possible.

There is an underlying assumption that the models these websites use now to come up with stuff like xGF are accurate to the point were 3-5% differences are enough to determine whether a player had a good game or bad game. As someone that writes mathematical models for a living, it's absurd. And the application and misuse of these numbers on a regular basis is infuriating.

Yeah, it's out of my wheelhouse and far closer to yours from the sound of it. But for sure, data being tacked-on in a biased argument as opposed to the start of a bigger question is the problem here and elsewhere. But that's human nature, I guess.

 

[Derps]

No they won't get lit up hard - they'll get on board.

I don't object to stats. I object to TODAY'S stats and those who hold them sacrosanct.

That's pretty much where I fall, too. Too many people using CF% as a sufficient condition to a player being/playing good. I don't get it at all.

Having a CF% of 55% is like your team outshooting the other team 17-14 while you're on the ice. What the hell are you supposed to gather from that? Making conclusive statements using terms like "outplayed" and "better" seem insane based on that knowledge. It's just part of the big puzzle.

 

[McDuffz88]

So the one thing that is really killing this team is the PK. How do we resolve this issue should be the coaches biggest concern.

 

[ChicksDigTheTrap]

So the one thing that is really killing this team is the PK. How do we resolve this issue should be the coaches biggest concern.

It is killing us this year, but I doubt it is a long term problem. Most important thing IMO is the ES play. The younger players are leading the team which is what you want to see. Future is so bright.......

 

[Triumph]

That's pretty much where I fall, too. Too many people using CF% as a sufficient condition to a player being/playing good. I don't get it at all.

Having a CF% of 55% is like your team outshooting the other team 17-14 while you're on the ice. What the hell are you supposed to gather from that? Making conclusive statements using terms like "outplayed" and "better" seem insane based on that knowledge. It's just part of the big puzzle.

In one game, a Corsi like that is meaningless. Over several games, it's not, though it is a limited bit of information to hang on. And everything else is working in Johnsson's favor in that regard too - goals scored, goals against, etc. I'd take that up against 'I remember him turning the puck over a few times several nights ago'.

The crux of all of this is that goals are semi-random events. It's very hard to score them consistently and just looking at who is on the ice for goals for and against is going to leave so much of the hockey game out. Corsi helps us find those other parts of the game, even if it is imperfect.

 

[Devils731]

So the one thing that is really killing this team is the PK. How do we resolve this issue should be the coaches biggest concern.

I think the bad news is that it’s going to be very hard.

Random luck will probably see the PK improve but looking at it, the Devils are running a diamond PK that requires a lot of shifting and position swaps. Guys need to know what their teammates are going to do without looking and there isn’t practice to get better at that, so the games are like practice.

Team can be more aware of allowing players behind their back down low in the slot if they’re covering nobody. Devils likely do expect the goalie to cut some of the passes off too. So I think these 2 areas can be improved before the others, as their individual fixes rather than team fixes.

 

[Whaddagoal]

So the one thing that is really killing this team is the PK. How do we resolve this issue should be the coaches biggest concern.

Its really hard to fathom how bad the PK is right now. -Coleman and -Greene really makes the difference. Seems hard to believe

Same (Assistant) Coach. Slight adjustment of personnel.......... and we are at bottom.....

We still have Zajac, Zacha, McLeod, Hischier.......

 

[jkrdevil]

Its really hard to fathom how bad the PK is right now. -Coleman and -Greene really makes the difference. Seems hard to believe

Same (Assistant) Coach. Slight adjustment of personnel.......... and we are at bottom.....

We still have Zajac, Zacha, McLeod, Hischier.......

while the assistant is the same, the coaching overall isn’t. The assistant coach implements the head coach’s systems. This seems to be a blind spot for Lindy.

 

[Derps]

In one game, a Corsi like that is meaningless. Over several games, it's not, though it is a limited bit of information to hang on. And everything else is working in Johnsson's favor in that regard too - goals scored, goals against, etc. I'd take that up against 'I remember him turning the puck over a few times several nights ago'.

The crux of all of this is that goals are semi-random events. It's very hard to score them consistently and just looking at who is on the ice for goals for and against is going to leave so much of the hockey game out. Corsi helps us find those other parts of the game, even if it is imperfect.

I don't really disagree. It's all part of the equation and it all has a place.

 

[Whaddagoal]

while the assistant is the same, the coaching overall isn’t. The assistant coach implements the head coach’s systems. This seems to be a blind spot for Lindy.

I know, just still hard to put together, i can't say the system looks different at the outset.

I do notice, those goals seem to be a guy that sneaks side-backdoor behind the defenseman.

Today the Buf guy snuck behind Murray (i think), A a game or so ago it was Sevs (or Kullikov?), and a game or so before that it was Vats........ All same style play. Of course we played Buffalo I think in all 3 of those games, so they're probably running same play each time. DOH. Whatever they are doing on their PP is killing our PK. Literally each PP they had we were mostly trapped in our zone with their Point man in full control.

 

[Nubmer6]

I know, just still hard to put together, i can't say the system looks different at the outset.

I do notice, those goals seem to be a guy that sneaks side-backdoor behind the defenseman.

Today the Buf guy snuck behind Murray (i think), A a game or so ago it was Sevs (or Kullikov?), and a game or so before that it was Vats........ All same style play. Of course we played Buffalo I think in all 3 of those games, so they're probably running same play each time. DOH. Whatever they are doing on their PP is killing our PK. Literally each PP they had we were mostly trapped in our zone with their Point man in full control.

I think we got scored on a few times from point shots that got tipped mid-slot earlier in the season, so we over-compensated and pulled the D men further up. Personally, I'd risk the mid-slot tip to get better coverage down low. If the opponent has that kind of hand-eye coordination and the puck manages to make it thru traffic in front, so be it.

 

[Whaddagoal]

I think we got scored on a few times from point shots that got tipped mid-slot earlier in the season, so we over-compensated and pulled the D men further up. Personally, I'd risk the mid-slot tip to get better coverage down low. If the opponent has that kind of hand-eye coordination and the puck manages to make it thru traffic in front, so be it.

The D are definitely creeping a bit higher than usual, now that you have pointed it out, it actually explains it better than anything i've thought of so far.

I agree that i'd rather keep them further back. Let's see if the coaches can figure this out over next few games. I think this team has an extra day before next one.. so better practice the hell out of these defensive reads.

 

[glenwo2]

Just wanted to repost this.


Anyway, this Weekend could be rough with the Caps coming to town on back-to-back games.

 

[Nubmer6]

Just wanted to repost this.

What... How our line shares COVID via water bottles?