GDT: Avs @ Devils, 7:00 PM, MSG +

[Hisch13r]

Shitting on Schmid's four games and two periods sample size is pretty telling.

Yeah what it told is that he was really f***ing awful in those four games and two periods

 

[NJDevs26]

I mean if you want to say the advanced stats with a goalie are subjective because everyone's definition of a stoppable goal (or a high danger chance) is going to differ and not be uniform that's a fair criticism, but at least it's an honest attempt to measure something we all debate on this board daily and Bleed has his own thread for. And when the advanced stats actually agree with the eye test in saying our goalies have been awful for most of the season, there really shouldn't be much of a debate in this case.

 

[Devs3cups]

Shitting on Schmid's four games and two periods sample size is pretty telling.

Don’t know if that was meant at my post referring to Schmid being in the lower ranks of the Goals Saved Above Expected category, but it wasn’t meant to crap on him at all. I’m actually pretty excited for his development and am keeping a close eye on him in Utica. He has a lot of potential.

 

[guitarguyvic]

Amazing how many people still don’t understand the Expected Save stat literally takes into account the quality of shots. Seems like this stat would be gospel to them if they stop yelling, calm down and actually look at what that stat takes into account.

We understand. We just don't trust the subjectivity of what's being labeled HD.

 

[Tao Jersey Jones]

Don’t know if that was meant at my post referring to Schmid being in the lower ranks of the Goals Saved Above Expected category, but it wasn’t meant to crap on him at all. I’m actually pretty excited for his development and am keeping a close eye on him in Utica. He has a lot of potential.

It wasn't your post. That's old. I just don't think it's fair to throw Schmid in with Gillies and injured Blackwood without the context of how the team was playing in front of him then vs. now.

 

[MadDevil]

Ah good old HF Devils. A win. Followed by an argument. Never change.

 

[PizzaAndPucks]

Schmid and Daws aren't really NHL ready. They weren't supposed to play even 1 game in the NHL this year. As disappointing as it was to see Schmid play poorly in his few games, we also have to take into account we were playing bad during that stretch. Daws has been pretty good at the NHL level though. He has a pretty respectable record this year. He had some good games like his NHL debut and his game against the Pens where he was dominant. Ruff has been playing him lot latley , it's pretty cool to see him handle this somewhat of a starting role the way he has.

 

[NJDevs26]

Ah good old HF Devils. A win. Followed by an argument. Never change.

Yep this place is definitely living up to the meme

 

[Auto Pilot]

It’s scary how good hughes is. My devils feed was messed up so I had to watch the Avs feed. They were laughing and praising Hughes when ever he cut through their team

 

[billingtons ghost]

I’d honestly like to learn more on if/why these specific stats can’t be relied upon. Do you have sources or articles that discuss them?

High Danger ONLY takes into account position on the ice. The weakness of that should be obvious - as a two-on-zero shot off a cross ice pass from the circle is less dangerous than a weak push of a puck into the pads of the goalie. There's evidently little danger from a point shot or a screen or a high slot deflection yet they seem to go in quite a bit against very good goalies.

Expected goals is a little more sophisticated and it depends upon the many public models that are used, but as explained in alot of places, most models do not have access to pre-shot passing data. Most models don't have access to shot speed without puck tracking or AWS sports data on every shot. The attempt to get 'shot quality' firmly defined is a noble but really tough thing to do - and I think it is even harder without some subjective component from real goaltenders, or some sort of truly objective data set where you take a bazillion shots from a certain spot, with the same speed, off the same pass, with the same screen against a large number of goalies to actually compare how many goalies stop the shot. I don't know which model your graphs are using - you can look them up to see how they determine 'shot quality'.

Basically - these are nice placeholders and oversimplifications of complex systems - with the hopes that someone, someday, with real data can actually do somethng that brings real insight. For now, their nice toys for people to justify their narratives with little understanding of what the underlying causes of the goals tell us about the game.

I'm not saying they are completely worthless - especially over years of data - just not the Holy Grail of what is happening on the ice.

 

[Tao Jersey Jones]

Goaltending has been assisted by the return of Hamilton, the concurrent benching of Ty Smith, the reemergence of Shara, Mercer's slump ending,...I'm sure I'm missing someone else's improvement and contributions.

Also true,

Goaltending has assisted the return of Hamilton, the concurrent benching of Ty Smith, the reemergence of Shara, Mercer's slump ending,...I'm sure I'm missing someone else's improvement and contributions.

 

[NJDevs26]

As far as the Avs...When they went up 3-0, I remember thinking to myself this is one of the most undeserved 3-0 leads I've ever seen. Two of their goals were flukey and weak from Daws. They clearly played an extremely lackluster game, which I suspect was a combination of them playing the night before and just going through the motions while in the middle of a road trip against a team with a poor record.

That's not to take anything from the Devils...because they played great in all facets and even if the Avs were at the level we'd expect this would have still been a competitive game. And we've seen so many times over the last several years examples of good teams coming in and still kicking our ass despite putting in a half-assed effort like the Avs did today. So this win is very much a good sign.

TLDR; the Avs thought they could sleepwalk their way through this game and still beat the Devils, but they were wrong. So f*** 'em, we deserved this W.

And let's not forget us losing the captain right before faceoff when people want to give excuses to the Avs. Because once again this team's medical staff and trainers can't diagnose/manage injuries or lets them fester until they become a problem.

 

[billingtons ghost]

I mean if you want to say the advanced stats with a goalie are subjective because everyone's definition of a stoppable goal (or a high danger chance) is going to differ and not be uniform that's a fair criticism, but at least it's an honest attempt to measure something we all debate on this board daily and Bleed has his own thread for. And when the advanced stats actually agree with the eye test in saying our goalies have been awful for most of the season, there really shouldn't be much of a debate in this case.

Unless the rudimentary stats give an accurate picture of team defense along with goaltending, and that both are complicit in a goaltender's success.

I can't defend Gillies. I don't think Schmid is at all as bad as anyone says here. I don't even think Blackwood was particularly much worse than mediocre. I cannot see how giving up breakaway goals 14 seconds into two periods in a single game, or 4 ppgs against or three successive 2-on-1s is entirely on a goaltender's shoulders.

Our team has been BAD this season. Our goaltenders have been WORSE.

But the stats crowd just sees the bottom line numbers and, predictably, when the team turns in a very solid effort, even from Ty Smith, we get a silly post about how Nico Daws would've led us to the promised land, because xGF and HDC tells us so.

 

[billingtons ghost]

It’s scary how good hughes is. My devils feed was messed up so I had to watch the Avs feed. They were laughing and praising Hughes when ever he cut through their team

I enjoyed the Avs announcers - I think Peter McNab has a little love for his old squad too - they absolutely adored the way Hughes played and were effusive with their praise... especially when he walked that line behind Ty Smith and danced through for a chance.

He seems to be getting stronger by the game.

 

[Buck Dancer]

High Danger ONLY takes into account position on the ice. The weakness of that should be obvious - as a two-on-zero shot off a cross ice pass from the circle is less dangerous than a weak push of a puck into the pads of the goalie. There's evidently little danger from a point shot or a screen or a high slot deflection yet they seem to go in quite a bit against very good goalies.

Expected goals is a little more sophisticated and it depends upon the many public models that are used, but as explained in alot of places, most models do not have access to pre-shot passing data. Most models don't have access to shot speed without puck tracking or AWS sports data on every shot. The attempt to get 'shot quality' firmly defined is a noble but really tough thing to do - and I think it is even harder without some subjective component from real goaltenders, or some sort of truly objective data set where you take a bazillion shots from a certain spot, with the same speed, off the same pass, with the same screen against a large number of goalies to actually compare how many goalies stop the shot. I don't know which model your graphs are using - you can look them up to see how they determine 'shot quality'.

Basically - these are nice placeholders and oversimplifications of complex systems - with the hopes that someone, someday, with real data can actually do somethng that brings real insight. For now, their nice toys for people to justify their narratives with little understanding of what the underlying causes of the goals tell us about the game.

I'm not saying they are completely worthless - especially over years of data - just not the Holy Grail of what is happening on the ice.

Someone frame this post and cover it in gold, please!

 

[billingtons ghost]

Amazing how many people still don’t understand the Expected Save stat literally takes into account the quality of shots. Seems like this stat would be gospel to them if they stop yelling, calm down and actually look at what that stat takes into account.

What does your expected save model actually use to calculate quality of shots? If it has passing data (which it doesn't) I'll shut up. If it truly includes rebound shots and rush metrics instead of just kludging some pad for these situations (which it doesn't) I'll shut up.

By nature, most if not all of these models use logistic regression in order to grossly simplify things. They use distance from net, angle of shot, whether it is a 'snap, slap or wrist shot' (why that should matter is kinda beyond me, except that on 'average' more of X types of shots go in, I guess), and they take into account whether it was PP/ES/SS.

That's it. Shot speed? nah. Player's skill? Nah. Shot Location? Nah. Where the pass came from? One timer? cross slot?

Here's the public code for one model if you're interested.
RPubs - A New Expected Goal Model for Predicting Goals in the NHL

That's expected goals. And Expected SV is in some models inexplicably:
xGA/Fenwick.

Go figure.

 

[Bcap88]

Kids are thriving and our resident cranky pants is no where to be found. Prob in tbe woods writing a thesis on how Hughes is the worst first overall ever

 

[Devs3cups]

High Danger ONLY takes into account position on the ice. The weakness of that should be obvious - as a two-on-zero shot off a cross ice pass from the circle is less dangerous than a weak push of a puck into the pads of the goalie. There's evidently little danger from a point shot or a screen or a high slot deflection yet they seem to go in quite a bit against very good goalies.

Expected goals is a little more sophisticated and it depends upon the many public models that are used, but as explained in alot of places, most models do not have access to pre-shot passing data. Most models don't have access to shot speed without puck tracking or AWS sports data on every shot. The attempt to get 'shot quality' firmly defined is a noble but really tough thing to do - and I think it is even harder without some subjective component from real goaltenders, or some sort of truly objective data set where you take a bazillion shots from a certain spot, with the same speed, off the same pass, with the same screen against a large number of goalies to actually compare how many goalies stop the shot. I don't know which model your graphs are using - you can look them up to see how they determine 'shot quality'.

Basically - these are nice placeholders and oversimplifications of complex systems - with the hopes that someone, someday, with real data can actually do somethng that brings real insight. For now, their nice toys for people to justify their narratives with little understanding of what the underlying causes of the goals tell us about the game.

I'm not saying they are completely worthless - especially over years of data - just not the Holy Grail of what is happening on the ice.

Thanks for the info!

For your information, most analytics in regards to GSAx are posted from JFresh, which takes its data from TopDownHockey.

Here are the explanations in regards to his GSAx model and the variables it takes into account, etc.: A New Expected Goal Model That is Better Than Corsi at Predicting Future Goals

EDIT: Twitter thread from the creator explaining a little also:

 

[Brodeur]

Forgot to post the Newfoundland showdown, but I guess there's always the rematch next month:

 

[pulkit1989]

It’s actually starting to look like he might be back. The poor goaltending is looking like it’s gonna give him a stay of execution. Fitzy knows the goaltending sucks as much as we do and has publicly acknowledged as such.

I was not a fan of the Ruff hire, but I don’t think he’s really worse now than he ever was.

The team scores quite a bit for him at 5 on 5.

I do think if he’s back it’s probably playoffs or bust for him. If we miss for any reason next year, even if half the team is injured then I think he’s not coaching a second past next season. And that’s what I originally thought earlier in the year before we really got in trouble and it looked like he might be gone at any minute. Fitzy’s public vote on confidence for him around Christmas is usually what you hear right before a coach is fired.

If we have a record in December like we had this season in December then he probably don’t make it to New Years or maybe even earlier.

i wonder if Kevin Deneen will step up to NHL and coach the devils. He’s done a good job with Utica and is familiar with the Devils’ vision and culture

 

[Guttersniped]

Yes but have you considered that the Devils are somehow uniquely bad and far worse defensively than everyone else and those numbers don't adequately portray us?

Also we aren’t the worst defensive team or the sloppiest or most problematic etc.

You can argue with the fine details here but just every public statistical, as well as the ol’ eye test, can find teams that are doing worse defensively and have better goaltending because… they have <drum roll> NHL goaltenders.

At this point I’m not bothering to even provide a single shred of stats for this. For one thing I’m going to bed soon but also if people don’t just don’t want to believe goalie is a position that skill makes a difference, then… ok. Sure.

And the “there’s elite goalies but good goalies don’t exist because reasons” argument … ok. Sure. Teams endlessly invest in this position for the occasional elite and to pass the time I guess.

I can’t stress enough that no one is saying the defense is great or has been rock solid though out the year. But if our barely average at best and sometimes worse than that defense was better it wouldn’t have saved these goalies from themselves in many of the losses.

Finally who the f*** cares how many games would have been won vs lost in the alternate universe with healthy average goalies? We don’t live there.

It also doesn’t sound like we’re changing coaches either so if you were hoping we box out guys in front of the net well or protect the middle anytime soon… ok. Sure.

 

[Capt Nico Poo]

Watched highlights only. At 3-0 i thought we were getting a reality check by the #1 team in the league.

Someone mentioned after the Rags loss that this team is showing spirit and future is looking better. I think its well said, still few years away from contending but the key guys seem to do well.

 

[Devils Army]

Thought zacha was fine tonight honestly. You guys are making him the whipping boy even in a win. Thought he was good on the PP as well. 2/4 made some good passes.

 

[Better Call Sal]

Forgot to post the Newfoundland showdown, but I guess there's always the rematch next month:

Knows, Dawson, knows!

 

[Clam Jensen]

Ah good old HF Devils. A win. Followed by an argument. Never change.

Holy shit I fell asleep with everyone vibing and wake up to BG ranting past midnight. Good lord.