Speculation: Caps Roster General Discussion (Coaching/FAs/Cap/Lines etc) - 2022-23 Season Part 2: Regular Season

Status
Not open for further replies.
So visual observation of how the player executes actions, and not just simple measurements of quantifiable metrics, are used to modify the statistics themselves?

Probably both. I imagine the authors of these models see where other models are better and look to improve their own using data from others. On the other hand a lot of these modelers are absolute freaks and watch the games themselves and test hypotheses based on what they see in game action.
 
I don’t disagree that they could use more dynamic playmakers up front but we’ve also seen situations where a team has stars and those stars go cold or get injured and there’s no depth to pick them up. Best on best this team probably falls short of some of the truly elite teams but they’re deeper and more balanced than some of the teams with great top end talent and replacement level players in the bottom six. I’d love to see them get a Timo Meier type to close the gap on top end talent.
I don't think anyone wants the team to lose depth. But we shouldn't be putting the cart before the horse. Don't over-allocate money to depth at the expense of the top six/top pairing. For most of the past 15 years, the top of the line-up was pretty much set, so it made sense to invest in depth. But now we're reaching the stage in the careers of Ovechkin, Backstrom, Carlson, and Oshie that they can't really carry the entire top of the line-up on their own anymore (and we're missing half of them on any given game anyway). So it's time to once again invest resources in shoring up the top of the line-up. Don't give up on depth, but utilize cheaper/younger (and higher upside) players as that depth rather than expensive assets.
 
Take everyone on the roster. Guys have some impressive attributes. Physicality of some sort. Or skating, or shooting.

CMM doesn't have anything of that. I didn't even see some sophisticated pass from him yet (could miss though).

even if he does some things right way that's not a given he can displace proven NHL guys which have their niche each.
 
Probably both. I imagine the authors of these models see where other models are better and look to improve their own using data from others. On the other hand a lot of these modelers are absolute freaks and watch the games themselves and test hypotheses based on what they see in game action.

Ok, so as CCR said that means "eye test" is a factor even in the fanciest of fancy stats, because the more you try to "normalize" or troubleshoot the raw data the more you have to account for actual in-game observations.

So I would be VERY SKEPTICAL of declarations made based on any of these stats, in particular those that carry fairly small differences between what's considered good and bad (fractions of a percentage, for example) since it's very possible that human error or unaccounted for factors that escape the normalization have buggered the model/data just enough to make such razor thin deltas within an undeclared margin of error.

A pattern over years that can be cited as validation of such a model might simply be a pattern of error.
 
Some real cherrypicking going on here.

This is again a stat about imaginary goals, and RAPM apparently factors GF while on ice but not GA, because of goaltender variations? How does that work ...if a shitty goaltender gives up a goal you benefit, but if your goaltender gives up a shitty goal there's no penalty. Seems like it's tilted toward offense-only players by default.

Anyway.
For better or worse I’ve gotten to the point I don’t pay attention to any stat that has a little x infront of it. I think there is a place for advanced stats but for common hockey board discussions they are usually miss used and miss understood.
 
Ok, so as CCR said that means "eye test" is a factor even in the fanciest of fancy stats, because the more you try to "normalize" or troubleshoot the raw data the more you have to account for actual in-game observations.

So I would be VERY SKEPTICAL of declarations made based on any of these stats, in particular those that carry fairly small differences between what's considered good and bad (fractions of a percentage, for example) since it's very possible that human error or unaccounted for factors that escape the normalization have buggered the model/data just enough to make such razor thin deltas within an undeclared margin of error.

A pattern over years that can be cited as validation of such a model might simply be a pattern of error.

Subjective tests such as one's visual interpretation of the game (i.e. "eye tests") certainly inform the construction of statistical models. I don't think that was ever in doubt. But without the additional step of measuring how much one's eye test is actually predictive of future goal differential said eye test is basically meaningless.
 
Subjective tests such as one's visual interpretation of the game (i.e. "eye tests") certainly inform the construction of statistical models. I don't think that was ever in doubt.

1665250424136.png


But without the additional step of measuring how much one's eye test is actually predictive of future goal differential said eye test is basically meaningless.

This is partly circular logic that completely ignores the discussion that came before it.

Something to think about: maybe we CAN'T predict future goal differentials with any real consistency. If we could do so with significant accuracy then some people would be getting very rich betting correctly on every single game.

The "eye test" is not mere observation. When applied at high professional levels it's the combination of knowledge, experience, and access to relevant information. These things are absolutely valuable in any business or pursuit.

And apparently it has so much value the fanciest of fancy stats require it.
 
This is partly circular logic that completely ignores the discussion that came before it.

Something to think about: maybe we CAN'T predict future goal differentials with any real consistency. If we could do so with significant accuracy then some people would be getting very rich betting correctly on every single game.

The "eye test" is not mere observation. When applied at high professional levels it's the combination of knowledge, experience, and access to relevant information. These things are absolutely valuable in any business or pursuit.

And apparently it has so much value the fanciest of fancy stats require it.

If we can't predict future goal differential with any real consistency then the eye test would similarly be worthless. Of course this probably isn't true, otherwise teams like Arizona, Chicago, and Philadelphia would have equal odds of winning the Stanley Cup as Colorado and Tampa Bay.
 
If we can't predict future goal differential with any real consistency then the eye test would similarly be worthless. Of course this probably isn't true, otherwise teams like Arizona, Chicago, and Philadelphia would have equal odds of winning the Stanley Cup as Colorado and Tampa Bay.

You can't always predict it, you can provide educated analysis to inform decisions that hopefully give the best chance of success.

This is how most things work.

I think eye test people fully understand this while fancy stats people think they have a pure form of data that only requires this "best available" excuse when the model fails. Not to mention the role the eye test has in the refinement of the stats.
 
Last edited:
CMM didn't act like he wanted to make the team tonight. Hard to see any way he gets a sweater over Protas and I think they're worried about Snivleys waiver status so he's #13 until Wilson comes back(assuming Oshie isn't made of paper again this year... Not looking good already). Kind of discouraging on one hand but exciting to see the rapid development of Protas on the other. Lujo stays on the team because it's sink or swim time. Irwin is Lavis man crush so he's on the team as #8 for now who can paper transaction to Hershey if needed for forward changes. I say it looks like this Monday after cuts.



Ovechkin-Kuznetsov-Brown
Mantha-Strome-Oshie
Protas-Eller-Sheary
Johansson-Dowd-Hathaway
Snivley


Fehervary-Carlson
Orlov-Jensen
Gustafsson-van Riemsdyk
Johansen
Irwin
 
  • Like
Reactions: PlushMinus
CMM didn't act like he wanted to make the team tonight. Hard to see any way he gets a sweater over Protas and I think they're worried about Snivleys waiver status so he's #13 until Wilson comes back(assuming Oshie isn't made of paper again this year... Not looking good already). Kind of discouraging on one hand but exciting to see the rapid development of Protas on the other. Lujo stays on the team because it's sink or swim time. Irwin is Lavis man crush so he's on the team as #8 for now who can paper transaction to Hershey if needed for forward changes. I say it looks like this Monday after cuts.



Ovechkin-Kuznetsov-Brown
Mantha-Strome-Oshie
Protas-Eller-Sheary
Johansson-Dowd-Hathaway
Snivley


Fehervary-Carlson
Orlov-Jensen
Gustafsson-van Riemsdyk
Johansen
Irwin

I want this line but with Protas instead of Oshie and Mantha playing RW. A real beefy line then with Mantha the sniper, Strome the playmaker and Protas the puckhound.
 
CMM didn't act like he wanted to make the team tonight. Hard to see any way he gets a sweater over Protas and I think they're worried about Snivleys waiver status so he's #13 until Wilson comes back(assuming Oshie isn't made of paper again this year... Not looking good already). Kind of discouraging on one hand but exciting to see the rapid development of Protas on the other. Lujo stays on the team because it's sink or swim time. Irwin is Lavis man crush so he's on the team as #8 for now who can paper transaction to Hershey if needed for forward changes. I say it looks like this Monday after cuts.



Ovechkin-Kuznetsov-Brown
Mantha-Strome-Oshie
Protas-Eller-Sheary
Johansson-Dowd-Hathaway
Snivley


Fehervary-Carlson
Orlov-Jensen
Gustafsson-van Riemsdyk
Johansen
Irwin
1 shot, -1 in 8+ minutes….pretty disappointing after the last game.
 

I don’t hate this. I think Leason has the potential to be a great 4th liner or good 3rd liner and I can see a team taking a shot at him, but won’t be sad to see LuJo, AJF or Borgstrom hit waivers. Not sure if any are realistic long term NHLers.

Really liked Protas last year, and although I haven’t been able to watch more than highlights of preseason (thank you NHL.tv) it sounds like he’s taken another step forwards and had a great camp.
 
No real surprises. Personally I liked LuJo the best out of the 3LD options so I’m not thrilled with that decision and I also think it’s going to be a waste to keep McMike, Protas, and Snively in DC if only one of them is going to play on a consistent basis but generally this went as expected.

LuJo and TvR seem like they’d be a steady and mobile bottom six. I didn’t really see anything with Gustafson that made me think he earned the 3LD spot but I’ll be happy to be proven wrong.
 
CMM goes from top game score in the last game to least minutes of the forwards in this past game. No wonder the kids confidence is shot. WTF
 
  • Like
Reactions: Misery74
CMM didn't act like he wanted to make the team tonight. Hard to see any way he gets a sweater over Protas and I think they're worried about Snivleys waiver status so he's #13 until Wilson comes back(assuming Oshie isn't made of paper again this year... Not looking good already). Kind of discouraging on one hand but exciting to see the rapid development of Protas on the other. Lujo stays on the team because it's sink or swim time. Irwin is Lavis man crush so he's on the team as #8 for now who can paper transaction to Hershey if needed for forward changes. I say it looks like this Monday after cuts.



Ovechkin-Kuznetsov-Brown
Mantha-Strome-Oshie
Protas-Eller-Sheary
Johansson-Dowd-Hathaway
Snivley


Fehervary-Carlson
Orlov-Jensen
Gustafsson-van Riemsdyk
Johansen
Irwin
Waive MoJo and Gus.
CMM didn't act like he wanted to make the team tonight. Hard to see any way he gets a sweater over Protas and I think they're worried about Snivleys waiver status so he's #13 until Wilson comes back(assuming Oshie isn't made of paper again this year... Not looking good already). Kind of discouraging on one hand but exciting to see the rapid development of Protas on the other. Lujo stays on the team because it's sink or swim time. Irwin is Lavis man crush so he's on the team as #8 for now who can paper transaction to Hershey if needed for forward changes. I say it looks like this Monday after cuts.



Ovechkin-Kuznetsov-Brown
Mantha-Strome-Oshie
Protas-Eller-Sheary
Johansson-Dowd-Hathaway
Snivley


Fehervary-Carlson
Orlov-Jensen
Gustafsson-van Riemsdyk
Johansen
Irwin
Waive MoJo and Gus. Would rather inject youth then play these middling vets.
 
No real surprises. Personally I liked LuJo the best out of the 3LD options so I’m not thrilled with that decision and I also think it’s going to be a waste to keep McMike, Protas, and Snively in DC if only one of them is going to play on a consistent basis but generally this went as expected.

LuJo and TvR seem like they’d be a steady and mobile bottom six. I didn’t really see anything with Gustafson that made me think he earned the 3LD spot but I’ll be happy to be proven wrong.
Agreed. Gustafsson has begun to show a little something up the ice but he's not dynamic nor an upgrade over Orlov on PP2 really. I'm not sure he's better than Nardella where you'd at least get some developing upside. Nardella may have helped play a part in being comfortable waiving Johansen. It's a super busy day so if he's going to clear the time is now. The rest is whatever.

Scratching two forwards and going with Gustafsson/Irwin at 6/7 wouldn't have been how I would've played it. But, y'know, vets.
 
  • Like
Reactions: AlexModvechkin8
Status
Not open for further replies.

Ad

Upcoming events

Ad