Speculation: Caps Roster General Discussion (Coaching/FAs/Cap/Lines etc) - 2023 Off-season

Status
Not open for further replies.

CapitalsCupReality

It’s Go Time!!
Feb 27, 2002
66,483
21,512
If there was a much easier way for me to do my job that required much less work and improved my performance, I'm not sure I would be telling my boss about it either :D

I do think the point about institutional change being a slow beast is fair. Baseball teams didn't copy Billy Beane instantly, they adopted over time. The real question I would love to know, that we'll likely never ever know, is which NHL teams are at the top of weighting analytics vs at the bottom, so we can compare their results with data.

Twabby's point being that there is data that says a certain analytical method appears to be better than league wide current results, shouldn't be dismissed just because we don't know the weights any teams are using. And organizational fit can never really be answered by raw stats, so scouting and human interactions are always going to be a part of the whole process.

I’ve already acknowledged the best teams use both methodologies to identify good prospects. And sure, if one team has a math nerd who has found a better formula to deduce it all, then yeah, that’s called IP and who shares that freely? But it would be neat to know which teams use all this information the best.

The only thing being dismissed is the notion you can simply sort a spreadsheet of points and boom you’re going to dominate the draft, and hell you don’t even have to know the game of hockey! Lol…
 
  • Like
Reactions: kicksavedave

kicksavedave

I'm just here for the memes and gifs.
Sponsor
Apr 29, 2009
11,397
14,588
Fallbrook, CA
www.tiasarms.org
I’ve already acknowledged the best teams use both methodologies to identify good prospects. And sure, if one team has a math nerd who has found a better formula to deduce it all, then yeah, that’s called IP and who shares that freely? But it would be neat to know which teams use all this information the best.

The only thing being dismissed is the notion you can simply sort a spreadsheet of points and boom you’re going to dominate the draft, and hell you don’t even have to know the game of hockey! Lol…


Well everyone knows, if we knew the game, we'd be in it. :D

mcphee_george640.jpg
 

YippieKaey

How you gonna do hockey like that?
Apr 2, 2012
3,022
2,563
Stockholm Sweden
I evaluate players based on cool names, if they're fun in interviews and if they play in an entertaining way. I really liked that little agro dude we had on the 4th line who played with a big edge and had a cool name but then he turned out to be a turd unfortunately. What was that guys name? Leipsic?
 
  • Haha
Reactions: racingmoose

HTFN

Registered User
Feb 8, 2009
12,566
11,486
I evaluate players based on cool names, if they're fun in interviews and if they play in an entertaining way. I really liked that little agro dude we had on the 4th line who played with a big edge and had a cool name but then he turned out to be a turd unfortunately. What was that guys name? Leipsic?
That one shift he had where he absolutely freight trained a guy with like 40 pounds on him was awesome, but f*** that guy.

Outside of the big reasons, when you're making fun of your teammates and their choice of wife because she's not "attractive enough" to be a hockey wife (because you don't know what love is) you can f*** out of the room. Nobody needs that.
 

twabby

Registered User
Mar 9, 2010
14,198
15,788
But part of their argument is that the bad teams are all just not using them enough.

There's no room for, say, a team using analytics a whole lot but having a bad/different model and not seeing the results twabby is advocating for... A different statistician could value a different stat and make a compelling enough case that they take the "wrong" player, but the framework of twabby's argument would blame old-head hockey scouts or somebody else getting in the way because it's a flawed perspective with no available insight for us to refute the claim.

Usually in situations like these twabby would have the burden of proof, but it's never proven that these picks fail due to archaic scouting practices... it's asserted pretty arrogantly that all the failures must be due to archaic scouting practices because they failed. We need to know the balances teams are coming in with if we want to take this argument seriously, or else it's just cherry picking and reverse engineering the position to suit the premise.

Shit, we were just talking about this a few pages ago. Building backwards from the conclusion is bad science, but there's no room in this discussion for there to be another conclusion for some people. Even when statistics fail, they're just "not perfect yet" but still infallibly better than anything humans do... that's zealotry, not open discussion.

You’re suggesting the reason why current drafting performs worse than a public analytic model is that analytics teams are relied on too heavily?

Feel free to believe that I guess. There’s no proof one way or another but it feels like a slap in the face to William of Ockham.
 

HTFN

Registered User
Feb 8, 2009
12,566
11,486
You’re suggesting the reason why current drafting performs worse than a public analytic model is that analytics teams are relied on too heavily?

Feel free to believe that I guess. There’s no proof one way or another but it feels like a slap in the face to William of Ockham.
No, I'm suggesting you need to stop playing word games and re-read the post, but I'll make it simpler.

Remember when the Coyotes brought in a young, analytics driven GM and said they were going to basically moneyball the shit out of things and change the direction of the team? How's that going now? Where's the draft success, exactly?

You can commit to an analytical approach, still hire the wrong guys or follow the wrong model, and get f*** all for it. This is a team that was very vocal about this stance, too, so we have no reason to believe the balance between analytics and scouting wasn't heavily tipped in the favor of analytics.

Your current argument asserts that teams fail because they put too much value on traditional scouting, but here's a team that did no better while claiming to fully embrace analytics. Your argument lumps these teams in to make a point, but fails to recognize that they did what you wanted, just not how you wanted, and it didn't work. There are teams who have likely tried way more than you give them credit to find an analytical model and still came up short in the draft, but you simply don't give that possibility any space to exist.


I'm not saying it's all teams, this isn't a game of spin and the argument you're making here is really disingenuous. Until you can prove without a doubt that underfunded analytics departments correlates to draft failure, you have no real case here, and one fun wrinkle to that is the post a few pages ago that notes that the Capitals exceeded their expected WAR while having a very small and seldom updated analytics group. Meanwhile Ross Mahoney's been running and crushing drafts at various stages of his career, was shown to outperform the spreadsheets, has worked through various phases of the game, and that guy's... what, not good because he's not an algorithm? Do you really not appreciate what a scout has done for this franchise?
 

twabby

Registered User
Mar 9, 2010
14,198
15,788
No, I'm suggesting you need to stop playing word games and re-read the post, but I'll make it simpler.

Remember when the Coyotes brought in a young, analytics driven GM and said they were going to basically moneyball the shit out of things and change the direction of the team? How's that going now? Where's the draft success, exactly?

You can commit to an analytical approach, still hire the wrong guys or follow the wrong model, and get f*** all for it. This is a team that was very vocal about this stance, too, so we have no reason to believe the balance between analytics and scouting wasn't heavily tipped in the favor of analytics.

Your current argument asserts that teams fail because they put too much value on traditional scouting, but here's a team that did no better while claiming to fully embrace analytics. Your argument lumps these teams in to make a point, but fails to recognize that they did what you wanted, just not how you wanted, and it didn't work. There are teams who have likely tried way more than you give them credit to find an analytical model and still came up short in the draft, but you simply don't give that possibility any space to exist.


I'm not saying it's all teams, this isn't a game of spin and the argument you're making here is really disingenuous. Until you can prove without a doubt that underfunded analytics departments correlates to draft failure, you have no real case here, and one fun wrinkle to that is the post a few pages ago that notes that the Capitals exceeded their expected WAR while having a very small and seldom updated analytics group. Meanwhile Ross Mahoney's been running and crushing drafts at various stages of his career, was shown to outperform the spreadsheets, has worked through various phases of the game, and that guy's... what, not good because he's not an algorithm? Do you really not appreciate what a scout has done for this franchise?

John Chayka is even worse than a Hockey Man™. He’s a businessman.

His background was not in analytics at all but rather in Stathletes, a microstats company that he founded. He used his business and sales acumen to convince the Coyotes that he had some great approach that would revolutionize hockey.

I’ve consistently expressed my skepticism of microstats because they are completely descriptive and have yet to show value for predictive analysis. They’re just neat numbers to look at. Like hits or takeaways.

So sure, maybe a few organizations make too many decisions based on microstats. I could buy that. But microstats are just quantified eye-tests. They’re not analytics.
 

Langway

In den Wolken
Jul 7, 2006
32,994
10,206
You’re suggesting the reason why current drafting performs worse than a public analytic model is that analytics teams are relied on too heavily?
You repeat this an awful lot. What are the sources? All I'm seeing is a Reddit post and this:
But a number of questions arise. For one, taking goaltenders out of the equation. That's easier perhaps but not realistic. Most crucially I think judging performance by WAR essentially always devalues the fringe player and doesn't properly assess the value of complete busts. Complete busts contribute 0 WAR yet that's actually preferable to a negative WAR NHL player. Yet which is the superior selection? The bust that never sniffs action? Or the theoretical sub-replacement level player that somehow keeps earning a jersey and finds some way to earn trust and carve out a career? There's a general competitive element that such an isolated stats approach is bound to bog down an organization if left unchecked. Organizations can't afford to view value through a singular lens, even if it happens to be true within a certain limited framework. The Stat to End All Stats is doomed to failure because there's inevitably too much nuance and context it can't capture. Teams aren't built piece by piece irrespective of the whole. If it were just about stacking WAR why not just dress 18 high quality wingers? Etc. Absurd. Eventually that becomes a corrosive system because if it can't be calculated it doesn't exist. Its limitations are unaccounted for. A more pluralistic stance is absolutely necessary to avoid the pitfalls of believing It's All Been Figured Out. Adopt that mentality and...good luck.

Just looking at HockeyProspecting's first rounds from 2011-16 it's not immediately apparent as the superior method. It hits very nicely in 2011 between Kucherov, Gaudreau and others in the first round vs. not but it's the only year it appears to clearly outperform in the first round field vs. reality. 2016 is also pretty solid thanks to DeBrincat and Fox but always with a high bust rate that's essentially written off. It doesn't appear to be so uniformly dominant as to elevate it to such heights. (Maybe I'm missing something.) Teams aren't going to be comfortable with such a wide variance in outcomes. They're going to mitigate risk to a far greater extent. They have staffs. Of course they're going to use them and piece together their own preference toward optimizing decisions. Not penalizing drastic whiffs whatsoever stands out as a pretty severe model limitation. A backward-looking model that actually incentivizes being more wrong seems to me to be kind of hard to follow without a great deal of skepticism.
 
Last edited:

YippieKaey

How you gonna do hockey like that?
Apr 2, 2012
3,022
2,563
Stockholm Sweden
That one shift he had where he absolutely freight trained a guy with like 40 pounds on him was awesome, but f*** that guy.

Outside of the big reasons, when you're making fun of your teammates and their choice of wife because she's not "attractive enough" to be a hockey wife (because you don't know what love is) you can f*** out of the room. Nobody needs that.

Yeah i remember that, and he was quick and pretty good hands too. But yeah, f him for sure. Karma rightfully ruined his career i guess.
 

HTFN

Registered User
Feb 8, 2009
12,566
11,486
John Chayka is even worse than a Hockey Man™. He’s a businessman.

His background was not in analytics at all but rather in Stathletes, a microstats company that he founded. He used his business and sales acumen to convince the Coyotes that he had some great approach that would revolutionize hockey.

I’ve consistently expressed my skepticism of microstats because they are completely descriptive and have yet to show value for predictive analysis. They’re just neat numbers to look at. Like hits or takeaways.

So sure, maybe a few organizations make too many decisions based on microstats. I could buy that. But microstats are just quantified eye-tests. They’re not analytics.
And here we go. They say "we're going to take an analytics defined approach", you don't like the analytics they used (not that you actually know what was applied at all) and then find a way to compartmentalize and minimize the point because it's not something you personally see value in.... which was the point in the first place. You don't know what teams are doing, what analytics they do or don't value, hardly anything because teams don't give that information out easily. You don't know if a team is 100% analog scouting just because their GM has some buzzwords for the interview. These are all things you've assumed about teams you think you know to prove a point you can't possibly prove.

Until you can tell us which teams are doing what with which models (many secret and proprietary) this is just howling at the wind. You're arguing backwards from the conclusion.
 

CapitalsCupReality

It’s Go Time!!
Feb 27, 2002
66,483
21,512
Err ….. not to be a downer or anything, but don’t we already HAVE a fancy stat thread? I certainly don’t mind a few comments, but this is now hijacking the thread from discussing our team and our players too ……fancy stats.

Please have mercy!!!
Oh stop it’s fine…at least we are not being spammed by spreadsheets. This discussion is about talent evaluation. Stats are a part of that process…
 

HTFN

Registered User
Feb 8, 2009
12,566
11,486
It’s been multiple pages of arguing fancy stats vs the eye test/knowledge of the game.

What’s the point of having a thread for it then?
Because that's a haven for it. If we were having that conversation in there, there wouldn't really be a safe place to post them on their own merits.

That thread is a great place to theorize and post things for numbers related feedback, but some people like to bring that methodology unchecked out into the streets and that's when it gets all... well...

spongebob-how-many-times-do-we-need-to-teach-you.gif


It's just a healthy learning experience
 

Jags

Mildly Disturbed
May 5, 2016
1,989
2,398
Central Florida
It hits very nicely in 2011 between Kucherov, Gaudreau and others in the first round vs. not but it's the only year it appears to clearly outperform in the first round field vs. reality. 2016 is also pretty solid thanks to DeBrincat and Fox but always with a high bust rate that's essentially written off.

I'm of the opinion that some of these models are designed (at least in part) to explain the Kucherovs, Guentzels, DeBrincats, and Gaudreaus of the world. And of course they should be; that's kinda the point. So they reverse engineer a model that values those players by whatever other metrics they find, then they try to use that model predictively and it's as shitty as the old way, just different.

Now maybe I'm wrong or oversimplifying because most of what I just wrote was conjecture, but it feels obvious to me. If there were some metric (especially a singular one) that nailed past draft classes and then yielded similar, inarguable results predictively, that nerd would be very rich and very famous. No one could keep that secret.

People arguing in that direction are simply buying magic beans. There is zero chance that most teams haven't looked at all models against past drafts and predictively and analyzed the results. No one would ignore something so simple that would give them such a massive advantage. It doesn't exist. You're gonna end up with a giant beanstalk in your backyard and your neighbors are gonna call the HOA.
 

twabby

Registered User
Mar 9, 2010
14,198
15,788
You repeat this an awful lot. What are the sources? All I'm seeing is a Reddit post and this:
But a number of questions arise. For one, taking goaltenders out of the equation. That's easier perhaps but not realistic. Most crucially I think judging performance by WAR essentially always devalues the fringe player and doesn't properly assess the value of complete busts. Complete busts contribute 0 WAR yet that's actually preferable to a negative WAR NHL player. Yet which is the superior selection? The bust that never sniffs action? Or the theoretical sub-replacement level player that somehow keeps earning a jersey and finds some way to earn trust and carve out a career? There's a general competitive element that such an isolated stats approach is bound to bog down an organization if left unchecked. Organizations can't afford to view value through a singular lens, even if it happens to be true within a certain limited framework. The Stat to End All Stats is doomed to failure because there's inevitably too much nuance and context it can't capture. Teams aren't built piece by piece irrespective of the whole. If it were just about stacking WAR why not just dress 18 high quality wingers? Etc. Absurd. Eventually that becomes a corrosive system because if it can't be calculated it doesn't exist. Its limitations are unaccounted for. A more pluralistic stance is absolutely necessary to avoid the pitfalls of believing It's All Been Figured Out. Adopt that mentality and...good luck.

Just looking at HockeyProspecting's first rounds from 2011-16 it's not immediately apparent as the superior method. It hits very nicely in 2011 between Kucherov, Gaudreau and others in the first round vs. not but it's the only year it appears to clearly outperform in the first round field vs. reality. 2016 is also pretty solid thanks to DeBrincat and Fox but always with a high bust rate that's essentially written off. It doesn't appear to be so uniformly dominant as to elevate it to such heights. (Maybe I'm missing something.) Teams aren't going to be comfortable with such a wide variance in outcomes. They're going to mitigate risk to a far greater extent. They have staffs. Of course they're going to use them and piece together their own preference toward optimizing decisions. Not penalizing drastic whiffs whatsoever stands out as a pretty severe model limitation. A backward-looking model that actually incentivizes being more wrong seems to me to be kind of hard to follow without a great deal of skepticism.

Posted earlier:


I’m particular high NHLe correlates strongly with overperforming a draft spot while lower NHLe correlates strongly with underperforming.

There was another article that had this spelled out more clearly but I can’t for the life of me find it. Regardless, the above article has an equivalent conclusion.

The goaltender point is a fair one. Throw in some goalie scouts I guess since there isn’t a good model for them yet.

I guess I don’t really care about fringe players. Boom or bust is a much more fruitful strategy IMO because the middling players are the ones you can get in trades and free agency. We’re seeing the results of a team with little high-end talent yet some good depth, and it’s not great!

Penalizing whiffs is fine but you have to penalize missing out on elite talents as well, and that’s where current scouting fails much more. I’d rather get 1 elite talent and 6 whiffs than 3 or 4 fringe to decent players and 3 or 4 whiffs. The rest can be filled in later. There will always be Sonny Milanos or Nick Jensens or Nic Dowds available for the low low cost of money.

And again you are completely mischaracterizing my position. NHLe hasn’t figured everything out. It’s just better than whatever nonsense is going on now!
 

HTFN

Registered User
Feb 8, 2009
12,566
11,486
Posted earlier:


I’m particular high NHLe correlates strongly with overperforming a draft spot while lower NHLe correlates strongly with underperforming.

There was another article that had this spelled out more clearly but I can’t for the life of me find it. Regardless, the above article has an equivalent conclusion.

The goaltender point is a fair one. Throw in some goalie scouts I guess since there isn’t a good model for them yet.

I guess I don’t really care about fringe players. Boom or bust is a much more fruitful strategy IMO because the middling players are the ones you can get in trades and free agency. We’re seeing the results of a team with little high-end talent yet some good depth, and it’s not great!

Penalizing whiffs is fine but you have to penalize missing out on elite talents as well, and that’s where current scouting fails much more.

And again you are completely mischaracterizing my position. NHLe hasn’t figured everything out. It’s just better than whatever nonsense is going on now!
Wow.

We're not seeing the results of a team with little high end talent, we're seeing high end talent at age 35... Now, maybe that's the same to you, because it's just numbers on a board, but it's completely ignoring a decade+ of high caliber elite talent and warping the framework of how the Capitals draft.

Bad, bad take.
 

twabby

Registered User
Mar 9, 2010
14,198
15,788
Wow.

We're not seeing the results of a team with little high end talent, we're seeing high end talent at age 35... Now, maybe that's the same to you, because it's just numbers on a board, but it's completely ignoring a decade+ of high caliber elite talent and warping the framework of how the Capitals draft.

Bad, bad take.

The team is very bottom heavy is what I’m saying and they are having a hard time acquiring high-end pieces to make them competitive. They’ve had no trouble filling out the bottom of their roster with good, budget acquisitions on the market. So maybe don’t try to draft those types when higher reward targets are still on the board.
 

HTFN

Registered User
Feb 8, 2009
12,566
11,486
The team is very bottom heavy is what I’m saying and they are having a hard time acquiring high-end pieces to make them competitive. They’ve had no trouble filling out the bottom of their roster with good, budget acquisitions on the market. So maybe don’t try to draft those types when higher reward targets are still on the board.
But they haven't recently, so... yeah, I think they agree. Miro, Suzdalev, Cristall are all swings on top of the safe success in Leonard. They aren't the only ones either, Cam Allen and some other late round picks show they're rolling for upside.

The problem is contracts that can't move. Why do you think GMBM looked so frustrated about Backstrom down the stretch and in the offseason? He wants that money to take the swing you want, but it's tied up. Some of that goes away this year, more goes away next year.

If you don't think he knows that you're fooling yourself, he's just not as willing as you to make a desperate trade and lose on value to finalize one run even if it buries them after.
 

twabby

Registered User
Mar 9, 2010
14,198
15,788
But they haven't recently, so... yeah, I think they agree. Miro, Suzdalev, Cristall are all swings on top of the safe success in Leonard. They aren't the only ones either, Cam Allen and some other late round picks show they're rolling for upside.

The problem is contracts that can't move. Why do you think GMBM looked so frustrated about Backstrom down the stretch and in the offseason? He wants that money to take the swing you want, but it's tied up. Some of that goes away this year, more goes away next year.

If you don't think he knows that you're fooling yourself, he's just not as willing as you to make a desperate trade and lose on value to finalize one run even if it buries them after.

I was mainly responding to @Langway saying the model didn’t pick as many fringe players as the current draft does, to which I say I don’t care because fringe players are a dime a dozen.

Also while Cristall was a great pick I’m still not sold on Leonard being great value at #8 when Benson was on the board for instance.
 

Ridley Simon

HFBoards Sponsor
Sponsor
Feb 27, 2002
19,094
10,453
Marin County — SF Bay Area, CA
I was mainly responding to @Langway saying the model didn’t pick as many fringe players as the current draft does, to which I say I don’t care because fringe players are a dime a dozen.

Also while Cristall was a great pick I’m still not sold on Leonard being great value at #8 when Benson was on the board for instance.
Benson over Leonard!!… said no one, as we approached the actual draft.
 
Status
Not open for further replies.

Ad

Upcoming events

Ad

Ad