• Moderators, please send me a PM if you are unable to access mod permissions. Thanks, Habsy.

Leafs' Prospect/Marlies Discussion Thread!

BeLeafer

resurrected
Stat is no good for individual players but okay for teams? How does that work, exactly? I distinctly recall you finding it problematic because it isolates defense and offense and the two are supposedly indistinguishable in hockey.
 

zeke

Well-known member
Stat is no good for individual players but okay for teams? How does that work, exactly? I distinctly recall you finding it problematic because it isolates defense and offense and the two are supposedly indistinguishable in hockey.

2 issues:


1. using xga and xgf on their own is always incomplete, for teams and individuals. a team or individual can choose to hunker down and play 100% defense and take no offensive chances to lower their xga, but that doesn't actually mean they are necessarily better defensively. a team or individual can go all out 100% attacking with no D to get a higher xgf but that doesnt mean they are necessarily better offensively than a team that produces moderately less xgf but does it in a system that allows significantly less xga. So its always incomplete. the true sign of a good team or player is how much they outproduce what they allow - regardless of what kind of system or style they play. that's a huge deal by itself that makes uaing the number in isolation fairly useless unless you're trying to describe a very specific result rather than describe actual quality, but at least that's the extent of the issue when it comes to using them individually for teams - but not so for individuals.

2. teams as a whole have the same situational use and quality of competition as each other. they all play all the roles against all the competition in all the different usages. individuals, on the other hand, play in vastly different roles with vastly different usage against drastically different quality of competition with drastically different teammates in different systems. and all that is piled ON TOP of the already incomplete nature of using only one half of the equation to describe a player's offense or defense. it is totally unsurprising that a guy who plays against elite tough qoc on one of the worst xga teams in the league would end up with one of the worst xga in the league. this doesn't actually tell us he is bad defensively though.
 

LeafOfFaith

Well-known member
Oh shit, guyssssss! One of our prospects turned 23 today. Matty, I miss the days of your having only potential. Yesterday, all my troubles seemed so far away.
 

MindzEye

Wayward Ditch Pig
According to xGF%, Nick Holden is the best defensemen in the NHL.

Me: Usage matters
You: but how do you account for it
Me: (produces examples of analytics darlings in mediocre usage who became pretty neutral xGF% guys when facing top competition)
You: But how, maybe my math nerd can figure it out
Me: Throws feces


There, I just saved us both an hour, you're welcome.
 

BeLeafer

resurrected
2 issues:


1. So its always incomplete. the true sign of a good team or player is how much they outproduce what they allow - regardless of what kind of system or style they play. that's a huge deal by itself that makes uaing the number in isolation fairly useless unless you're trying to describe a very specific result rather than describe actual quality, but at least that's the extent of the issue when it comes to using them individually for teams - but not so for individuals.

2. teams as a whole have the same situational use and quality of competition as each other. they all play all the roles against all the competition in all the different usages. individuals, on the other hand, play in vastly different roles with vastly different usage against drastically different quality of competition with drastically different teammates in different systems. and all that is piled ON TOP of the already incomplete nature of using only one half of the equation to describe a player's offense or defense. it is totally unsurprising that a guy who plays against elite tough qoc on one of the worst xga teams in the league would end up with one of the worst xga in the league. this doesn't actually tell us he is bad defensively though.

Didn't have time for this word soup yesterday but ...

1. You talk about xGA/60 being in isolation, as if that's something unique about it. xGF% is an isolated stat -- it measures shot location. Why is that 'isolation' okay? Further, there's nothing isolated about it. It's one half of what goes into xGF%. This is akin to saying that we don't like scoring chance measurement because it is isolated from shots. It's nonsense.

2. This is true for every single player stat used. Nothing suggests that usage is uniquely troublesome for xGA. xGA is not incomplete; it's expected goals against. It's only 'incomplete' if you make a fetish of xGF%.
 

zeke

Well-known member
Didn't have time for this word soup yesterday but ...

1. You talk about xGA/60 being in isolation, as if that's something unique about it. xGF% is an isolated stat -- it measures shot location. Why is that 'isolation' okay? Further, there's nothing isolated about it. It's one half of what goes into xGF%. This is akin to saying that we don't like scoring chance measurement because it is isolated from shots. It's nonsense.

2. This is true for every single player stat used. Nothing suggests that usage is uniquely troublesome for xGA. xGA is not incomplete; it's expected goals against. It's only 'incomplete' if you make a fetish of xGF%.

I don't understand why this is so tough for you.

Just because a soft usage guy on a team that plays a trap no-offense style has a great xga doesn't actually tell us he's good defensively.

Using them alone for team stats is bad enough - it shouldn't be done without referencing what that team is able to do on the opposite of the ice as well.

But for individual players it becomes noisy beyond reason.
 

BeLeafer

resurrected
Nonesense is always tough to swallow. All individual stats are affected by team play. Just because you have a stat that measures both the offensive and defensive dimensions does not erase this affect or even mitigate it.
 

zeke

Well-known member
we've gone through examples of this before.

I can try again.

Here's one example of a fairly big minute defenseman who went from one good team to another good team, both that played very different styles, and played approximately the same minutes and role - Justin Faulk.

Last 3 years:

CAR: 53.5xgf% (#5) ----> 2.76xgf (#1), 2.40xga (#16)
STL: 52.4xgf% (#7) -----> 2.38xgf (#18), 2.16xga (#3)

So, 2 very good teams, with opposite styles of play.

Faulk went from similar role to similar role:

CAR '18: 17:46evtoi, -3.4dzdiff,
CAR '19: 18:48evtoi, -1.7dzdiff
STL '20: 18:23evtoi, -0.9dzdiff

And put up similar xgf%:

CAR '18: 51.6% (0.0rel)
CAR '19: 51.9% (-0.4rel)
STL '20: 51.2% (0.0rel)

But saw a significant change in the individual elements:

CAR '18: 2.67xgf, 2.51xga
CAR '19: 2.78xgf, 2.58xga
STL '20: 2.35xgf, 2.24xga


Are we actually supposed to conclude that Faulk became significantly worse offensively and better defensively in equal measure upon switching teams, or should we make the much more obvious conclusion?
 

MindzEye

Wayward Ditch Pig
 
Top