The Advanced Stats Thread Episode VII: An Ode to the Sanity of Silverfish

[GeorgeKaplan]

Got one for Yandle?

 

[silverfish]

Got one for Yandle?

Leopold - 61 (10-11)
Leopold - 59 (11-12)
Willie Mitchell - 51 (09-10)
Mike Green - 48 (16-17)
Kris Russell - 48 (17-18) UH OH

Smell test - fail

 

[GeorgeKaplan]

Leopold - 61 (10-11)
Leopold - 59 (11-12)
Willie Mitchell - 51 (09-10)
Mike Green - 48 (16-17)
Kris Russell - 48 (17-18) UH OH

Smell test - fail

I wish I remembered what hockey writer/person said it, but I have a memory of someone saying Yandle was having a Norris caliber year this year

 

[Machinehead]

After an awful regular season by analytics standards, the Capitals are outplaying their opposition in the playoffs.

How did we get here?

 

[GeorgeKaplan]

After an awful regular season by analytics standards, the Capitals are outplaying their opposition in the playoffs.

How did we get here?

Maybe they took a page from the Kings’ regular season excuse guide and they were just #SavingItForThePlayoffs

 

[Machinehead]

Maybe they took a page from the Kings’ regular season excuse guide and they were just #SavingItForThePlayoffs

Those Kings teams were some of the best regular season analytics teams ever.

 

[GeorgeKaplan]

Those Kings teams were some of the best regular season analytics teams ever.

Fair enough, I’ve just always hated that reasoning as to why they’re always on the outside looking in for the majority of every seasons playoff race. Replace them with like, any Penguins season before December

 

[PlamsUnlimited]

After an awful regular season by analytics standards, the Capitals are outplaying their opposition in the playoffs.

How did we get here?

Anomalies

 

[Irishguy42]

After an awful regular season by analytics standards, the Capitals are outplaying their opposition in the playoffs.

How did we get here?

Welcome to the Stanley Cup Playoffs. Where the reffing is made up and the sample sizes don't matter.

 

[silverfish]

Pionk blog up. Am I allowed to link to my website here? I feel like no. Anyway, y'all reading this thread know where it is. But the biggest item up for debate is the below. These are the metrics and weights for each metric that I'm using. Ridicule the f*** out of this. Let's make it better.

What's weighted too much? What's weighted too little? What's missing? What's here but shouldn't be? Come after me.

Individual:
P160 - 3
iCF60 - 3
ixGF60 - 3

Relative:
relTCF60 - 2
relTCA60 - 2
relTxGF60 - 3
relTxGA60 - 3
relxFSh - 1
relxFSv - 1

Usage/Context:
TOI/GP - 1
relZSR - 1
xGFQoT - 1
xGFQoC - 1

 

[Irishguy42]

Pionk blog up. Am I allowed to link to my website here? I feel like no. Anyway, y'all reading this thread know where it is. But the biggest item up for debate is the below. These are the metrics and weights for each metric that I'm using. Ridicule the **** out of this. Let's make it better.

What's weighted too much? What's weighted too little? What's missing? What's here but shouldn't be? Come after me.

Individual:
P160 - 3
iCF60 - 3
ixGF60 - 3

Relative:
relTCF60 - 2
relTCA60 - 2
relTxGF60 - 3
relTxGA60 - 3
relxFSh - 1
relxFSv - 1

Usage/Context:
TOI/GP - 1
relZSR - 1
xGFQoT - 1
xGFQoC - 1

Is your weighting scale only 1, 2, or 3?

 

[silverfish]

Is your weighting scale only 1, 2, or 3?

No, good question! Can be any number, but they need to total 25.

 

[SA16]

Is ixGF60 expected goal rate for that individual player or for the team while he’s on the ice? If it’s individual (which it seems to be) I feel like you’re double counting stuff with also including shot attempt rate (should be a pretty direct correlation) and point rate.

 

[silverfish]

Is ixGF60 expected goal rate for that individual player or for the team while he’s on the ice? If it’s individual (which it seems to be) I feel like you’re double counting stuff with also including shot attempt rate (should be a pretty direct correlation) and point rate.

Individual. I wanted individual attempts for quantity, individual attempts for quality, and then points. They are definitely related, but I think they serve different purposes. What do you think?

 

[silverfish]

So what are the Caps? Sample sizes be damned. They didn't really do anything at the deadline, right? They added Kempny, who I like a lot, a week before...

via Corsica

Pre-deadline, 5v5 unadjusted: 48.09 CF% | 45.79 xGF%
Post-deadline, 5v5 unadjusted: 47.67 CF% | 50.49 xGF% (20 gp)

Playoffs, 5v5 unadjusted: 49.89 CF% | 52.35 xGF% (14 gp)

The change in xGF% is noticeable. What are they doing differently?

Top-players get an uptick in icetime starting around game 60: WSH 1718 F Icetime
Kempny added, Orpik falls, Carlson sees a reduction: WSH 1718 D Icetime

 

[JimmyG89]

Kempny has taken tons of ES TOI away from Orpik. He's the clear cut #4 on this roster. Niskanen, Carlson, and Orlov get used pretty much the same, all logging 20+ per game. Kempny is around 17 minutes. Orpik and Djoos are under 13.

Would have to look at TOI the last 20 games, but not having Orpik out there is big for Washington. He's in a role for him to succeed (3rd pair, penalty killer).

That trade is very similar to the Kevin Klein trade. It stabilized their defense. Oddly enough, if Washington was smarter, they could have:

Orlov-Niskanen
Schmidt-Carlsson
Kempny-Djoos

Instead, they kept Orpik last June.

 

[silverfish]

Kempny has taken tons of ES TOI away from Orpik. He's the clear cut #4 on this roster. Niskanen, Carlson, and Orlov get used pretty much the same, all logging 20+ per game. Kempny is around 17 minutes. Orpik and Djoos are under 13.

Would have to look at TOI the last 20 games, but not having Orpik out there is big for Washington. He's in a role for him to succeed (3rd pair, penalty killer).

That trade is very similar to the Kevin Klein trade. It stabilized their defense. Oddly enough, if Washington was smarter, they could have:

Orlov-Niskanen
Schmidt-Carlsson
Kempny-Djoos

Instead, they kept Orpik last June.

Well, yeah, that's why I included the hockeyviz d-men ice-time

Orpik was declining in TOI/gp well before Kempny came to town. Trotz got woke.

 

[aufheben]

Thought this was kinda funny lol. @silverfish

 

[DanielBrassard]

LOL. That's something else.

 

[silverfish]

Thought this was kinda funny lol. @silverfish

Josh Gorges wtf?

Not even gonna mention Heed and Franson bc I am not surprised.

 

[DanielBrassard]

Josh Gorges wtf?

Not even gonna mention Heed and Franson bc I am not surprised.

Lovejoy and Gustafsson too is weird.

 

[YoSoyLalo]

also our good hated friend john gilmour

 

[JimmyG89]

Well, yeah, that's why I included the hockeyviz d-men ice-time

Orpik was declining in TOI/gp well before Kempny came to town. Trotz got woke.

That is overall TIO. They weren't taking Orpik's SH TOI away. They were taking his ES TOI away. That graph shows overall. That was my only comment.

 

[silverfish]

That is overall TIO. They weren't taking Orpik's SH TOI away. They were taking his ES TOI away. That graph shows overall. That was my only comment.

Well, right, a dip like that generally displays a dip in ES TOI because there aren't enough special teams minutes to impact that large.

But it feels like we're arguing the same point here

 

[Machinehead]

If the Caps got better because they replaced garage with an analytics darling... I'll take that!