2018-19 stats and underlying metrics thread

mcpw

WPG
Jan 13, 2015
10,024
2,072
Could be right. I'll be interested to see how this develops. Seems that a lot of crappy teams are running up big shot totals while some good teams are valuing possession more. The game last night was a good example. The Jets tortured the Devils in the 3rd while NJ racked up a Corsi advantage.

Honestly, this can be mathematically proven. Take a pair of correlated random variables (each binomially distributed), use a very small sample (like n<100) of one variable (which results in a huge variance = n/4), and watch the correlation fade on average. You can basically take any 15-ish set of games and watch the same effect.

About crappy teams: we'll need a classifier of what teams are supposed to be crappy (unless you mean by eye test, and yes, the Devils don't look good right now). Let's focus on four data points in the graph I posted (early 17-18 results):
46% CF / 52% GF --- Jets
50% CF / 35% GF --- Penguins
52% CF / 58% GF --- Blues
56% CF / 51% GF --- Hurricanes
Which teams are the crappy ones? Clearly the answer is: can't tell from such a small sample size.
 

Whileee

Registered User
May 29, 2010
46,061
33,043
Honestly, this can be mathematically proven. Take a pair of correlated random variables (each binomially distributed), use a very small sample (like n<100) of one variable (which results in a huge variance = n/4), and watch the correlation fade on average. You can basically take any 15-ish set of games and watch the same effect.

About crappy teams: we'll need a classifier of what teams are supposed to be crappy (unless you mean by eye test, and yes, the Devils don't look good right now). Let's focus on four data points in the graph I posted (early 17-18 results):
46% CF / 52% GF --- Jets
50% CF / 35% GF --- Penguins
52% CF / 58% GF --- Blues
56% CF / 51% GF --- Hurricanes
Which teams are the crappy ones? Clearly the answer is: can't tell from such a small sample size.
Yeah, I think it's a function of sample size, as you've shown. I didn't do the math. It's not just the low sample size but also the narrow range for goal differentials that creates statistical instability.

Still, I remain a bit skeptical about the huge shot totals that some teams are putting up. The Canes and Panthers might be the best examples, along with the Yotes and Hawks. Maybe there are just a few teams that focus on high shot volumes. There clearly are some stylistic differences.
 

Saidin

Registered User
Mar 18, 2015
1,250
1,043
upload_2018-11-19_11-18-21.png


I post this because of where I saw Copp.
 

garret9

AKA#VitoCorrelationi
Mar 31, 2012
21,738
4,380
Vancouver
www.hockey-graphs.com
Yeah, I think it's a function of sample size, as you've shown. I didn't do the math. It's not just the low sample size but also the narrow range for goal differentials that creates statistical instability.

Still, I remain a bit skeptical about the huge shot totals that some teams are putting up. The Canes and Panthers might be the best examples, along with the Yotes and Hawks. Maybe there are just a few teams that focus on high shot volumes. There clearly are some stylistic differences.

A Panther scout once told me the only player they think that tries to game Corsi or ignore shot volume for shot quality is one of their worst shot differential players.
 

Whileee

Registered User
May 29, 2010
46,061
33,043
A Panther scout once told me the only player they think that tries to game Corsi or ignore shot volume for shot quality is one of their worst shot differential players.
I don't really think that individual players can have that much influence on their on-ice shot metrics, for obvious reasons. However, I do think that some coaches / teams employ systems that focus on shot volume and others focus more on offensive zone puck possession and shot quality. That goes for both offensive and defensive systems. Some teams (like the Jets), clearly focus their defense on pushing play to the sides of the rink and clogging the middle / slot, but allowing more shooting opportunities from outside. On their PK, Maurice has been explicit about that strategy, with shot volume / positions that result from that. Conversely, the Jets often cycle in the offensive zone building zone time before they create opportunities for shots. They don't always result in prime scoring chances, but watching the game you can see the difference in styles between say the Hurricanes or Devils and the Jets. Over the past 2+ seasons the Canes are #1 in CF/60, #22 in GF/60 and #26 in GF%.

I also think that there are vagaries in the assessment of xGF, though I must admit I haven't delved into the methodologies enough (I wonder if there are any good public, peer-reviewed articles on this). An example... last game Ehlers had a shot at a wide open net from the deep slot near the faceoff circle. He missed the net, but I'd bet that he scores on about 50% of those chances. In the same game, a Sabres player had a shot from the exact same location, but it didn't come after a lot of movement and cycling and Brossoit was in position to easily make the save. I'd guess the chances of scoring on that sort of play with the goalie set would be around 5% or less. Both shots came after a cross-ice pass, produced a wrist shot from the same location, but one was perhaps 5-10 times more likely to result in a goal (xG = 0.5 vs. 0.02, or so).

This is a long-winded way of saying that I think that although the statistical associations are quite robust in general, there are vagaries that probably require better study to understand outliers. I'm sure there are some insightful statisticians looking at this as we speak. :thumbu:
 

garret9

AKA#VitoCorrelationi
Mar 31, 2012
21,738
4,380
Vancouver
www.hockey-graphs.com
I don't really think that individual players can have that much influence on their on-ice shot metrics, for obvious reasons. However, I do think that some coaches / teams employ systems that focus on shot volume and others focus more on offensive zone puck possession and shot quality. That goes for both offensive and defensive systems. Some teams (like the Jets), clearly focus their defense on pushing play to the sides of the rink and clogging the middle / slot, but allowing more shooting opportunities from outside. On their PK, Maurice has been explicit about that strategy, with shot volume / positions that result from that. Conversely, the Jets often cycle in the offensive zone building zone time before they create opportunities for shots. They don't always result in prime scoring chances, but watching the game you can see the difference in styles between say the Hurricanes or Devils and the Jets. Over the past 2+ seasons the Canes are #1 in CF/60, #22 in GF/60 and #26 in GF%.

I also think that there are vagaries in the assessment of xGF, though I must admit I haven't delved into the methodologies enough (I wonder if there are any good public, peer-reviewed articles on this). An example... last game Ehlers had a shot at a wide open net from the deep slot near the faceoff circle. He missed the net, but I'd bet that he scores on about 50% of those chances. In the same game, a Sabres player had a shot from the exact same location, but it didn't come after a lot of movement and cycling and Brossoit was in position to easily make the save. I'd guess the chances of scoring on that sort of play with the goalie set would be around 5% or less. Both shots came after a cross-ice pass, produced a wrist shot from the same location, but one was perhaps 5-10 times more likely to result in a goal (xG = 0.5 vs. 0.02, or so).

This is a long-winded way of saying that I think that although the statistical associations are quite robust in general, there are vagaries that probably require better study to understand outliers. I'm sure there are some insightful statisticians looking at this as we speak. :thumbu:

I just brought it up because you said Florida and I thought it was funny a funny story about Florida.
 

Maukkis

EZ4ENCE
Mar 16, 2016
10,538
7,218
Those who seem to have targetted Hellebuyck as the easy scapegoat for the early season struggles, here's something for y'all.

2017-18, 5v5 numbers

SV%: 93.07
xSV%: 92.92
GSAA: 2.16

2018-19

SV%: 91.06
xSV%: 91.93
GSAA: -2.15


You might point out that Helle's numbers have gone down a lot, but so has the quality of the team's defense. If you account for special teams, Hellebuyck has overperformed his xSV this year.

What have we learned from this?

1) Hellebuyck needs to be better.
2) Anybody who belonged to the 'won't miss Enstrom' -crowd was terribly wrong.
3) We have a team-wide problem with regards to icing a competent defensive structure to help Hellebuyck.

And before anybody goes there, Brossoit and his three games are largely irrelevant. Sample sizes are fun and all that, I know.
Update:

SV%: 90.96
xSV%: 91.77

You guys had better not point fingers at Hellebuyck, when the team in front of him is dropping the ball just as much. Funnily enough, Hellebuyck is carrying the dysfunctional PK. SV% of 90.29 on xSV% of 86.14.

Also, did anyone else notice that 'league average goaltending' is at .909 this year? Much lower than previously.
 

kylbaz

Winnipeg <3
Nov 14, 2015
4,894
4,808
www.movingtowinnipeg.ca
Update:

SV%: 90.96
xSV%: 91.77

You guys had better not point fingers at Hellebuyck, when the team in front of him is dropping the ball just as much. Funnily enough, Hellebuyck is carrying the dysfunctional PK. SV% of 90.29 on xSV% of 86.14.

Also, did anyone else notice that 'league average goaltending' is at .909 this year? Much lower than previously.

Keep feeding them fancy stats pretending he's playing good
 

Ducky10

Searching for Mark Scheifele
Nov 14, 2014
19,809
31,386
Nope. I mean look how good he's doing in the NHL right now with those facts .
Lol, you keep acting as if that means something. Enstrom would be a big help to how the Jets are leaking more chances from danger areas. Your dislike of him does not make it untrue.
 

Mud Turtle

Registered User
Jul 26, 2013
8,058
18,234
Garret, what are your thoughts on Poolman? Should he be up with the Jets on the third pairing?
 

ps241

The Ballad of Ville Bobby
Sponsor
Mar 10, 2010
34,831
30,981
Update:

SV%: 90.96
xSV%: 91.77

You guys had better not point fingers at Hellebuyck, when the team in front of him is dropping the ball just as much. Funnily enough, Hellebuyck is carrying the dysfunctional PK. SV% of 90.29 on xSV% of 86.14.

Also, did anyone else notice that 'league average goaltending' is at .909 this year? Much lower than previously.

No I missed that thank you. I wonder if it’s their new equipment or perhaps that teams tend to play a bit more open early in seasons?
 

ps241

The Ballad of Ville Bobby
Sponsor
Mar 10, 2010
34,831
30,981
Please God I hope you don't mean enstrom



:banghead::tmi:



Warning: this is not a “fancy stat” it’s an actual heat map that shows where the shots are coming from but more importantly where shots are not coming from.

I have an issue with Logan Stanley.....if someone showed me this same heat map with and without him on the ice over a large enough sample size I would admit my eyes are lying to me and Logan is very good at protecting the house and keeping the shots to the outside.

So far this season our goalie can be better but our team needs to get way better defensively too.
 
  • Like
Reactions: Ducky10

kylbaz

Winnipeg <3
Nov 14, 2015
4,894
4,808
www.movingtowinnipeg.ca
It was well reported that Enstrom received NHL offers, but none were enough to take him away from home and his other half (who has been fighting cancer).

So there is those facts.
Yes probably in the million dollar range, made the decision much easier. No one was offering him several million like you claim his worth to be
 

garret9

AKA#VitoCorrelationi
Mar 31, 2012
21,738
4,380
Vancouver
www.hockey-graphs.com
Yes probably in the million dollar range, made the decision much easier. No one was offering him several million like you claim his worth to be

1) So? My point still stands regardless.

2) I didn’t claim any worth. Worth in the NHL is very... context driven, amongst other things.

Age of player, depth in that position, depth in that role, how close to the 50 contract limit, is the team rebuilding or competing, location of team, marketability of player, player’s leverage, etc.

There’s a lot beyond just player’s impact on wins that goes into contract value. The NHL player market is highly inefficient in determining value (point totals drives value more than impact on wins), and even then it’s hard to determine how much a player will go for.
 

Ad

Upcoming events

Ad

Ad

-->