HFBoards

Go Back   HFBoards > General Hockey Discussion > By The Numbers
Mobile Hockey's Future Become a Sponsor Site Rules Support Forum vBookie Page 2
Notices

By The Numbers Hockey Analytics... the Final Frontier. Explore strange new worlds, to seek out new algorithms, to boldly go where no one has gone before.

High danger scoring chances

Reply
 
Thread Tools
Old
05-28-2016, 01:20 AM
  #1
irishock
Fire Ken Holland
 
Join Date: Jan 2012
Location: Vancouver, BC
Country: Canada
Posts: 605
vCash: 50
High danger scoring chances

Do you guys know where I can get data for high danger scoring chances per game? Thanks in advance

irishock is offline   Reply With Quote
Old
05-28-2016, 01:37 AM
  #2
ElfanuReinhard*
Curry Ironhook
 
Join Date: Mar 2016
Location: Tuanorse, Ardhglaur
Country: North Korea
Posts: 733
vCash: 500
Quote:
Originally Posted by irishock View Post
Do you guys know where I can get data for high danger scoring chances per game? Thanks in advance
Hmm I'm not too sure about it per game, but I know that you can get per game by goaltender on http://war-on-ice.com/goalietable.html, but the thing is high danger scoring chance is skewed by absolute numbers. What I mean is it's much easier to face a high danger scoring chance that is visible and slow then it is a medium danger scoring chance while you're heavily screened.

ElfanuReinhard* is offline   Reply With Quote
Old
06-21-2016, 03:57 PM
  #3
pokerface1
Registered User
 
Join Date: Sep 2010
Posts: 203
vCash: 500
with war-on-ice offline, does anyone know of a currently operating website with sortable hd sv%,md sv%, ld sv% ?

pokerface1 is offline   Reply With Quote
Old
06-27-2016, 10:21 AM
  #4
eperry
Registered User
 
Join Date: Jun 2016
Posts: 42
vCash: 500
Quote:
Originally Posted by pokerface1 View Post
with war-on-ice offline, does anyone know of a currently operating website with sortable hd sv%,md sv%, ld sv% ?
I'm hoping to add this to Corsica over the Summer. It's been the most common request since day one.

eperry is offline   Reply With Quote
Old
07-01-2016, 07:53 PM
  #5
jc17
Registered User
 
jc17's Avatar
 
Join Date: Jun 2013
Posts: 1,643
vCash: 90
Check out sporting charts.

Won't give you a list like war-on-ice did but you can compare shot charts for 2 players at a time.

jc17 is offline   Reply With Quote
Old
07-20-2016, 05:25 AM
  #6
pokerface1
Registered User
 
Join Date: Sep 2010
Posts: 203
vCash: 500
Quote:
Originally Posted by eperry View Post
I'm hoping to add this to Corsica over the Summer. It's been the most common request since day one.
looking forward to it!

Quote:
Originally Posted by jc17 View Post
Check out sporting charts.

Won't give you a list like war-on-ice did but you can compare shot charts for 2 players at a time.
thanks for the suggestion-I'll definitely take a look.

pokerface1 is offline   Reply With Quote
Old
07-20-2016, 04:42 PM
  #7
oilerbear
Registered User
 
oilerbear's Avatar
 
Join Date: Jun 2008
Posts: 2,826
vCash: 500
Quote:
Originally Posted by eperry View Post
I'm hoping to add this to Corsica over the Summer. It's been the most common request since day one.
As the guy who saw langes Shot quality curves in 2006.
I recognized importance of the Bubble in the curves near the net.
It is a shame he did not recognize the HSCA area.

I had been looking for work that related to a concept that stuck in my brain since I first saw the Flin Flon bombers of Clarke; Leach; Stoughton.

The violence was about :
go to the net
Defend the net.

I was never going to be able to Present the databases that we see now.

But anyone who's been on lowetide's site know's the concepts I have fought up river to defend for the last 10 years.

1. Situational expected performance ranges for
-Competition based on Even production
- Teamates
- ZS

Behind the net was what made me consider this approach.
to get expected range of
CF; SF; GF;
CA; SA; GA:
Cdiff; S diff; Gdiff
for any given player.

the 3 values are broken up into 864 groups:
on 3 axis graph.

not much long after Rob Vollman came up with a single basic 2 D graph
were you could
Pick you 2 axis
Comp; team; ZS
results were displayed as a color based on neg or positive results with increased bubble amplitude.

Most use Comp/ZS
And look at Corsi
Which excludes team affect.

it does not tell you what a players performance is compared to expected.
If you take the 96 groups.
1st;2nd; 3rd; 4th comp - 4variables
1st; 2nd; 3rd; 4th comp - 4 variables
Zone starts broken up by STnd devations - 6 variables.

4 X 4 X 6 = 96

you take all the Forawrd and Dman and can establish an average for a given situation.
Adam Larson;
1st comp; 1st team; 31% ZS
Expected GD -17


You can further break each group into a series of groups that forams a cube around the situational average.
Comp Upr; mid; lower - 3 variables
Team Upr; mid; Lower - 3 variables
3 X 3 = 9 groups
UPR C - UPR T; UPR C - MID T; UPR C - LWR T
MID C - UPR T; MID C - MID T; MID C - LWR T
LWR C - UPR T; LWR C - MID T; LWR C - LWR T

So we can take Larsson further situation
Upr 1st comp; LWR 1st team; 31% ZS
Datas average says
expected Goal diff is -24

Expected range for U1st C; lwr 1st T ; 30% ZS
is -17 to -24

Larsson was +15 Last year
32 to 41 Goal diff better than expected.

I would like to see the Corsi, Shot; Gola data further broken down into
Low scoring chance area shots.
High scoring chance Area shots.

Now to my point!

War on ice established there line in the sand. For HSCA shots.
I loved seeing there site.
But comparing their Database results to my Extensive years of manual stripping of data results.

I realized they used the wrong line in the sand.
I discussed this on lowetides site and 2 days later they changed it.

You want your low scoring chance area to be all shots that are below League average shooting %.
0.0 to 8.5%


You want your your High scoring chance area to be all shots above league average
shooting %. 8.5 to 20%


There is still the most important factor affecting GF and GA.
I have to continue to fight up river defending.

There are Posters on Lowetides site Woodguy/Gmoney that are breaking down players measure
for the levels of comp Faced based on Standard deviations and presenting them as expected Dangerous Fenwick.
understanding that HSCA is the critical area of the game.

it largely ignores teammates and ZS.
and grouping is different.

But does tell who is strong or weak against diffrent levels of comp.

Which takes it to a much more simplified presentation like Vollman's.

your site is amazing:
But your work was flawed the instant you took out shooting affect.

Players like eberle who have that flin flon bomber tough.
Stone to go to HSCA area.
deliver an elite shooting %

Your data needs that LSC area and HSC area separation.


Last edited by oilerbear: 07-20-2016 at 04:59 PM.
oilerbear is offline   Reply With Quote
Old
07-21-2016, 12:39 AM
  #8
eperry
Registered User
 
Join Date: Jun 2016
Posts: 42
vCash: 500
Quote:
Originally Posted by oilerbear View Post
As the guy who saw langes Shot quality curves in 2006.
I recognized importance of the Bubble in the curves near the net.
It is a shame he did not recognize the HSCA area.

I had been looking for work that related to a concept that stuck in my brain since I first saw the Flin Flon bombers of Clarke; Leach; Stoughton.

The violence was about :
go to the net
Defend the net.

I was never going to be able to Present the databases that we see now.

But anyone who's been on lowetide's site know's the concepts I have fought up river to defend for the last 10 years.

1. Situational expected performance ranges for
-Competition based on Even production
- Teamates
- ZS

Behind the net was what made me consider this approach.
to get expected range of
CF; SF; GF;
CA; SA; GA:
Cdiff; S diff; Gdiff
for any given player.

the 3 values are broken up into 864 groups:
on 3 axis graph.

not much long after Rob Vollman came up with a single basic 2 D graph
were you could
Pick you 2 axis
Comp; team; ZS
results were displayed as a color based on neg or positive results with increased bubble amplitude.

Most use Comp/ZS
And look at Corsi
Which excludes team affect.

it does not tell you what a players performance is compared to expected.
If you take the 96 groups.
1st;2nd; 3rd; 4th comp - 4variables
1st; 2nd; 3rd; 4th comp - 4 variables
Zone starts broken up by STnd devations - 6 variables.

4 X 4 X 6 = 96

you take all the Forawrd and Dman and can establish an average for a given situation.
Adam Larson;
1st comp; 1st team; 31% ZS
Expected GD -17


You can further break each group into a series of groups that forams a cube around the situational average.
Comp Upr; mid; lower - 3 variables
Team Upr; mid; Lower - 3 variables
3 X 3 = 9 groups
UPR C - UPR T; UPR C - MID T; UPR C - LWR T
MID C - UPR T; MID C - MID T; MID C - LWR T
LWR C - UPR T; LWR C - MID T; LWR C - LWR T

So we can take Larsson further situation
Upr 1st comp; LWR 1st team; 31% ZS
Datas average says
expected Goal diff is -24

Expected range for U1st C; lwr 1st T ; 30% ZS
is -17 to -24

Larsson was +15 Last year
32 to 41 Goal diff better than expected.

I would like to see the Corsi, Shot; Gola data further broken down into
Low scoring chance area shots.
High scoring chance Area shots.

Now to my point!

War on ice established there line in the sand. For HSCA shots.
I loved seeing there site.
But comparing their Database results to my Extensive years of manual stripping of data results.

I realized they used the wrong line in the sand.
I discussed this on lowetides site and 2 days later they changed it.

You want your low scoring chance area to be all shots that are below League average shooting %.
0.0 to 8.5%


You want your your High scoring chance area to be all shots above league average
shooting %. 8.5 to 20%


There is still the most important factor affecting GF and GA.
I have to continue to fight up river defending.

There are Posters on Lowetides site Woodguy/Gmoney that are breaking down players measure
for the levels of comp Faced based on Standard deviations and presenting them as expected Dangerous Fenwick.
understanding that HSCA is the critical area of the game.

it largely ignores teammates and ZS.
and grouping is different.

But does tell who is strong or weak against diffrent levels of comp.

Which takes it to a much more simplified presentation like Vollman's.

your site is amazing:
But your work was flawed the instant you took out shooting affect.

Players like eberle who have that flin flon bomber tough.
Stone to go to HSCA area.
deliver an elite shooting %

Your data needs that LSC area and HSC area separation.
I'm sorry, I'm really trying to understand what is is you're trying to say and I just can't. You say my work is flawed and I would like to understand how that is in a manner that is comprehensible.

eperry is offline   Reply With Quote
Old
07-24-2016, 02:22 AM
  #9
oilerbear
Registered User
 
oilerbear's Avatar
 
Join Date: Jun 2008
Posts: 2,826
vCash: 500
Quote:
Originally Posted by eperry View Post
I'm sorry, I'm really trying to understand what is is you're trying to say and I just can't. You say my work is flawed and I would like to understand how that is in a manner that is comprehensible.
the work no!

I refer everyone I talk with re Analytics to your site.

your data is correct.

the game is situational.

So your info will suggest some players to be better than others.
When they are not.

you present you data.
and
your site and data development is the best available.

desjardins first presented
Comp; teammates; ZS measures.

you can establish
96 combinations of comp (4); teammates (4); ZS by std deviation (6)
so you can get an expected avg result for the 96 situations.

in any of the data you share.
but it needs the situational average to be a measure versus performance.

to take it further.
there should be an expected range for each of those groups broken up into a more specialized combination.
comp and team needs to be Upr, Mid, Lower to establish a situation range for the more specialized case.

it allows you to measure players versus there situational pears.
otherwise you are saying a player who
faces
upr 1st comp
with
lower 3rd teammates
in a
36% ZS

has equal challenges to a

Player
who faces
bottom 4th comp
with
upr 2nd teamates
in a
57% zs

we know this not to be true.

you state you ignore these situational factors.

there are further factors that create LSCA and HSCA results
that most of the analytics community are not willing to accept at this point.

i struggle to convince lowetides site related to these factors.

1. just like situational results as a baseline.

2. D had an influence on shot results.

3. LSCA and HSCA

you are skilled enough to take your data base to the next level.

oilerbear is offline   Reply With Quote
Old
07-24-2016, 01:00 PM
  #10
eperry
Registered User
 
Join Date: Jun 2016
Posts: 42
vCash: 500
Quote:
Originally Posted by oilerbear View Post
you state you ignore these situational factors.
I did no such thing. That contextual factors are not included in raw xGF% does not mean I ignore usage effects altogether. Perhaps you missed my score and zone-start adjustment method? I wrote about it here.

My goals added metric K is pending review, but it accounts for all the factors you mentioned and more.

eperry is offline   Reply With Quote
Old
07-27-2016, 05:09 AM
  #11
Neo1978
Registered User
 
Join Date: Aug 2015
Posts: 125
vCash: 500
Quote:
Originally Posted by irishock View Post
Do you guys know where I can get data for high danger scoring chances per game? Thanks in advance
There isn't any that I am aware of and there shouldn't be to be honest. There simply isn't any data available that is accurate enough to be useful. I have tracked quality shots manually for the Blackhawks the last two seasons. The data coming from the NHL site is so wrong it is a joke.

Average shot locations from each stadium vary dramatically. Some sites put everything much closer to the net than others. Even different official scorers from the same stadium are radically different. Individual shots in the same game vary greatly as some shots are located from the beginning of the shot and some from the follow through.

And the overall accuracy itself is mostly random. Shots directly in front of the net are routinely measured in the 20 foot range. And something that shows up in the 30 foot range can be a shot from the point. Until there is tracking in uniforms, this data will produce random results.

Neo1978 is offline   Reply With Quote
Old
08-02-2016, 02:02 PM
  #12
JaegerDice
Q Gonna Q
 
JaegerDice's Avatar
 
Join Date: Dec 2014
Posts: 12,428
vCash: 500
FYI, according to their twitter, naturalstattrick will also be adding scoring chance data to their site.

JaegerDice is offline   Reply With Quote
Old
08-05-2016, 07:29 PM
  #13
jc17
Registered User
 
jc17's Avatar
 
Join Date: Jun 2013
Posts: 1,643
vCash: 90
Quote:
Originally Posted by JaegerDice View Post
FYI, according to their twitter, naturalstattrick will also be adding scoring chance data to their site.
Hallelujah!

jc17 is offline   Reply With Quote
Old
08-08-2016, 12:34 PM
  #14
Irishguy42
Registered User
 
Irishguy42's Avatar
 
Join Date: Sep 2015
Location: Upstate NY Wildernes
Country: United States
Posts: 6,569
vCash: 500
Corsica.hockey just added L/M/H danger scoring chances to their website.

Irishguy42 is online now   Reply With Quote
Old
08-09-2016, 04:46 PM
  #15
Love
Fire Benning
 
Love's Avatar
 
Join Date: Feb 2012
Posts: 8,077
vCash: 500
Quote:
Originally Posted by Irishguy42 View Post
Corsica.hockey just added L/M/H danger scoring chances to their website.
Can you send me a link for this? Im trying to find it but can't seem to locate it on the site

Love is offline   Reply With Quote
Old
08-09-2016, 04:57 PM
  #16
Irishguy42
Registered User
 
Irishguy42's Avatar
 
Join Date: Sep 2015
Location: Upstate NY Wildernes
Country: United States
Posts: 6,569
vCash: 500
Quote:
Originally Posted by Love View Post
Can you send me a link for this? Im trying to find it but can't seem to locate it on the site
Meant to split that up, sorry. My mistake.

They added L/M/H Sv% splits for goalies, and Scoring Chances stats to skaters.

Added GSAA for goalies as well.

You will have to scroll sideways to find it for both.

Irishguy42 is online now   Reply With Quote
Old
08-10-2016, 01:07 PM
  #17
triggrman
Registered User
 
triggrman's Avatar
 
Join Date: May 2002
Location: Murfreesboro, TN
Country: United States
Posts: 22,254
vCash: 557
Quote:
Originally Posted by Irishguy42 View Post
Meant to split that up, sorry. My mistake.

They added L/M/H Sv% splits for goalies, and Scoring Chances stats to skaters.

Added GSAA for goalies as well.

You will have to scroll sideways to find it for both.
I finally found a stat for defense I like, thank you..


SCA/60 tells a pretty good story...

triggrman is offline   Reply With Quote
Old
08-10-2016, 04:11 PM
  #18
eperry
Registered User
 
Join Date: Jun 2016
Posts: 42
vCash: 500
Quote:
Originally Posted by triggrman View Post
I finally found a stat for defense I like, thank you..


SCA/60 tells a pretty good story...
Adjusted Rel xGA/60 is probably a better choice.

eperry is offline   Reply With Quote
Old
08-10-2016, 10:36 PM
  #19
Canadiens1958
Registered User
 
Canadiens1958's Avatar
 
Join Date: Nov 2007
Posts: 13,813
vCash: 500
Further

Quote:
Originally Posted by Neo1978 View Post
There isn't any that I am aware of and there shouldn't be to be honest. There simply isn't any data available that is accurate enough to be useful. I have tracked quality shots manually for the Blackhawks the last two seasons. The data coming from the NHL site is so wrong it is a joke.

Average shot locations from each stadium vary dramatically. Some sites put everything much closer to the net than others. Even different official scorers from the same stadium are radically different. Individual shots in the same game vary greatly as some shots are located from the beginning of the shot and some from the follow through.

And the overall accuracy itself is mostly random. Shots directly in front of the net are routinely measured in the 20 foot range. And something that shows up in the 30 foot range can be a shot from the point. Until there is tracking in uniforms, this data will produce random results.
With or without access to specific game strategies?

Canadiens1958 is offline   Reply With Quote
Reply

Forum Jump


Bookmarks

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off



All times are GMT -5. The time now is 07:34 AM.

monitoring_string = "e4251c93e2ba248d29da988d93bf5144"

vBulletin Copyright ©2000 - 2016, Jelsoft Enterprises Ltd.
HFBoards.com is a property of CraveOnline Media, LLC, an Evolve Media, LLC company. 2016 All Rights Reserved.