Hey everyone, I've just created a chart for 2016-2017 season high danger impact by defenseman. The top 3 defenseman (subjective) of every team that has played at least 750 minutes 5 on 5 is used. For Anaheim, Carolina, St. Louis, and Nashville I decided to use their top 4. If it's not very easy for you to navigate, please open the image in a new tab to get the full view, and if it's still hard, just tell me what I should change in the comments below.
Which players positions on the chart surprised you the most?
Relative in this case means the impact that the respective defenseman have on their team's high danger chances when they are on the ice. So for example Braun has a strong negative impact on his team's overall high danger chances generation when he's on the ice, but has a strong positive impact on his team's overall high danger chance suppression. For high danger chances against, having a negative number means better as you're allowing that much less chances when you're on the ice than when you're not. For high danger chances for, having a positive number (Obviously) means better for the vice versa reason of high danger chances against.
Relative isn't by the overall situations, it's by when the said player is not on the ice. For example, Pesce's high danger generation when he is on the ice is 11.86 per 60. The amount of high danger chances he allows is 8.41. The amount of high danger chances he suppresses is 3.51 when he's on the ice versus when he's not (Aka a negative 3.51 on the chart, as in this scenario the lower in the number the better), while he generates 0.22 more when he's on the ice than when he's not. Carolina as a whole gives up 10.67 HDCA and produces 11.72 HDCF. Pesce's relative numbers (Well all players really) are not indicated by his team's overall numbers. Relative means when he's on the ice versus when he's not; how much is he impacting high danger chances.
As the creator of high danger theory 10+ years ago.
I spent much time trying to have people get it.
I have been asked to generate 5 books on my specific theories for academic use.
I finally have te free time to Do my ork.
Plus I am tired of lking at people us portions of my theories
and
coming to disgustingly awful analytical conclusions.
Thier is suttle portions to the affected theory that sites like Corsica Hockey and Puck IQ did not get.
One of the theories I developed is Situational means and ranges for Sets of Data.
Corsi; Goals; ... etc.
You have to look at each situation and mean based on Comp Team and ZS.
I finally came to 3 groups of 8
Upper, Lower 1st; 2nd; 3rd; 4th comp and Team
8 std deviations of ZS.
you end up with a data cube 8 high x 8 wide x 8 deep = 512 groups of means and ranges.
An example would be Goal differential:
GF - GA = GD
The variance in mean by the 3 axis situations was +29 to -30 goal dif.
Rob Volman: later created a 2 axis version of my chart.
which greatly scewed results.
He allowed you to look at portions of the cube.
allowing you to make it 1st line comp rather than upper or 1st+ 2nd comp
which gave you a partial reference to mean.
Which end up Statistical lies.
A full usage chart needs to deal with the 3 variables a coach can control.
Team; Comp; Zone start (with and without procession coming off the bench as part of Zone start affect)
So Volman excludes one axis.
A simple test to see if that is a smart move is:
we should be able to hold the 2 displayed variables constant.
work the length axis for the excluded variable.
if its exclusion is to have little affect the Mean and Range for the excluded ariabl should not vary results.
The removal of any of the 3 axis does not pass the smell test.
2 axis usage charts are a lie.
People in the analytics world want to simplify it for the masses.
but exsclusion often leads to Analytic delusion.
Most of my theories was breaking down the defensive aspect of the game.
As the person who created what you are looking at.
Corsi for and Corsi against is a reflection of Fwd and Offensive Dman Success or Failure.
Their is No defensive Measure of the attacking players Starting from procession to Pocession change.
When I looked at trying to capture Elite HD dmen I found that Corsi to GA (90%)
or Fenwick to GA (95%) were the best 2 captures of the 60 worst and Best HD Dmen.
! looked at individual
Groups:
Corsi to Fenwick 43.7%
Fenwick to SA 47.7%
SA to GA 85%
Corsi to Fenwick to SA 45%
Fenwick to Sa to GA (95%)
Corsi to fewick to Sa to GA (90%)
What became clear right away is
1. their are 2 ways to affect Play defensively.
2. you must see what is the hardest situational play a Dman can repeat his performance. ( Belichek way) You do not look at ceilng play.
The 2 ways to affect Defence.
1. HD reduction:
Take a cummulative total of Shot scoring Potential based on x,y Corsi
2. Open Hole shot suppression:
looking at the best Dman that have the lowest % of Corsi be a open hole shot.
What this is really saying is have the highest % of Corsi have 0% chance of going in the net.
We know what Corsi; Blocks and Misses are.
Closed hole shot: is what I call a table hockey goalie shot.
In the game today golaies move side to side just like a table hockey Goalie.
Their is a large % Shots that hit the goalie.
They had 0% chance of going in.
1. Shot into glove before movement ; then the arm swing
2. hits the pads
3. Hits the shoulder
4. Hits the Chest protector
5. hits the blocker
6. Off the helmet.
So an elite Open hole Shot suppression Dman.has a Standard rate
Open hole shots = Corsi - ( blocks + misses + closed hole shots)
that rate is;
Open hole shots/ corsi
You want them to Repeat that level of Perormance for as high a % of EVTOI.
This 2nd measure is how you Define Dmen like Kris Russell.
one of the 2 best in the league.
So sorry Volman's Player usage charts tell you nothing!
Analytic People trying to come up with a More simple number to define player are being Silly.
So remember:
Forwards:
CF x,y to CA x,y establishes the entry to turnover HD affect of forward play.
THe hd aspect of that must be included.
Dmen:
I do not look at Dman offence because all but 5-10 are upper 4th to #14 press box forward even offence.
All that really matters is the Baseline EVGA Mean that a Dpair can establish.,
Cause if you want to be Playoff Wild card competitive.
The forward line has to generate .15 EVGF60 more than the dpairs EVGA60.
If I become a Prof I will give a 0 to anyone who
1. uses a 2 axis usage chart,
2. tries to reduce play to a single variable.
3. Makes any portion of offensive play a measure of Defence.
a. HD corsi rlease
b. open hole Shot suppression
c. Repeatablity of Performance. (belicjek way)
Realative high danger chances for gets you a Zero!
Ps. my HD theory came from the idetic memories of a 3-6 year old watching Flin flon Bombers of Clarke, Leach, Stoughton; Hart; Morrison; Carr; Baird; Arnason; Howatt, poonch, Mcilargey, R.Wlson
Violence about the space in front of the net.