Narrative Analysis: Leafs D sucks & goalie saves them

Pookie

Wear a mask
Oct 23, 2013
16,172
6,684
Going strictly by eye test, I have no idea who is winning the Zeke/Pookie debate.

I’m not sure he’s even debating ;)

The basic point is that whether we use Scoring Chances, Corsi Against, Expected Goals... limitations of those aside... our Leafs should be giving up more goals than they are.

That points squarely at Freddy. Data suggests we rely more on him than most other contending teams do.

This shouldn’t be surprising to anyone. 75% of the games so far. With a .923 vs Sparks’ .905.

30 GP for Freddy which is more than most other contenders (eg Tampa 19 GP for their starter).

The other stuff.... like zeke saying these 40 games count and these 40 don’t when it’s convenient... or learning what a meta analysis is... I just chalk it up to the school system.
 
  • Like
Reactions: JT AM da real deal

Marshy

Behind Enemy Lines
Oct 3, 2007
8,143
9,193
Ottawa
I’m not sure he’s even debating ;)

The basic point is that whether we use Scoring Chances, Corsi Against, Expected Goals... limitations of those aside... our Leafs should be giving up more goals than they are.

That points squarely at Freddy. Data suggests we rely more on him than most other contending teams do.

This shouldn’t be surprising to anyone. 75% of the games so far. With a .923 vs Sparks’ .905.

30 GP for Freddy which is more than most other contenders (eg Tampa 19 GP for their starter).

The other stuff.... like zeke saying these 40 games count and these 40 don’t when it’s convenient... or learning what a meta analysis is... I just chalk it up to the school system.


Freddy is playing out of his mind relative to the shots he faces. League average goaltending or worse and we are in a heap of trouble. Now I realize not all shots are created equal but any goalie among the leaders in shots faced AND save percentage usually would mean we are getting quite a few wins because of goaltending (or at least not losing as much as we should be).
 
  • Like
Reactions: Pookie

Pookie

Wear a mask
Oct 23, 2013
16,172
6,684
Freddy is playing out of his mind relative to the shots he faces. League average goaltending or worse and we are in a heap of trouble. Now I realize not all shots are created equal but any goalie among the leaders in shots faced AND save percentage usually would mean we are getting quite a few wins because of goaltending (or at least not losing as much as we should be).

And that sir is the essence of the “debate”

We win with an above average offence, relying on one above average goalie in particular, riding him through more games than other contenders. Our D needs to improve relative to the others.

And I’d say we need a little sandpaper but that’s just a feeling.

This is what we all know to be true. Zeke seems intent on fudging and ignoring both the eye test and tests of statistical evidence he claims to be an expert in.

I tried logic. I tried using stats. It’s just a lost cause.

You and I are in agreement. A happy new year to you.
 
  • Like
Reactions: Marshy

zeke

The Dube Abides
Mar 14, 2005
66,937
36,957
Going strictly by eye test, I have no idea who is winning the Zeke/Pookie debate.

Hint: It's not the guy who claimed r^2 justification for his stats then ran away when confronted with the actual r^2 for his stats, then lied about sample size to cover his tracks.
 
  • Like
Reactions: stickty111

Gallagbi

Formerly Eazy_B97
Jul 5, 2005
48,490
11,109
Hint: It's not the guy who claimed r^2 justification for his stats then ran away when confronted with the actual r^2 for his stats, then lied about sample size to cover his tracks.
What's the lie on sample size? Seems like he has a quote, I don't think it's made up.
 

zeke

The Dube Abides
Mar 14, 2005
66,937
36,957
Freddy is playing out of his mind relative to the shots he faces. League average goaltending or worse and we are in a heap of trouble. Now I realize not all shots are created equal but any goalie among the leaders in shots faced AND save percentage usually would mean we are getting quite a few wins because of goaltending (or at least not losing as much as we should be).

This is demonstrably false, by every measure we have.

Corsica provides dsv%, which is the difference between save percentage and expected save percentage based on shot quality.

dSV% (min 1000 minutes)

1.Gibson +2.09
2.Halak +1.74
3.Greiss +1.68
4.Rittich +1.43
5.Desmith +1.21
6.Holtby +1.08
7.Rinne +0.93
8.Andersen +0.90
9.Bishop +0.89
10.Howard +0.87

Andersen ranks pretty much exactly as well in dsv% as he does in sv%.
 
  • Like
Reactions: stickty111

zeke

The Dube Abides
Mar 14, 2005
66,937
36,957
What's the lie on sample size? Seems like he has a quote, I don't think it's made up.

I quoted the methodology from the link I posted more than once.

Read it.

Then realize that Pookie is calling that a 41gm sample size.
 

Gallagbi

Formerly Eazy_B97
Jul 5, 2005
48,490
11,109
I quoted the methodology from the link I posted more than once.

Read it.

Then realize that Pookie is calling that a 41gm sample size.
And your thought is that it is a ~12,300 game sample since it's 2 segments of 41 games for all 30 teams over 5 years?
 

Marshy

Behind Enemy Lines
Oct 3, 2007
8,143
9,193
Ottawa
This is demonstrably false, by every measure we have.

Corsica provides dsv%, which is the difference between save percentage and expected save percentage based on shot quality.

dSV% (min 1000 minutes)under

1.Gibson +2.09
2.Halak +1.74
3.Greiss +1.68
4.Rittich +1.43
5.Desmith +1.21
6.Holtby +1.08
7.Rinne +0.93
8.Andersen +0.90
9.Bishop +0.89
10.Howard +0.87

Andersen ranks pretty much exactly as well in dsv% as he does in sv%.



Am I understanding it right? Andersen is 8th in the league (5th or so among starters with heavy workload) in saving BETTER than his expected save percentage? All while being top 3 in shots faced in the league. Help me understand why that isn't helping my case?
 
Last edited:
  • Like
Reactions: Pookie

Pookie

Wear a mask
Oct 23, 2013
16,172
6,684
And your thought is that it is a ~12,300 game sample since it's 2 segments of 41 games for all 30 teams over 5 years?

If I can read between his personal jabs, I believe he thinks we have 5 seasons worth of data... not a first half season comparison to the second half done 5 times.
 

Pookie

Wear a mask
Oct 23, 2013
16,172
6,684
This is demonstrably false, by every measure we have.

Corsica provides dsv%, which is the difference between save percentage and expected save percentage based on shot quality.

dSV% (min 1000 minutes)

1.Gibson +2.09
2.Halak +1.74
3.Greiss +1.68
4.Rittich +1.43
5.Desmith +1.21
6.Holtby +1.08
7.Rinne +0.93
8.Andersen +0.90
9.Bishop +0.89
10.Howard +0.87

Andersen ranks pretty much exactly as well in dsv% as he does in sv%.

Thanks for this.

In 5v5, Andersen is 1.14... playing well above the 0 indicator for "average" goaltending. 4th overall in that category.

Halak (BOS) is doing better... 1.75... but playing in 8 less games. Rask is playing below average in 19 games.

Meaning that we have an above average Goaltender playing in more games. This would be true for the majority of our contending competition.

Want to try expected Save Percentage next? ;)
 

zeke

The Dube Abides
Mar 14, 2005
66,937
36,957
Am I understanding it right? Andersen is 8th in the league (5th or so among starters with heavy workload) in saving BETTER than his expected save percentage? All while being top 3 in shots faced in the league. Help me understand why that isn't helping my case?

you are reading it right.

Andersen ranks 8th in the league in sv% over expected save percentage based on shot quality.
 
  • Like
Reactions: stickty111

zeke

The Dube Abides
Mar 14, 2005
66,937
36,957
And your thought is that it is a ~12,300 game sample since it's 2 segments of 41 games for all 30 teams over 5 years?

actually it's 4yrs but yes, that is the proper description of the sample size.
 

Notsince67

Papi and the Lamplighters
Apr 27, 2018
15,854
11,052
And that sir is the essence of the “debate”

We win with an above average offence, relying on one above average goalie in particular, riding him through more games than other contenders. Our D needs to improve relative to the others.

And I’d say we need a little sandpaper but that’s just a feeling.

This is what we all know to be true. Zeke seems intent on fudging and ignoring both the eye test and tests of statistical evidence he claims to be an expert in.

I tried logic. I tried using stats. It’s just a lost cause.

You and I are in agreement. A happy new year to you.
Is it possible that you are both right? I figure that anyone who doesn't point to the right side D as the weakest part of the team is blind. That being said, they are occasionally capable of mounting an acceptable performance that allows the leafs to win if the opposing team doesn't target their right side like Boston. They are good enough to not see an incremental team lift by a Faulk or Pesce if the tradeoff is an impactful forward. They would however see a notable improvement with a gritty, puck moving RHD
 

Pookie

Wear a mask
Oct 23, 2013
16,172
6,684
Is it possible that you are both right? I figure that anyone who doesn't point to the right side D as the weakest part of the team is blind. That being said, they are occasionally capable of mounting an acceptable performance that allows the leafs to win if the opposing team doesn't target their right side like Boston. They are good enough to not see an incremental team lift by a Faulk or Pesce if the tradeoff is an impactful forward. They would however see a notable improvement with a gritty, puck moving RHD

I don’t know. His point is that we don’t have a weak D and don’t rely on goaltending.

Not sure how that can coexist with the opposite view.
 

Gallagbi

Formerly Eazy_B97
Jul 5, 2005
48,490
11,109
This is demonstrably false, by every measure we have.

Corsica provides dsv%, which is the difference between save percentage and expected save percentage based on shot quality.

dSV% (min 1000 minutes)

1.Gibson +2.09
2.Halak +1.74
3.Greiss +1.68
4.Rittich +1.43
5.Desmith +1.21
6.Holtby +1.08
7.Rinne +0.93
8.Andersen +0.90
9.Bishop +0.89
10.Howard +0.87

Andersen ranks pretty much exactly as well in dsv% as he does in sv%.
Wouldn't his dSv% be 0 if he's performing as expected? Not exactly a list of Allstars at the top of that chart either (though I do like Gibson)
 
Last edited:

Pookie

Wear a mask
Oct 23, 2013
16,172
6,684
Wouldn't his dSv% be 0 if he's performing as expected? Not exactly a list of Allstars at the top of that chart either (though I do like Gibson)

You are correct.

He is performing above expectations in heavy usage... more than most other contenders... suggesting that the goalie is contributing significantly to the positive results thus far (ie. saving them).

Not sure why zeke posted that and drew the opposite conclusion.
 

Pookie

Wear a mask
Oct 23, 2013
16,172
6,684
@Gallagbi

One final thought on the sample size topic that has dominated the last few pages.

The author zeke cites suggested that because they used data over 4 seasons (my bad) that his sample size was 120.

The author is incorrect. Sample size is the number of data events being measured. In this case, Corsi over 41 games. n=41. We can repeat this 30 times (for each team) or 120 times (for each team over 4 seasons) or 10,000 times if we want... but it doesn't change the event being measured.

If this causes a headache, let's simplify.

For example, let's say we are 5 games into a new season. And we want to look at shots and try to make some comparisons across the NHL.

What's our sample size?

a) 31 teams x 5 games = 155 or
b) 5 games

155 is a number suggestive of a very large sample. But we all know you cannot make inferences after just 5 games.

This is what zeke's blogger... and zeke... are inferring from the data. That we have 4 years worth of data. It's true but it's just the same 41 events repeated many times. Very much a meta analysis of the same study repeated based on a small sample of data.

That's it for me on this. He either gets it or he doesn't.
 

4thline

Registered User
Jul 18, 2014
14,371
9,674
Waterloo
@Gallagbi
The author zeke cites suggested that because they used data over 4 seasons (my bad) that his sample size was 120.

The author is incorrect. Sample size is the number of data events being measured. In this case, Corsi over 41 games. n=41. We can repeat this 30 times (for each team) or 120 times (for each team over 4 seasons) or 10,000 times if we want... but it doesn't change the event being measured.

What you're saying is equivalent to saying "it doesn't matter how many people take the test, it has 41 questions so the sample is 41". The event being measured is Team Seasons, specifically the predictive ability of the 1st half to the second.

When you graph the result of the study, does it have 120 data points or 41?
 
Last edited:

Ad

Upcoming events

Ad

Ad

-->