Narrative Analysis: Leafs D sucks & goalie saves them

[Pookie]

Thanks, this usually matches me eyes closest in games,but those are always basically a direct for/against comparison in game.

TB is the best team and seem unbeatable but they are playing with fire these days and are worse than the leafs in GAA. What is their excuse. They are lucky to have such a powerhouse offense or they'd be in trouble.

I feel that their luck is about to run out soon, they've been playing horrible defensively and in net lately, they allowed the Habs to score 5 goals last night, the game before that, the Flyers (who we beat 6-0) scored 5 goals, the Oilers scored 3 goals on 28 shots. BTW, Vasilevskiy played all 3 of those games with a sub .900 S% in each game. Before that, the Flames scored 4 on Domingue.

What I'm trying to ask here is that everybody says that TB has a defense (with Hedman) far better than the Leafs, and Vasilevskiy is far better than Andersen so then why is our GAA better. We are 7th in the league with a GAA of 2.79 and TB is 17th with a GAA of 2.97.

Well the facts are that Vasilevsky isn't better than Andersen.

Andersen has a better save percentage (.923 vs .920) but more importantly has put those numbers up in 30 games relative to Vasilevsky's 19

He has faced 325 fewer shots than Andersen which makes his save percentage more susceptible to variation whereas Andersen's is more established over a longer term with more data. Just posting save percentage is a weak comparison. The context is important.

Their defense does limit more chances that ours does. They are stronger back there.

If we meet them in the postseason, we'd have to hope our goaltending/offense can outplay them.

 

[Pookie]

he used 5 seasons of data.

stop constantly lying.

be an honest poster for once.

You don't understand how studies are constructed.

It's 41 games compared with 41 other games... repeated 5 times. This would be akin to a meta-analysis of 5 independent studies each with n=41 (sample size).

And in his review, he also looked at autocorrelation... how well the first and second halves of the season match up. He concluded "better than random"

For each of his 3 metrics, he found 3 different measures of autocorrelation ranging from .856 (Corsi) to .676 (HDSA). 1 is good. 0 is not.

The first half and second half of a small sample size don't match up... they match up better than perhaps random chance but what happens in the 2nd half of the season? Why don't they match up?

Teams tank. They sell at the deadline. They buy at the deadline. They rest stars. Rosters change. Important variables that need to be isolated. This is why more review is necessary.

Perhaps you have a better understanding now.

 

[zeke]

You don't understand how studies are constructed.

It's 41 games compared with 41 other games... repeated 5 times. This would be akin to a meta-analysis of 5 independent studies each with n=41 (sample size).

And in his review, he also looked at autocorrelation... how well the first and second halves of the season match up. He concluded "better than random"

For each of his 3 metrics, he found 3 different measures of autocorrelation ranging from .856 (Corsi) to .676 (HDSA). 1 is good. 0 is not.

The first half and second half of a small sample size don't match up... they match up better than perhaps random chance but what happens in the 2nd half of the season? Why don't they match up?

Teams tank. They sell at the deadline. They buy at the deadline. They rest stars. Rosters change. Important variables that need to be isolated. This is why more review is necessary.

Perhaps you have a better understanding now.

lmao. this is where pookie realizes he lied about the sample size, and instead of admitting to his lie, he immediately spins off in another nonsensical direction, while using rather pathetic faux-arrogance to try and cover his cowardly tracks.

The methodology, for the record:

"I gathered four statistics: Corsi, scoring chances, high-danger chances, and goals, for all 30 teams that were in the NHL for each season from 2013-14 to 2016-17. That’s all four full 82 game seasons that have been played since the last lockout, which works out to 120 team seasons, which I think is a pretty good sample size. I then split the data into two halves: the first 41 games of each season and the last 41 games of each season. The goal was to see how well you can predict goal scoring in the second half of the season with a team’s results from the first half."

he also refers to a number of other studies with similar sample size.

 

[Pookie]

lmao. this is where pookie realizes he lied about the sample size, and instead of admitting to his lie, he immediately spins off in another nonsensical direction, while using rather pathetic faux-arrogance to try and cover his cowardly tracks.

The methodology, for the record:

"I gathered four statistics: Corsi, scoring chances, high-danger chances, and goals, for all 30 teams that were in the NHL for each season from 2013-14 to 2016-17. That’s all four full 82 game seasons that have been played since the last lockout, which works out to 120 team seasons, which I think is a pretty good sample size. I then split the data into two halves: the first 41 games of each season and the last 41 games of each season. The goal was to see how well you can predict goal scoring in the second half of the season with a team’s results from the first half."

he also refers to a number of other studies with similar sample size.

Again. You don’t understand what you are reading.

Using a small base of games to predict the results in another small base of games is the methodology. 41.

Doing the same thing 5 times (seasons) does not increase the value of the data if our 41 game base remains the same.

A meta analysis of 5 independent studies, each with n=41 is not a significant review to conclude one of our variables is more important than another. It is suggestive.

Not understanding how studies are constructed isn’t a shame. It’s tough stuff.

Back to the program. Your OP is flawed. Not realizing why is quite comical.

 

[Mr Hockey]

That blogger is equal to Stephen Burtch, he is not to be taken seriously, he used to claim Gardiner was the best dman in the league years ago because of his "corsi" He will put out uninformed info, etc as fact because he doesn't know any better.

 

[zeke]

Again. You don’t understand what you are reading.

Using a small base of games to predict the results in another small base of games is the methodology. 41.

Doing the same thing 5 times (seasons) does not increase the value of the data if our 41 game base remains the same.

A meta analysis of 5 independent studies, each with n=41 is not a significant review to conclude one of our variables is more important than another. It is suggestive.

Not understanding how studies are constructed isn’t a shame. It’s tough stuff.

Back to the program. Your OP is flawed. Not realizing why is quite comical.

Every single study you have ever read on hockey stats has analyzed the predictability of stats accumulated over a certain number of games (20, 40, what have you) on the next sample of that number of games.

You are inventing issues that don't exist to try and cover your ignorance and/or dishonesty.

(Even better, you don't seem to realize that the fictions you invent only prove you don't know what you're talking about. You actually seem to be implying that we should be studying how 5 collective seasons of data predict the following season, which is obvious nonsense, and you don't even realize it.)

This is nothing new - everyone who has ever argued with you on any subject knows this is what you always do.

 

[Pookie]

That blogger is equal to Stephen Burtch, he is not to be taken seriously, he used to claim Gardiner was the best dman in the league years ago because of his "corsi" He will put out uninformed info, etc as fact because he doesn't know any better.

Maybe.

But I would say that maybe he found something interesting. It’s just not robust enough though to draw any conclusions based on the design.

Not to mention not accounting for significant roster changes in the second half of the season.

 

[Pookie]

Every single study you have ever read on hockey stats has analyzed the predictability of stats accumulated over a certain number of games (20, 40, what have you) on the next sample of that number of games.

You are inventing issues that don't exist to try and cover your ignorance and/or dishonesty.

(Even better, you don't seem to realize that the fictions you invent only prove you don't know what you're talking about. You actually seem to be implying that we should be studying how 5 collective seasons of data predict the following season, which is obvious nonsense, and you don't even realize it.)

This is nothing new - everyone who has ever argued with you on any subject knows this is what you always do.

I just seem to pick apart your shit zeke.

Call it low hanging fruit.

Back to the topic if you like. High scoring and good goaltending is saving our ass.

 

[zeke]

I just seem to pick apart your **** zeke.

Call it low hanging fruit.

Back to the topic if you like. High scoring and good goaltending is saving our ass.

And pookie's last resort when he gets both his ignorance and dishonesty clearly exposed - the famous Pookie Chicken Run.

 

[Mr Hockey]

Maybe.

But I would say that maybe he found something interesting. It’s just not robust enough though to draw any conclusions based on the design.

Not to mention not accounting for significant roster changes in the second half of the season.

These people will not acknowledge the flaws in the hockey advanced stats, there are too many moving parts in the game of hockey. It's getting better though, they used to say Corsi with zero context was the best way to judge a player lol

 

[Pookie]

And pookie's last resort when he gets both his ignorance and dishonesty clearly exposed - the famous Pookie Chicken Run.

I’m running?

Weird interpretation. Why would I leave with all this low hanging fruit around?

 

[Pookie]

These people will not acknowledge the flaws in the hockey advanced stats, there are too many moving parts in the game of hockey. It's getting better though, they used to say Corsi with zero context was the best way to judge a player lol

I’ll give you a like for that.

It’s weird to me how many.... not all of course... of the stats backers don’t even take the time to understand the methodology of the data collected.

Zeke can’t even acknowledge that the stats data he likes highlights we have a weak D relative to the contenders.

 

[zeke]

I’m running?

Weird interpretation. Why would I leave with all this low hanging fruit around?

ah, now it's the almost as famous POOKIE POOKIE BOO BOO I CAN'T HEAR YOU tactic.

solid.

 

[Gary Nylund]

You are inventing issues that don't exist to try and cover your ignorance and/or dishonesty.

This is nothing new - everyone who has ever argued with you on any subject knows this is what you always do.

Truer words have never been spoken.

 

[Pookie]

One of my favourite @zeke quotes, written on Jan 22, 2017... Leafs had played 40 games as of this date:

Freddy ranks 29th in save percentage. but then again a great goale like Schneider is 30th, while the #1 guy on that list is.....Jhonas Enroth.

sample size is too small to make any statistical conclusions, eithee way.

by the eye test, though, he seems to get not just beat but beat badly much of the time. it's weird.

Look at all this fruit:

1) Sample size is a MASSIVE and VALID criticism. It is hugely important and is a major reason why most stats are useless. It's not funny to observe this - in fact, if you don't understand how critical this is this only makes your understanding look bad.

... the real power of numbers comes with larger sample size.

You're absolutely right that we simply cannot take a backup goalie's stats seriously. Not only is it a small sample, but it's a biased sample of selected games against selected opponents.

i would argue the difference is mostly due to sample size - special teams are just a fraction of the minutes of even stength play, so of course they will be much noisier and less consistent than even strength numbers.

And let's end this on a delicious one:

the biggest problem was a dummy like Steve Burtch not understanding the fundamentals of statistics, such as sample size, when he decided to push his Bozak wowy/Kessel numbers to prove that Bozak was the worst.

You want us to take your OP seriously using a small sample size... one not good enough for you (see above).... and take your blog reference seriously based on 41 games... 1 more than 40 which again wasn't good enough for you?

This stuff writes itself.

 

[zeke]

One of my favourite @zeke quotes, written on Jan 22, 2017... Leafs had played 40 games as of this date:

​

Look at all this fruit:

​

And let's end this on a delicious one:

​

You want us to take your OP seriously using a small sample size... one not good enough for you (see above).... and take your blog reference seriously based on 41 games... 1 more than 40 which again wasn't good enough for you?

This stuff writes itself.

you either were wrong or lied about the sample size, and either way, demonstrated you don't actually understand how sample size works or applies to these studies.

it's ok.

 

[Pookie]

Update:

GA/gm

1.NSH 2.56
7.TOR 2.79
9.WPG 2.82
11.WSH 2.84
15.PIT 2.92
17.TBL 2.97


SA/gm

4.NSH 28.5
20.WSH 31.9
21.TBL 32.4
22.TOR 32.5
23.WPG 32.6
25.PIT 33.1


SV%

6.TOR 91.40
7.WPG 91.36
8.PIT 91.17
9.WSH 91.12
10.NSH 91.02
11.TBL 90.82

pretty cool - all 6 contenders mooshed right in there back to back to back in the 6-11 slots in team save percentage.

Meh.

I remember this one dude told us that he couldn’t make any comparison/conclusion for Freddy and the league based on a small sample size.

That was after 40 games. Looked a lot like you. Oh wait... I still have the quote from the dude. Here it is:

Freddy ranks 29th in save percentage. but then again a great goale like Schneider is 30th, while the #1 guy on that list is.....Jhonas Enroth.

sample size is too small to make any statistical conclusions, eithee way.

by the eye test, though, he seems to get not just beat but beat badly much of the time. it's weird.

So.... according to him... who looked a lot like you.... this is pretty worthless based on a small sample size.

 

[Joey Hoser]

This is nothing new - everyone who has ever argued with you on any subject knows this is what you always do.

I'm sure trading Kadri will solve whatever problem he's on about now.

 

[zeke]

Meh.

I remember this one dude told us that he couldn’t make any comparison/conclusion for Freddy and the league based on a small sample size.

That was after 40 games. Looked a lot like you. Oh wait... I still have the quote from the dude. Here it is:



So.... according to him... who looked a lot like you.... this is pretty worthless based on a small sample size.

the more you repeat this attempted gotcha, the more you expose you don't know what you're talking about.

 

[Pookie]

the more you repeat this attempted gotcha, the more you expose you don't know what you're talking about.

OP makes a point of comparing data from 40ish games.

A guy who looks like you says you can’t make any kind of conclusion... ironically about goaltending... based on a small sample of 40ish games.

What’s the deal zeke... do you agree with the guy who looks like you or not?

 

[zeke]

OP makes a point of comparing data from 40ish games.

A guy who looks like you says you can’t make any kind of conclusion... ironically about goaltending... based on a small sample of 40ish games.

What’s the deal zeke... do you agree with the guy who looks like you or not?

this is painful.

 

[Pookie]

this is painful.

No doubt.

It’s ridiculous that a “stats expert” doesn’t know the basics of study design.

I’m not blaming you. Our school system doesn’t focus on critical thinking.

But just so you know, when you have 5 studies that are pooled together, that’s called a meta analysis.

However, a “meta-analysis of several small studies does not predict the results of a single large study.“ - New England Journal of Medicine.

Hopefully, you can incorporate this into your “expert” analysis the next time.

 

[razkaz]

Hey cool. Thanks for sharing.

Can you link me to that report? I looked but came up empty.

My wife calls that “man looking” when I go into the fridge and ask where the milk is. Turns out it was behind something.

Here you go

Player Season Totals - Natural Stat Trick

 

[JT AM da real deal]

Nice fight boys. We need a little of that from our defense and we will be good to go for playoffs. Has any of our D got a fighting major this year?

 

[zeke]

No doubt.

It’s ridiculous that a “stats expert” doesn’t know the basics of study design.

I’m not blaming you. Our school system doesn’t focus on critical thinking.

But just so you know, when you have 5 studies that are pooled together, that’s called a meta analysis.

However, a “meta-analysis of several small studies does not predict the results of a single large study.“ - New England Journal of Medicine.

Hopefully, you can incorporate this into your “expert” analysis the next time.

oh pookie you poor thing. just stop.