[Evolving Hockey] Defenseman GAR Ranking for 2019-2020

[North Cole]

Probably the best demonstration possible for failing to account for quality of competition.

"BUt lOok AT tHe ReGrESsiOn"

 

[Fixed to Ruin]

As a draftkings player, these are the guys who helped me win the most money this year at the defense position.

 

[VoluntaryDom]

GAR uses regression of on-ice GF, not points, to calculate EV offense.

At 5-on-5 with Ellis the Predators scored 4.09 GF/60 and without him they scored 2.29 GF/60. It makes sense that any regression model would suggest he had an absolutely insane impact on his team’s on-ice goal rates.

While Ellis is an excellent defenseman, the fact that he ranks so high shows a few of the key flaws with this stat. Using on-ice GF (or individual scoring) for any player in a single year sample will tend to produce some wacky results due to variations in on-ice shooting and GAR is no different in that respect. The even strength offense component of GAR explains a huge portion of the variance observed in skater GAR from year to year, but quite often it can be heavily skewed by luck. This is an even bigger problem for defensemen because conventional wisdom and research both tell us that unlike forwards, defensemen have little control over their on-ice shooting rates to begin with; it’s probably not actually Ellis’ doing that the Predators are shooting 11.28% with him and 7.43% without him, yet a glance at the results just says “Ellis was the best.”

It’s still a very useful metric, but it’s obviously not something that anybody should just take as gospel or a definitive ranking. Ideally, you’d see analytics fans use it the way that “eye test” fans use points today; not a definitive judge of player value, but the list of best players by position will generally look similar to the list of best players by this stat. In practice I think that mostly holds.

theres multiple years of ellis doing this though. 13-14 onwards these are his D GAR finishes

13-14: 2nd
14-15: 29th
15-16: 6th
16-17: 43rd
17-18: 12th
18-19: 25th
19-20: 1st
overall: 1st, barely ahead of Victor Hedman

even if we remove 19-20 (his best/maybe luckiest season), and just use 13-19, hes 3rd behind Hedman and Giordano

 

[BringTheReign]

TBH I've never heard of Matt Roy.

Most Kings fans would've told you the same thing before the start of this year. Now we're penciling him into our 2nd pairing RD spot for the next 8 years and running.

 

[Do Make Say Think]

As much as people try, making a WAR for hockey is just never going to be as clean as it is in baseball.

But how else will people who understand very little of the sport be able to give the appearance of authority if they can't talk about science?!

 

[TomasHertlsRooster]

theres multiple years of ellis doing this though.

No, there aren’t. This is his first season leading the league in GAR. This is his first season where his 5-on-5 on-ice shooting percentage exceeded 10.

 

[Sidney the Kidney]

John Marino > Adam Fox confirmed.

 

[VoluntaryDom]

No, there aren’t. This is his first season leading the league in GAR. This is his first season where his 5-on-5 on-ice shooting percentage exceeded 10.

ok maybe not THIS, but being an elite dman yeah

 

[TeddyBare]

One of GAR's biggest weaknesses is is the penalty kill, it is pretty much noise. So that should be noted on any player on here.

I like GAR so much better for forwards. It highlights Panarin and shows why he may be a Hart Trophy winner

 

[TomasHertlsRooster]

ok maybe not THIS, but being an elite dman yeah

My post literally said that Ellis is an excellent defenseman, I don't know why you're responding to me as if I said anything otherwise.

 

[KirkAlbuquerque]

But how else will people who understand very little of the sport be able to give the appearance of authority if they can't talk about science?!

f***in poindexters will never learn, the EYE TEST >>>> all!

 

[KirkAlbuquerque]

One of GAR's biggest weaknesses is is the penalty kill, it is pretty much noise. So that should be noted on any player on here.

I like GAR so much better for forwards. It highlights Panarin and shows why he may be a Hart Trophy winner

You can look at his goals, assists, +/- , EV scoring , takeaways and watch him play to see that he deserves the Hart. No spreadsheets required.

 

[TomasHertlsRooster]

f***in poindexters will never learn, the EYE TEST >>>> all!

...

You can look at his goals, assists, +/- , EV scoring , takeaways

I thought the eye test was greater than all? Why have you suddenly switched to numbers from a spreadsheet?

Has it occurred to you that you may be one of the very “poindexters” that you’re criticizing?

 

[KirkAlbuquerque]

...



I thought the eye test was greater than all? Why have you suddenly switched to numbers from a spreadsheet?

Has it occurred to you that you may be one of the very “poindexters” that you’re criticizing?

shut up nerd

 

[TomasHertlsRooster]

shut up nerd

You’re the one using numbers such as goals, assists, EV scoring, takeaways, and plus/minus to support your argument. Wouldn’t that make you a nerd as well?

 

[tarheelhockey]

Wouldn't trust a thing that comes out of that company.

All of their 'advanced' stats seem to be mostly-garbage and are used as a tool to gain attention and notoriety. Great job in aiding them this way.

I think it's great to have companies out there pushing the envelope with this stuff. The only way the tools get better is to continually subject them to experiment. Can't sharpen a blade without friction.

That being said... I agree that this company specifically seems to be the source of more than their fair share of ridiculous arguments.

 

[The Macho King]

I think it's great to have companies out there pushing the envelope with this stuff. The only way the tools get better is to continually subject them to experiment. Can't sharpen a blade without friction.

That being said... I agree that this company specifically seems to be the source of more than their fair share of ridiculous arguments.

There's also the fact that I think the companies that produce the really good stuff end up working directly/contracting with teams. A lot of it is because the shot quality metrics these guys use are based exclusively on shot location, which is not great. A fourth liner two inches from the net banging the puck directly into a goaltenders pads isn't as high danger as a Stamkos one-timer off of a cross-ice pass from Nikita Kucherov.

There was an Athletic article from someone that contracts with some NHL teams that gave some insight into how they calculate shot quality, and the best way to do it is *actually looking at all of the shots*. There's too much context in shots in the NHL. There's where the shot is from. There's who's shooting. There's pre-shot movement. There's did they scuff the shot. Was the shot off the ice or was it right in the middle of the net.

The article was discussing goaltending, but for most of these factors they look at can be used to judge both shooters and defenders as well.

 

[Merrrlin]

I wonder how Pulock did with and without Pelech. As underrated as Pulock is, Pelech is far more underrated (and a steal at 1.6 the next two or three years).

The Pulock/Pelech pairing is by far the best the Islanders have seen in a while. Adding Dobson and Toews as a 2nd pairing will give the Isles an excellent top 4 for years.

Pulock/Pelech so good. Finally starting to get the cred they deserve. Not as sold on Toews or especially Dobson but the future still looks bright.

 

[bambamcam4ever]

There's also the fact that I think the companies that produce the really good stuff end up working directly/contracting with teams. A lot of it is because the shot quality metrics these guys use are based exclusively on shot location, which is not great. A fourth liner two inches from the net banging the puck directly into a goaltenders pads isn't as high danger as a Stamkos one-timer off of a cross-ice pass from Nikita Kucherov.

There was an Athletic article from someone that contracts with some NHL teams that gave some insight into how they calculate shot quality, and the best way to do it is *actually looking at all of the shots*. There's too much context in shots in the NHL. There's where the shot is from. There's who's shooting. There's pre-shot movement. There's did they scuff the shot. Was the shot off the ice or was it right in the middle of the net.

The article was discussing goaltending, but for most of these factors they look at can be used to judge both shooters and defenders as well.

Completely agree, I don't doubt there are ways to accurately measure shot quality, but the publicly available data is simply inadequate, no matter how hard these Twitter statisticians contort the numbers. Garbage in = garbage out.

As you said, right now we have just location. We need direction the shooter was skating, how long he's had control of the puck, distance from the shooter to defenders, location of the goalie, and direction goalie is moving, plus if the goalie's view is obstructed.

And look at Sheary's goal below. I remember this shot was worth 0.1 "expected goals." A wide open player shooting into a wide open net. That should be invalidating for xG, but people keep bringing it up like it's worth something.

 

[Do Make Say Think]

f***in poindexters will never learn, the EYE TEST >>>> all!

The eye test is but one element. Advanced stats do matter.

That being said, I will always trust someone who can properly explain the results of their eye test over someone who is good at Excel.

If you are able to explain to me what about the players game you like in plain language, you have infinitely more credibility than someone who does nothing except crunch numbers.

 

[majormajor]

Completely agree, I don't doubt there are ways to accurately measure shot quality, but the publicly available data is simply inadequate, no matter how hard these Twitter statisticians contort the numbers. Garbage in = garbage out.

As you said, right now we have just location. We need direction the shooter was skating, how long he's had control of the puck, distance from the shooter to defenders, location of the goalie, and direction goalie is moving, plus if the goalie's view is obstructed.

And look at Sheary's goal below. I remember this shot was worth 0.1 "expected goals." A wide open player shooting into a wide open net. That should be invalidating for xG, but people keep bringing it up like it's worth something.

I can't speak to the data that Evolving Wild is using here but a lot of people are putting pre-shot movement in their xG formulas. I've lost track of who is doing what but I can tell you it has been a frequent topic for the people in the field. One touch shot off of a cross ice pass gets a huge xG.

 

[elitepete]

Does Pulock play the tough mins for the Islanders?

 

[Machinehead]

Nice to see Pulock get some respect. He's an animal.

 

[PWJunior]

Does Pulock play the tough mins for the Islanders?

Yes. Pulock gets matched up against the top lines, gets #2PP time, and gets #2PK time. He is the top dog. Along with Pelech, they are the top pair.

The only thing he lacks are top end wheels. 105 mph clapper, he can rip it. A little Fulton Reed-ish though, scattershot yet devastating when he gets it on net. Built like a tank at 6'2 and 220-ish.

Think poor man's Shea Weber, Al MacInnis type.

 

[Aladyyn]

The eye test is but one element. Advanced stats do matter.

That being said, I will always trust someone who can properly explain the results of their eye test over someone who is good at Excel.

If you are able to explain to me what about the players game you like in plain language, you have infinitely more credibility than someone who does nothing except crunch numbers.

Most people can't even do that for players on their own team. Leaguewide? Pretty much nobody.