2018-19 stats and underlying metrics thread

[JetsWillFly4Ever]

Every time I see KC on the power play, all I see him do is get rid of the puck before it explodes. I really don't see what his purpose it there, he should be an alternative passer to Scheifele in the bumper spot but he's not. I think they should try Hayes there, good passer and a bigger body in front.

I thought for sure when we traded for Hayes he would replace Connor.

KC doesn't add anything to the PP at all at this point.

 

[Calendal]

For most people +/- is simply more easily accessible. I wouldn't call it next to useless at all. It's useful, provided its limitations are understood. Other stats, like 5x5 g diff are better, but if very few are accessing or talking about them on a social media platform, they are thus limited too.

I think the notion that the usefulness of a stat is dictated by it’s availability is fundmentally flawed. We can certainly say it’s available and it’s accessible.

I think sufficient context to make +/- representative of a player’s ability is significantly less accessible than for many other stats. In other words: it’s easier to pick another stat than it is to understand limitations of +/- in the context of an individual player.

 

[Gm0ney]

Studies have shown it to be close to random.

Brian Burke once called it a "horse shit stat".

 

[Duke749]

Brian Burke once called it a "horse **** stat".

Idk why people would want to use a stat that can interpreted about 100 different ways and can mean something different to 50 different people. It’s about as vague as anything out there that even with extreme context tells very little, like the tiniest of hints.

 

[DRW204]

Tyler Myers - Traditional (+/-): +8
Tyler Myers - 5v5 (+/-): -8

i am not one of those guys that try to constantly shit on (+/-) or claim it's useless. But it really should be paired with 5v5 (+/-) to give it some additional commentary & context

 

[YWGinYYZ]

Tyler Myers - Traditional (+/-): +8
Tyler Myers - 5v5 (+/-): -8

i am not one of those guys that try to constantly **** on (+/-) or claim it's useless. But it really should be paired with 5v5 (+/-) to give it some additional commentary & context

That's stratifying something that measures what are already rare events. +/- is descriptive of very specifically (and strangely) defined events, but it has no predictive value.

 

[DRW204]

That's stratifying something that measures what are already rare events. +/- is descriptive of very specifically (and strangely) defined events, but it has no predictive value.

i am not sure if you're for or against what i am saying lol. but yea i agree, stats are generally descriptive on what has happened

 

[YWGinYYZ]

i am not sure if you're for or against what i am saying lol. but yea i agree, stats are generally descriptive on what has happened

Some have proven to be useful for prediction though. +/- is not one of them.

Not arguing with you: just stating that breaking a stat that measures rare events (goals) into different bins makes for even fewer measured events in each bin. Add the fact that +/- is assigned "strangely", and it's usefulness regarding predictive abilities is negligible.

 

[garret9]

No it's not.

It's something the fancystats people want everyone to believe and it's now become a meme.

It's certainly an imperfect measure and shouldn't be taken in isolation, but it has value.

There - I said it.

No it’s fairly useless, because anything it tries to do has something that does that better, even when just answering the question who outscored more.

Stats are merely a tool, while +/- is an outdated and fairly useless one.

In a full season it doesn’t even pass a 0.5 alpha coefficient, which isn’t a hockey analytics invented test.

That means about half of plus minus at one season’s worth is noise. The *non* noise half would still include usage/deployment, team quality, goaltending, etc.

 

[JetsFan815]

I am not talking about advantage or perfect statistics (because I don't think they exist).

I think that a player who, year over year, is a plus demonstrates that they are a net positive.

That takes into account some of the quirks of plus minus and also things like team and partner impacts.

Again, I am not saying it's a perfect measure, nor should it be used in isolation to prove anything, but its hardly useless.

For most people +/- is simply more easily accessible. I wouldn't call it next to useless at all. It's useful, provided its limitations are understood. Other stats, like 5x5 g diff are better, but if very few are accessing or talking about them on a social media platform, they are thus limited too.

Using +/- is like putting your hand on the forehead of someone to determine if they have a fever which is impacted by- your own body temperature, the relative and subjective perception of whether it feels "hot" and if it feels "hot" how hot- a 104 degree temperature will require a very different urgency and remediation than a 100.5 degree one and it is unlikely your hand can tell the difference. Why do that when you can use a thermometer with decimal level precision?

This talk of context and accessibility but this is the amount of work an average person has to do to put +/- into proper context. Get +/- for all the players you are interested in, get number of goals into opposition ENs when each player was on the ice and subtract it from their +/- number, then get the number of goals into the player's team's EN when each player was on the ice and add it to their +/-, get the number of PP goals against and add it to their +/-, get the number of PK goals for and subtract it from their +/-. Some of these (EN Goals For/Against) are not easy to find and require some digging to unearth.

Now think about how many people who use this stat are willing to put all that work to properly contextualize it. Not that many I imagine.

 

[garret9]

Two players A and B are +8 and +6. Which player is better?

One of many ways to show the issues with plus/minus... starting with the issues that also impact “fancy” stats:

1) what if A plays with top-form Scheifele the most and B plays with Thorburn

2) what if A plays against Thorburn and B plays against top-form Scheifele

3) what if A mostly gets ozone faceoffs and B dzone

Now the plus-minus specific ones

1) what if A did that in 800 mins and B did that in 400 mins

2) what if A is +508/-500 (50.8% GF%) and B is +56/-50 (56.0% GF%)

3) what if A plays PK and empty net situations where he is +4/-50 (which would be bad, FYI) and +4 at 5v5 while B only plays at 5v5

Then the stuff where analytics comes in:

Players have some impact on how many goals per shot go in the net for offense, but when it comes to goals against the numbers act no differently than if it were purely random (this isn’t to say goals against are random. There is a reasonable explanation to this but this post is getting long*).

Whether you split by 20 or 40 games, a season or two, etc. we’ve seen no repeatability in players being able to sustain positive or negative impact on save percentage outcomes.



*the tl;dr version would be that outside factors (like goaltending performance variance and small samples) exude too much control to make their control significant

 

[WPGChief]

For those who care, SeaHAC is going on right now with some heavy hitters in terms of presenters. I can't watch much unfortunately so I'll wait for the recaps. Livestreams are found courtesy of @emilyjomichele:


One point / finding / conclusion thus far:

 

[Sperss1997]

I made a match-up with Leafs and Bruins forward line-up and thought I should post the Jets line-up in here as well
All RAPM data are from Evolving-Hockey.com. Only 2018/19 and even strength.
The attempt is to isolate a players impact and performance on the team, independant of both his teammates and opponents, compared to league average.
@garret: If RAPM data are invalid then please let me now! Some data seem rather dubious

Jets forward group impact has been mediocre. Scheifele and Wheeler dissapointing. Connor need to work things out in his own end, so does Laine.
(sorry about the puncuation mistakes in the table)

 

[Mud Turtle]

I made a match-up with Leafs and Bruins forward line-up and thought I should post the Jets line-up in here as well
All RAPM data are from Evolving-Hockey.com. Only 2018/19 and even strength.
The attempt is to isolate a players impact and performance on the team, independant of both his teammates and opponents, compared to league average.
@garret: If RAPM data are invalid then please let me now! Some data seem rather dubious

Jets forward group impact has been mediocre. Scheifele and Wheeler dissapointing. Connor need to work things out in his own end, so does Laine.
(sorry about the puncuation mistakes in the table)

View attachment 197871

Good thing they’re getting 25 minutes per game and also a good thing that they’re getting all that penalty killing time.
Should be fresh every game.

 

[YWGinYYZ]

Good thing they’re getting 25 minutes per game and also a good thing that they’re getting all that penalty killing time.
Should be fresh every game.

Well, at least it's trending in a better direction with Hayes here. Between Hayes' line taking more 5v5 time, and Hayes and KC getting PK time, this does help going forward.

 

[Board Bard]

I made a match-up with Leafs and Bruins forward line-up and thought I should post the Jets line-up in here as well
All RAPM data are from Evolving-Hockey.com. Only 2018/19 and even strength.
The attempt is to isolate a players impact and performance on the team, independant of both his teammates and opponents, compared to league average.
@garret: If RAPM data are invalid then please let me now! Some data seem rather dubious

Jets forward group impact has been mediocre. Scheifele and Wheeler dissapointing. Connor need to work things out in his own end, so does Laine.
(sorry about the puncuation mistakes in the table)

View attachment 197871

Like most charts and graphs (and eye test) this season, it paints a gloomy picture.

 

[Whileee]

Much has been written about the Jets' decline in defensive performance this season. Here are a couple of charts showing the Evolving Wild's expected goals against per 60 minutes (xGA_60) metric for 2018/19, along with the difference in performance from last season (2017/18). It's pretty evident that many of the Jets top forwards have poor xGA_60, and have declined from last season.

Among defensemen, only Chiarot has had a strong showing in xGA_60, with a big improvement from last season. You could see that it's understandable to pair him with an offensive D this season, based on his defensive performance in shutting down shots / chances against.

But overall, not a pretty picture...

 

[surixon]

Much has been written about the Jets' decline in defensive performance this season. Here are a couple of charts showing the Evolving Wild's expected goals against per 60 minutes (xGA_60) metric for 2018/19, along with the difference in performance from last season (2017/18). It's pretty evident that many of the Jets top forwards have poor xGA_60, and have declined from last season.

Among defensemen, only Chiarot has had a strong showing in xGA_60, with a big improvement from last season. You could see that it's understandable to pair him with an offensive D this season, based on his defensive performance in shutting down shots / chances against.

But overall, not a pretty picture...


View attachment 199495

View attachment 199493

Brian Little's performance is just Stunning.

Given the decline in performance is essentially team wide I'd think to a system tweak is I order.

 

[Sperss1997]

Much has been written about the Jets' decline in defensive performance this season. Here are a couple of charts showing the Evolving Wild's expected goals against per 60 minutes (xGA_60) metric for 2018/19, along with the difference in performance from last season (2017/18). It's pretty evident that many of the Jets top forwards have poor xGA_60, and have declined from last season.

Among defensemen, only Chiarot has had a strong showing in xGA_60, with a big improvement from last season. You could see that it's understandable to pair him with an offensive D this season, based on his defensive performance in shutting down shots / chances against.

But overall, not a pretty picture...


View attachment 199495

View attachment 199493

Ugly
If I read the graphs correctly only Roslo and Fly has improved on defense compared to last year among forwards and both were already good, while 4 defenders has improved, but only Chiarot is actually good defensively

I do not see a long post season for the Jets, unfortunately. To much has to change to be good

 

[HannuJ]

some ppl will strongly disagree, but faceoff stats are, on the whole, irrelevant.
situational faceoffs (i.e. powerplay, PK, pulled goalie)? i believe they're statistically relevant.
showing that Player A is 55% in the faceoff circle, in general, is useless and tells us nothing important.

 

[surixon]

Ugly
If I read the graphs correctly only Roslo and Fly has improved on defense compared to last year among forwards and both were already good, while 4 defenders has improved, but only Chiarot is actually good defensively

I do not see a long post season for the Jets, unfortunately. To much has to change to be good

Keep in mind the performance of the forwards plays a role in how the defense does. Morrissey and Trouba have been out with the Scheifele line most of the year and their numbers are fairly respectable given how poor the Scheifele line has been defending.

 

[JetsFan815]

This thread is pretty depressing as a fan, however sometime reality is depressing

 

[garret9]

I made a match-up with Leafs and Bruins forward line-up and thought I should post the Jets line-up in here as well
All RAPM data are from Evolving-Hockey.com. Only 2018/19 and even strength.
The attempt is to isolate a players impact and performance on the team, independant of both his teammates and opponents, compared to league average.
@garret: If RAPM data are invalid then please let me now! Some data seem rather dubious

Jets forward group impact has been mediocre. Scheifele and Wheeler dissapointing. Connor need to work things out in his own end, so does Laine.
(sorry about the puncuation mistakes in the table)

View attachment 197871

Just one thing to correct...

xGA is against (or defense)
xGPM is for AND against, not offense
xGF is for (or offense)

And obviously xGoals does not account for finishing talent, but it does count for most of shot quality.

 

[JetsFan815]

I was thinking about how some of the past Jets would fare with the current team. Here's our current top line's stats against some of the most infamous Jets lines from the past include the great GST line. Those faint of the heart need not read any further. Consider yourself warned

Season	Line	TOI	CF%	FF%	SF%	ScoringChancesFor%	HDCF%	GF%
18-19	Laine-Scheifele-Wheeler	168:45	44.21	44.94	43.84	43.98	34.98	49.71
11-12	Glass-Slater-Thorburn	439:25	41.43	44.17	43.10	42.27	43.85	34.85
11-12	Glass-Slater-Miettinen	122:04	44.31	43.60	43.21	39.18	29.27	22.08

[TBODY] [/TBODY]

I mean our top line guys are not bad players but wtf is going on?

 

[Imcanadianeh]

I was thinking about how some of the past Jets would fare with the current team. Here's our current top line's stats against some of the most infamous Jets lines from the past include the great GST line. Those faint of the heart need not read any further. Consider yourself warned

Season	Line	TOI	CF%	FF%	SF%	ScoringChancesFor%	HDCF%	GF%
18-19	Laine-Scheifele-Wheeler	168:45	44.21	44.94	43.84	43.98	34.98	49.71
11-12	Glass-Slater-Thorburn	439:25	41.43	44.17	43.10	42.27	43.85	34.85
11-12	Glass-Slater-Miettinen	122:04	44.31	43.60	43.21	39.18	29.27	22.08

[TBODY] [/TBODY]

I mean our top line guys are not bad players but wtf is going on?

It’s almost like Corsi is a limited stat that like everything else is extremely flawed and shouldn’t be blindly followed.

I mean just look at the GF% clearly shows the skill difference.