2018-19 stats and underlying metrics thread

[JetsFan815]

Lowry has has been producing good results with Tanev/Copp at 5 on 5 for a while now but even so Perreault's impact on the Lowry line has been absolutely unreal. Look at that shot rate go through the roof as soon mustard color starts showing up in Lowry's linemate graph. Also shows how wasted he is in limited minutes on the 4th line.

While Perreault has done an admirable job, his services are more desperately needed to get the 2nd line out of the blackhole that its in. Just put him on that line Maurice... he will solve that line's problems overnight.

 

[Whileee]

An interesting conversation with Maurice that veered into a discussion of shot attempts (Corsi) vs. shot quality (starting around 6:30, I think).



Clearly, Maurice and the team is tracking shot attempts and shot quality (location, situations), and they see trade-offs between taking low percentage shots (like Carolina) and offensive zone possession. I think that has been reflected in the Jets' play over the past couple of season, where defensively they try to limit chances from high danger areas and try to avoid taking too many low percentage shots that can result in lost possession and break opportunities for the opposition. In general, the Jets tend not to take very many shots from the outside, and they tend to push shot attempts towards the perimeter (see below for this season, and last season). At first glance, it seems that the Jets have been better at generating shots from the centre of the ice this season compared to last season, but have been a bit less effective at shutting down shot attempts from the slot area. That might reflect the loss of Enstrom, and the shaky start this season from Myers and Kulikov.

 

[garret9]

CAR isn't (always) a low quality issue (depending on the year)... it's a low finishing issue compounded with consistently bad goaltending and luck (again, depending on the year):

CAR may take fewer shots from inside *relative* to the outside, but they take *more* from prime areas than a good portion of teams.

 

[JetsWillFly4Ever]

Lowry has has been producing good results with Tanev/Copp at 5 on 5 for a while now but even so Perreault's impact on the Lowry line has been absolutely unreal. Look at that shot rate go through the roof as soon mustard color starts showing up in Lowry's linemate graph. Also shows how wasted he is in limited minutes on the 4th line.

While Perreault has done an admirable job, his services are more desperately needed to get the 2nd line out of the blackhole that its in. Just put him on that line Maurice... he will solve that line's problems overnight.

100% agree with this.

Third line has been great with Perreault but we can plug someone else in there and they will still be decent, need to move Perreault into the top 6 and try to get the second line going. I'm not sure how I would do it, Maurice will never break up Scheifele/Wheeler but I wouldn't be opposed to something like Perreault with Scheifele and Laine then Connor - Little - Wheeler as a 2nd line. I think Wheeler and Little can push the play enough and Connor can finish, then Perreault and Scheif can push play and Laine can finish. Then when Ehlers comes back either slide Connor onto the third line or slide down Perreault if you are confident enough in Ehlers - Scheifele - Laine as a line.

 

[ps241]

Nice work Whileee and JetsFan815. Man I am going to miss Matty P when he is gone and so will the Jets.

 

[Whileee]

CAR isn't (always) a low quality issue (depending on the year)... it's a low finishing issue compounded with consistently bad goaltending and luck (again, depending on the year):

CAR may take fewer shots from inside *relative* to the outside, but they take *more* from prime areas than a good portion of teams.

I agree about the finishing talent, but I would also contend that shot position doesn't necessarily translate directly into danger. So, I would hypothesize that shots from "prime areas" aren't all equivalent, and that teams differ in terms of the actual danger of their "prime area" shots.

Did Skinner become a more talented shooter when he left Carolina? What about E Staal? Lindholm? All three experienced significant jumps in their iSH% when they joined new teams. Maybe some of that is just luck (bad in Carolina, good elsewhere), but maybe it's also that what look like scoring chances from "prime areas" spot map, aren't all the same.

Here are the individual Corsi shooting percentages for those three players on 5v5 iHDCF (individual High Danger CF), for their last few seasons in Carolina, and their subsequent seasons...

Staal - CAR (14/15 and 15/16): 12% / NYR/MIN (15/16 to 18/19): 17%

Skinner - CAR (14/15 to 17/18): 22% / BUFF (18/19): 40%

Lindholm - CAR (14/15 to 17/18): 15% / CGY (18/19): 23%

Hypothesis 1: Players are unlucky when they play in Carolina, but not elsewhere.
Hypothesis 2: This is just random variation, notwithstanding the general observation that Carolina has a very low SH%.
Hypothesis 3: Measurement of shot quality / danger is imprecise, and the shot quality from Carolina is overestimated by sub-par classification methods.

I'm sure there are other hypotheses that could be proposed...

 

[Gm0ney]

An interesting conversation with Maurice that veered into a discussion of shot attempts (Corsi) vs. shot quality (starting around 6:30, I think).



Clearly, Maurice and the team is tracking shot attempts and shot quality (location, situations), and they see trade-offs between taking low percentage shots (like Carolina) and offensive zone possession. I think that has been reflected in the Jets' play over the past couple of season, where defensively they try to limit chances from high danger areas and try to avoid taking too many low percentage shots that can result in lost possession and break opportunities for the opposition. In general, the Jets tend not to take very many shots from the outside, and they tend to push shot attempts towards the perimeter (see below for this season, and last season). At first glance, it seems that the Jets have been better at generating shots from the centre of the ice this season compared to last season, but have been a bit less effective at shutting down shot attempts from the slot area. That might reflect the loss of Enstrom, and the shaky start this season from Myers and Kulikov.

View attachment 173801

View attachment 173803

I dunno...the Jets don't seem to be getting in close in the offensive zone like they were last year. More of the focus seems to be in "Scheifele's Office" in the mid-high slot (and the Byfuglien-Myers Nexus on the right point). Still a good place to shoot from, but that low slot/crease area is a blue hole of badness.

And on defense, the Royal Road has opened right up this year. Last year nothing was happening in close in the Jets zone. It wasn't all Enstrom. The team's defensive play has been a bit suspect all season. I think Hellebuyck is kicking out way more rebounds - last year he was absorbing everything coming his way. Small Wade Flaherty adjustments have undone him!

 

[Whileee]

I dunno...the Jets don't seem to be getting in close in the offensive zone like they were last year. More of the focus seems to be in "Scheifele's Office" in the mid-high slot (and the Byfuglien-Myers Nexus on the right point). Still a good place to shoot from, but that low slot/crease area is a blue hole of badness.

And on defense, the Royal Road has opened right up this year. Last year nothing was happening in close in the Jets zone. It wasn't all Enstrom. The team's defensive play has been a bit suspect all season. I think Hellebuyck is kicking out way more rebounds - last year he was absorbing everything coming his way. Small Wade Flaherty adjustments have undone him!

Jets overall shot heat map certainly seems to be more central and closer this season than last season...

I think that overall shot suppression has been poorer this season, and Enstrom was very good last season in that area.

 

[Ace]

I agree about the finishing talent, but I would also contend that shot position doesn't necessarily translate directly into danger. So, I would hypothesize that shots from "prime areas" aren't all equivalent, and that teams differ in terms of the actual danger of their "prime area" shots.

Did Skinner become a more talented shooter when he left Carolina? What about E Staal? Lindholm? All three experienced significant jumps in their iSH% when they joined new teams. Maybe some of that is just luck (bad in Carolina, good elsewhere), but maybe it's also that what look like scoring chances from "prime areas" spot map, aren't all the same.

Here are the individual Corsi shooting percentages for those three players on 5v5 iHDCF (individual High Danger CF), for their last few seasons in Carolina, and their subsequent seasons...

Staal - CAR (14/15 and 15/16): 12% / NYR/MIN (15/16 to 18/19): 17%

Skinner - CAR (14/15 to 17/18): 22% / BUFF (18/19): 40%

Lindholm - CAR (14/15 to 17/18): 15% / CGY (18/19): 23%

Hypothesis 1: Players are unlucky when they play in Carolina, but not elsewhere.
Hypothesis 2: This is just random variation, notwithstanding the general observation that Carolina has a very low SH%.
Hypothesis 3: Measurement of shot quality / danger is imprecise, and the shot quality from Carolina is overestimated by sub-par classification methods.

I'm sure there are other hypotheses that could be proposed...

I can actually answer what happened to Skinner. I saw a shot chart a while back showing where his average shot distance and goal distance was coming from in Buffalo compared to with Carolina. It’s basically the Eichel effect. He gets shots off from so much closer in on net now that he is more effective. I wish I knew where I saw them so I could post it here because it was legit nuts.

I don’t know about the other players at all. But the Skinner thing makes some sense when you see where his shots are coming from now. He’s super talented and now gets attempts from much better scoring positions regularly.

 

[Whileee]

I can actually answer what happened to Skinner. I saw a shot chart a while back showing where his average shot distance and goal distance was coming from in Buffalo compared to with Carolina. It’s basically the Eichel effect. He gets shots off from so much closer in on net now that he is more effective. I wish I knew where I saw them so I could post it here because it was legit nuts.

I don’t know about the other players at all. But the Skinner thing makes some sense when you see where his shots are coming from now. He’s super talented and now gets attempts from much better scoring positions regularly.

Very interesting. Yet when you look at the HDCF on a site like Natural Stat Trick, you don't see a big difference between his time in Carolina vs. Buffalo. That suggests that the NST measures aren't sensitive enough to differentiate the actual shot quality.

 

[garret9]

Very interesting. Yet when you look at the HDCF on a site like Natural Stat Trick, you don't see a big difference between his time in Carolina vs. Buffalo. That suggests that the NST measures aren't sensitive enough to differentiate the actual shot quality.

Binning data is bad

 

[Whileee]

Binning data is bad

Who has better data? Is it publicly available? What are the best current methods to measure shot quality?

 

[Psych0dad]

Nice work Whileee and JetsFan815. Man I am going to miss Matty P when he is gone and so will the Jets.

Yeah he is a much more important piece in Jets than most realize. I hope Jets manage to keep him and play him to his strengths. He could play 1st line. Would, if I had a say.

 

[lomiller1]

Deja-vu all over again.

Yesterday I was explaining to a co-worker that Ohms Law isn’t really a law so much as a slight bias to the random movement of electrons when they are exposed to an electric field.

Last night I was debating with a climate science denier that couldn’t understand the concept that year to year temperate changes are mostly random variations around an underlying trend and that even if they buck that trend for a decade or more it doesn’t mean the trend has changed (AKA there was no "pause" in global warming)

A few min ago I was discussing investments (vs bitcoin) and pointing out how investments with real earnings do not move randomly up/down they ultimately follow a trend determined by the earnings. The value at any instant may be above/below that trend but the investment value will untimely follow along with where the earnings trend leads.

Now I’m here…

Not that it's a topic right now, but I couldn't help but see the parallel in how so many discussion on hockey stats go. But "hockey is too complicated to understand with statistics!" right?

 

[Gil Fisher]

Deja-vu all over again.

Yesterday I was explaining to a co-worker that Ohms Law isn’t really a law so much as a slight bias to the random movement of electrons when they are exposed to an electric field.

Last night I was debating with a climate science denier that couldn’t understand the concept that year to year temperate changes are mostly random variations around an underlying trend and that even if they buck that trend for a decade or more it doesn’t mean the trend has changed (AKA there was no "pause" in global warming)

A few min ago I was discussing investments (vs bitcoin) and pointing out how investments with real earnings do not move randomly up/down they ultimately follow a trend determined by the earnings. The value at any instant may be above/below that trend but the investment value will untimely follow along with where the earnings trend leads.

Now I’m here…

Not that it's a topic right now, but I couldn't help but see the parallel in how so many discussion on hockey stats go. But "hockey is too complicated to understand with statistics!" right?

Who are you who is so wise in the ways of science?

[he's a witch. burn him.]

 

[DuckyChuck]

Binning data is bad

Speaking of quality data, do you know how the NHL plans to use the data they collected on Thursday during the Knights-Sharks game?

NHL tests player-tracking technology that could...

And not just that particular game, but all games going forward that use that tracking. They mention how it relates mostly to VR experience's and gambling, but are their any grumblings among those in the know that they will open it up to statistician's and entities outside the NHL and their partner's?

I'm sure you're just salivating over the prospects of such a rich data set.

Thanks,

 

[JetsUK]

Speaking of quality data, do you know how the NHL plans to use the data they collected on Thursday during the Knights-Sharks game?

NHL tests player-tracking technology that could...

And not just that particular game, but all games going forward that use that tracking. They mention how it relates mostly to VR experience's and gambling, but are their any grumblings among those in the know that they will open it up to statistician's and entities outside the NHL and their partner's?

I'm sure you're just salivating over the prospects of such a rich data set.

Thanks,

Can't speak to the NHL's plans for data usage, but I work with this kind and level of tech in the lab as part of my day job, so if there are specific questions related to the tech and/or the sorts of data it's able to generate I'm happy to pitch in. I suspect the league will look to license some portion of the data generated for gaming, and suspect the players will want to include this stream in their renegotiation of the CBA (though I'm no IP lawyer). The layers of data provided are certainly analytically rich, and they can be exported in a range of time-sensitive formats, depending on intended usage.

 

[Whileee]

Speaking of quality data, do you know how the NHL plans to use the data they collected on Thursday during the Knights-Sharks game?

NHL tests player-tracking technology that could...

And not just that particular game, but all games going forward that use that tracking. They mention how it relates mostly to VR experience's and gambling, but are their any grumblings among those in the know that they will open it up to statistician's and entities outside the NHL and their partner's?

I'm sure you're just salivating over the prospects of such a rich data set.

Thanks,

This should be a gold-mine for innovative and thoughtful analysts.

 

[DuckyChuck]

This should be a gold-mine for innovative and thoughtful analysts.

Just thinking of how it can affect goalie evaluation is insane.

Every shot by location, speed, player combined with relative distance to the goalie and how far they have to move to make the save. All of the sudden, not all saves are equal.

It will be interesting if they can include an accuracy component.

 

[lomiller1]

This should be a gold-mine for innovative and thoughtful analysts.

My guess is that it we won't see nearly as much informative analysis as we do data-mining for things to support a predetermined conclusion.

 

[garret9]

Who has better data? Is it publicly available? What are the best current methods to measure shot quality?

xG > binning data.
Behind the Numbers: The issues with binning, QoC, and scoring chances

 

[garret9]

This should be a gold-mine for innovative and thoughtful analysts.

And a goal-mine for bad analysis too.

 

[garret9]

Speaking of quality data, do you know how the NHL plans to use the data they collected on Thursday during the Knights-Sharks game?

NHL tests player-tracking technology that could...

And not just that particular game, but all games going forward that use that tracking. They mention how it relates mostly to VR experience's and gambling, but are their any grumblings among those in the know that they will open it up to statistician's and entities outside the NHL and their partner's?

I'm sure you're just salivating over the prospects of such a rich data set.

Thanks,

See basketball and that gives you a good idea of where this is going and what you cna do.

 

[garret9]

Jeff Skinner xFSh% / actual FSh%
2018-19: 7.3 / 15.4
2017-18: 6.6 / 6.1
2016-17: 7.8 / 9.3
2015-16: 7.6 / 7.3

This year looks not that abnormal in the underlying numbers and more just a good year for Skinner.

 

[Maukkis]

NHL Playoff Odds -MoneyPuck Analytics Based Playoff & Cup Odds

We stand at

- 95.6% chance of making the playoffs! And then...
- 38.7% chance of making round 2
- 15.3% chance of reaching the WCF...
- 5.7% chance of making the finals...
- ...and 2.2% at winning the whole thing.

Fifteen teams have a better chance of winning it all than we do.