News Article: Dellow: Jets' record is great but long-term success requires better shot generation

[garret9]

Special teams and goaltending?

Those are the two largest.
There's also just changes that happen over the season with teams.

 

[garret9]

Some fun pieces I suggest reading:
Objective NHL: Even Strength Shooting Percentage
Objective NHL: Team Even Strength Shooting Talent
Objective NHL: Even Strength Outshooting and Team Quality
Driving Play: How Much of Shooting Percentage Is Skill?
Shot quality revisited: A look at the correlation between scoring chances and shot totals - NHL Numbers
Shot quality matters, but how much? - NHL Numbers
On-Ice Shooting Percentage as a Player Talent
Luck vs Shot Quality in Shooting Percentage

 

[Mathmew Purrrr Oh]

What if you hate cats?

hey!

would also like to mention that the Jets were the number one team for time spent with a lead in 14-15 and also were 4th in cf%

 

[GNP]

I dont know. Maybe. But I got tired of all the previous years where we would out Corsi and out shoot openents and lose so many games. Last year we couldn't even get games to overtime for a few more loser points. Then we would talk about our awesome 5v5 Corsi and how great of a team we are. But then we'd finish typically bottom 5ish in the league. News flash - if you finish that low you suck. I always said that the success of a team is measured in the standings. You are as good as your record. I can tell that we are a better team this year. Want to know how I know? It has nothing to do with Corsi and a lot to do with our points in the standings.

I do agree with you Jet !!!-"good post here" you only are as good as your record, and right now we're a pretty good hockey club. In the end, it's your record in the standings, and not your Corsi score or other fancy stats.

What I am saying is we can "improve" quite a lot, especially our second line point production, and putting together a good 3rd scoring line, and we have the talent to do it. A nice 3rd line would be Perrault, Copp, and Armia.

I just know when your getting outshout by 7-10 shots every game, it will cause a lot of problems with the more prolific scoring clubs. No reason why we can't improve with guys like Ehlers, Laine and Little--"they have got to get going" and be with the right linemates. I'd like to see some line shuffling.

 

[Whileee]

I'm well aware. That doesn't lessen anything I've said by any degree.
I think "some of the rest is luck" is a bit of an understatement when about half of what's missing is luck.

Corsi kills Sh% though in any split-half reliability (~ 0.1 vs 0.8 r^2) or predictive testing...

Which is kinda the whole point:
Shot volume isn't the only thing that wins you games, but it's what you have the most control over.

Reliability of a measure is not the same as its strength in predicting outcomes. So, the fact that sh% is not as reliable (i.e. repeatable) as Corsi is a different concept than the fact that Corsi is not that predictive of outcomes.

I guess the main take-home message is that Corsi is more reliable and explains more of the future results than many metrics, but it still doesn't explain much of the variance in future results.

What it tells me is that much more work needs to be done to try to understand deviance from Corsi / results associations, and just saying that it's mostly "luck" doesn't seem that satisfying to me.

I know it's simply an anecdote, but I was thinking a bit about the Jets last night and how they are achieving results. Overall, the Jets' PP seemed to struggle in all the usual areas, but a few brilliant plays by highly skilled players resulted in one chance that was worth a large number of lower quality chances (Buff keeps it in, Laine to Wheeler, Wheeler finds Scheifele in an ideal spot, and he doesn't miss from there). However much pressure and shot attempts that the Yotes had, they never generated that sort of play sequence to generate a chance. Similarly, on Connor's goal, it was a singular sequence of brilliant play by Scheifele (in an otherwise meh game by him) that set up a bang-bang goal that Connor is always going to score. Maybe shooting % is not that repeatable, but talent is talent and I'm not sure that these sorts of specific differences between teams is well-documented or studied. Goals are rare events, and hard to study.

Still, I'd feel much more comfortable if the Jets were controlling more of the shot attempts, and I have particular concerns with the poor puck movement by their D in their zone. But if you look at the distribution of shot metrics at the individual level, the Jets' main problem this season is that their 5v5 workhorses (Scheifele at 46.2% and Wheeler at 44.3%) have had bottom end metrics. Has usage changed so much? Has usage? Or is their something deficient in their quality of play? If they were both at 50% or higher, they would likely pull the entire Jets' performance much higher. There's something weird going on with them, and if and when it straightens out the Jets overall team performance would rise.

 

[garret9]

Reliability of a measure is not the same as its strength in predicting outcomes. So, the fact that sh% is not as reliable (i.e. repeatable) as Corsi is a different concept than the fact that Corsi is not that predictive of outcomes.

I guess the main take-home message is that Corsi is more reliable and explains more of the future results than many metrics, but it still doesn't explain much of the variance in future results.

What it tells me is that much more work needs to be done to try to understand deviance from Corsi / results associations, and just saying that it's mostly "luck" doesn't seem that satisfying to me.

I know it's simply an anecdote, but I was thinking a bit about the Jets last night and how they are achieving results. Overall, the Jets' PP seemed to struggle in all the usual areas, but a few brilliant plays by highly skilled players resulted in one chance that was worth a large number of lower quality chances (Buff keeps it in, Laine to Wheeler, Wheeler finds Scheifele in an ideal spot, and he doesn't miss from there). However much pressure and shot attempts that the Yotes had, they never generated that sort of play sequence to generate a chance. Similarly, on Connor's goal, it was a singular sequence of brilliant play by Scheifele (in an otherwise meh game by him) that set up a bang-bang goal that Connor is always going to score. Maybe shooting % is not that repeatable, but talent is talent and I'm not sure that these sorts of specific differences between teams is well-documented or studied. Goals are rare events, and hard to study.

Still, I'd feel much more comfortable if the Jets were controlling more of the shot attempts, and I have particular concerns with the poor puck movement by their D in their zone. But if you look at the distribution of shot metrics at the individual level, the Jets' main problem this season is that their 5v5 workhorses (Scheifele at 46.2% and Wheeler at 44.3%) have had bottom end metrics. Has usage changed so much? Has usage? Or is their something deficient in their quality of play? If they were both at 50% or higher, they would likely pull the entire Jets' performance much higher. There's something weird going on with them, and if and when it straightens out the Jets overall team performance would rise.

I'm not even going to read the whole thing yet before responding because the first paragraph annoys me that you are straw man-ing me (or just misread):

Corsi kills Sh% though in any split-half reliability (~ 0.1 vs 0.8 r^2) OR predictive testing...

No where did I say reliability is same as predicting outcomes. In fact, it is obviously distinct. So why are you responding to me as if I had?

Now, sh% poor reliability matters in this case, because it's poor reliability is why it's such a terrible predictor.

Goals has two inputs:
* Shot quantity
* Shot conversion

Shot quantity is fairly stable.
Luck is a small component of it, and it can have small sample issues (generally you want about 20-25 games for higher confidence).
Depending on which test, methodology, situation (EV, overall, etc.), etc. for research you are looking at, shot quantity is between 1/3-2/3 of what matters after accounting for "randomness." (low estimate being from 5v5 predictive testing; the high estimate from research on what drives wins... truth is probably near the middle)
That's a big deal.

Shot conversion is HIGHLY unstable.
There are four components to shot conversion: luck, shot quality, finishing talent, and goaltender talent.
I've seen most estimates of shot quality being about 10% of non-randomness. Goaltending about equal. Finishing talent is a bit more controversial, although I'd estimate slightly above shot quality.

Penalty differential is another independent factor, but I've never seen a number placed on it... yet. And then you have special teams, as well.

These arguments that are coming up are no different than what we saw with Toronto or Colorado. The good news is that, as you pointed out, the Jets have players that have drove shot differentials before, but are struggling right now (Wheeler, Scheifele) and injured players who drive them and will return (Perreault). You have to ask though if Jets are slightly shooting themselves in the foot *more than necessary* due to fixable systematic issues.

 

[GNP]

If you took just a couple stats, like- per game shots on goal, per game -shots allowed, and -"scoring percentage on shots taken," goaltending save % and Goals against--you'd have a pretty good idea on predicting if that club is successful.

Example-if you look at the top 5 goalies in the NHL- SP% etc- usually 2-4 out of those 5 clubs are in 1-st or second place in their division-"simple stuff"

As far as I'm concerned, you can put away your slide rules, geometry sets, and enjoy the game, and look at the basics. It's not rocket science here, and these are not machines your graphing--their humans, and their performances are always changing, with their circumstances. ( life or health or whatever)

 

[garret9]

If you took just a couple stats, like- per game shots on goal, per game -shots allowed, and -"scoring percentage on shots taken," goaltending save % and Goals against--you'd have a pretty good idea on predicting if that club is successful.
Example-if you look at the top 5 goalies in the NHL- SP% etc- usually 2-4 out of those 5 clubs are in 1-st or second place in their division-"simple stuff"
As far as I'm concerned, you can put away your slide rules, geometry sets, and enjoy the game, and look at the basics. It's not rocket science here, and these are not machines your graphing--their humans, and their performances are always changing, with their circumstances. ( life or health or whatever)

That's not predicting. Predicting is saying, given what we've seen, how will the teams/players perform in the future.

When we say predicting, we mean something like this:
At the twenty game mark, sort the NHL teams in order of shot differentials. Then repeat this with goal differential, standings points, win% save%, shooting%, scoring chance differential.

Which one of these list orders will be most like the team win% or goal differential over games 21 through 82?

The answer is Corsi. (EDIT: there are things better than Corsi. I'm just specifically discussing the things I listed above)

Performances change, but hockey is not a coin flip. There are signals. There are things that can help. There are reasons why all NHL teams are investing into these things despite being better at looking at the basics better than the average person here. It still is helpful and illuminating.

 

[JetsFan815]

hey!

would also like to mention that the Jets were the number one team for time spent with a lead in 14-15 and also were 4th in cf%

I miss the 14-15 team . I was the best team TNSE ever built and apparently after 4 playoff games decided that that group wasn't worth keeping together. Sad.

 

[Whileee]

I'm not even going to read the whole thing yet before responding because the first paragraph annoys me that you are straw man-ing me (or just misread):

Corsi kills Sh% though in any split-half reliability (~ 0.1 vs 0.8 r^2) OR predictive testing...

No where did I say reliability is same as predicting outcomes. In fact, it is obviously distinct. So why are you responding to me as if I had?

Now, sh% poor reliability matters in this case, because it's poor reliability is why it's such a terrible predictor.

Goals has two inputs:
* Shot quantity
* Shot conversion

Shot quantity is fairly stable.
Luck is a small component of it, and it can have small sample issues (generally you want about 20-25 games for higher confidence).
Depending on which test, methodology, situation (EV, overall, etc.), etc. for research you are looking at, shot quantity is between 1/3-2/3 of what matters after accounting for "randomness." (low estimate being from 5v5 predictive testing; the high estimate from research on what drives wins... truth is probably near the middle)
That's a big deal.

Shot conversion is HIGHLY unstable.
There are four components to shot conversion: luck, shot quality, finishing talent, and goaltender talent.
I've seen most estimates of shot quality being about 10% of non-randomness. Goaltending about equal. Finishing talent is a bit more controversial, although I'd estimate slightly above shot quality.

Penalty differential is another independent factor, but I've never seen a number placed on it... yet. And then you have special teams, as well.

These arguments that are coming up are no different than what we saw with Toronto or Colorado. The good news is that, as you pointed out, the Jets have players that have drove shot differentials before, but are struggling right now (Wheeler, Scheifele) and injured players who drive them and will return (Perreault). You have to ask though if Jets are slightly shooting themselves in the foot *more than necessary* due to fixable systematic issues.

Sorry, I reacted because I think a lot of people don't distinguish between prediction and reliability. I know they are linked and I know you understand this, but you went from the issue of prediction to citing reliability. My apologies if you thought it was a straw man argument. It wasn't my intention.

I do think the Jets have a system problem, but I'm not sure the nature and how severe. The fact that normal drivers of positive shot metrics are now dragging team metrics now is a curious phenomenon, and I don't know how sustainable that will be. My own feeling is that the Jets D are too focused on moving the puck with their legs, and not enough with smart passing. I also think that they are clamping down D to protect the slot to a fault, after a season where they were way too loose. In the offensive zone they are very reluctant to take low percentage shots, perhaps because they are emphasizing zone time. All of these might contribute to their problems.

 

[Whileee]

I miss the 14-15 team . I was the best team TNSE ever built and apparently after 4 playoff games decided that that group wasn't worth keeping together. Sad.

I never thought that team had the raw talent and quickness to really compete with the best. Now, they need to find a way to get a much more talented line-up to play a consistent and effective system.

 

[ps241]

Sorry, I reacted because I think a lot of people don't distinguish between prediction and reliability. I know they are linked and I know you understand this, but you went from the issue of prediction to citing reliability. My apologies if you thought it was a straw man argument. It wasn't my intention.

I do think the Jets have a system problem, but I'm not sure the nature and how severe. The fact that normal drivers of positive shot metrics are now dragging team metrics now is a curious phenomenon, and I don't know how sustainable that will be. My own feeling is that the Jets D are too focused on moving the puck with their legs, and not enough with smart passing. I also think that they are clamping down D to protect the slot to a fault, after a season where they were way too loose. In the offensive zone they are very reluctant to take low percentage shots, perhaps because they are emphasizing zone time. All of these might contribute to their problems.

I do think it is the system they are deploying. I think PMo is fine attempting to trade off a higher number of lower quality shot attempts while trying to protect the danger area more than in the past. I like winning but I think the team is being poorly coached and riding a hot streak by the goalies. Hopefully the goaltending has a really good season. This group of skaters can play better.

 

[Jimby]

A couple of things stand out from last season. Scheifele has a lot more Takeaways and Hits this year over last (/60). His linemate Wheeler has a LOT more Giveways and the lowest rate of Takeaways of his career.

Quick question... Any theories as to why Scheifele and Wheeler are getting buried in shot metrics? I doubt they've regressed and yet they are performing well below a number of other Jets' forwards. Some of it must be usage, but is usage so much different from previous seasons?

 

[Whileee]

I do think it is the system they are deploying. I think PMo is fine attempting to trade off a higher number of lower quality shot attempts while trying to protect the danger area more than in the past. I like winning but I think the team is being poorly coached and riding a hot streak by the goalies. Hopefully the goaltending has a really good season. This group of skaters can play better.

I think that Maurice and his top players need to get to a better balance. I think the message that has been hammered home from the beginning of the season is to play a "tighter" defensive game. You can see this in the defensive zone with a compact structure leaving a lot of space to the outside. Also, player positioning in the defensive zone tends to be conservative, with the Jets being more tentative in pursuing the puck and letting opponents win the puck more frequently. I also think that the Jets have tended to be more conservative in moving the puck in their own zone. They aren't as crisp or creative with passes and tend to try to protect the puck more and look for an opportunity to "clear" rather than initiate a break-out. In the offensive zone, I think they are really trying to avoid pinches and giving up odd-man breaks (which is good), but this relieves some of the pressure from their fore-check. As I've noted before, I also think that they are focusing a lot on puck possession in the offensive zone. This seems to be particularly something for Scheifele and Wheeler, who are really looking for good opportunities.

Here are some interesting trends... Wheeler, and especially Scheifele and Little are way down in iCF/60 this season, while Laine has jumped and Ehlers and Lowry are about the same. I don't think these players are less effective, I think they have changed their style somewhat. It doesn't seem like a purely system issue, because Laine, Ehlers and Lowry (and most others) have maintained their shot production. On the other hand, it's possible that the coaches are focusing on having their forwards with tough match-ups on being conservative with shot selection as a way of blunting opportunities for top players on the opposing team...

Anyway, a number of hypotheses to work with...

 

[truck]

Anyone that watched the Yotes game will realize how often the Jets try to make passes or other plays instead of shooting. Scheifele's line does this a lot. They also get pinned in their own zone due to soft D sometimes, so....

Is the assertion here that Scheifele has discovered a more effective and efficient style if play than pretty much everyone else in the entire league? Is the assertion that Scheifele has found a sustainable way to outplay shot metrics - something that even Sidney Crosby has not been able to accomplish.

 

[truck]

I never thought that team had the raw talent and quickness to really compete with the best. Now, they need to find a way to get a much more talented line-up to play a consistent and effective system.

The interesting thing to me is that most of the players that they've added (save for Laine and Tanev) were already in the system.

That said, I never saw that tank as intentional. They eventually made an effort to support the tank, but I think they entered that season fully expecting to step forward. Same as ever, they were sabotaged by their unwarranted unwaivering faith in a bad goalie.

 

[Whileee]

Is the assertion here that Scheifele has discovered a more effective and efficient style if play than pretty much everyone else in the entire league? Is the assertion that Scheifele has found a sustainable way to outplay shot metrics - something that even Sidney Crosby has not been able to accomplish.

Empirical observation - Scheifele's "style" has changed, in that his shot attempts have reduced quite dramatically (almost 50% since 2 years ago). I say that this is a "style" issue because I don't think his talent has declined. If anything, he's a more effective player in all of the areas that should drive shot metrics (puck retrieval, puck possession, etc.). Also, he's playing with one of the best drivers of shots in the NHL (Wheeler), and I don't think Wheeler's skill has declined.

Watching him carefully, I think that the style is one that is very conservative in terms of taking low-percentage shot attempts. So, I don't think it's a diminution in skill and performance per se, but rather a deliberate approach to avoid low-danger shots. I might be completely wrong, but that's what I see and the data seem to back it up.

Whether this will be effective in the long run is highly questionable, as you've pointed out. I expect that we'll see some normalization of the shot metrics for Scheifele, Little and Wheeler, which will improve the overall team metrics. But I also expect that Scheifele's style will result in higher shooting percentages than one might expect, due both to shot selection and skill.

 

[Aavco Cup]

Is the assertion here that Scheifele has discovered a more effective and efficient style if play than pretty much everyone else in the entire league? Is the assertion that Scheifele has found a sustainable way to outplay shot metrics - something that even Sidney Crosby has not been able to accomplish.

I think it's this

 

[Ducky10]

I think it's this

No doubt......on so many levels.

 

[Whileee]

The interesting thing to me is that most of the players that they've added (save for Laine and Tanev) were already in the system.

That said, I never saw that tank as intentional. They eventually made an effort to support the tank, but I think they entered that season fully expecting to step forward. Same as ever, they were sabotaged by their unwarranted unwaivering faith in a bad goalie.

I agree that I don't think they deliberately tanked, but Chevy and Maurice were both pretty open after the season that they thought that the team needed to add more skill and quickness to take the next step. They were sending a clear signal that they weren't going to maintain the status quo. At the time I shared the skepticism about the likelihood that they could take the next step without reshaping the roster with more skill and quickness. Happily they already had some top talent developing in their system or new to the roster (Scheifele, Trouba, Morrissey, Ehlers, Hellebuyck), and they were very fortunate to get Connor in 2015 and the #2 pick in the lottery the next year. Another factor might have been their realization that they could likely only keep one of Ladd or Buff, so they'd be looking to add more prospect depth to supplement the core.

 

[Whileee]

I think it's this

No doubt......on so many levels.

...oooh, that's a bit harsh (and unexpected from you, AC).

Please share what you think explains the dramatic reduction in shot attempts by Scheifele. This season. Did he somehow turn into Burmistrov this season?

The argument isn't so much about whether his style will yield sustainable positive results, but more about what's driving it and to what extent it is system driven.

 

[Aavco Cup]

...oooh, that's a bit harsh (and unexpected from you, AC).

Please share what you think explains the dramatic reduction in shot attempts by Scheifele. This season. Did he somehow turn into Burmistrov this season?

The argument isn't so much about whether his style will yield sustainable positive results, but more about what's driving it and to what extent it is system driven.

They are spending an enormous amount of time in their own end this year. Tough to generate shots that far from the net. I think it's that simple.

That image was intended in a general sense for people who are trying to rationalize the early results. I do think they are unsustainable if it continues down this path. I do think they can fix things though. I have more faith in that than the sustainability of us "Avalanching" the league for 82 games.

 

[TheRocketFlash]

Didn't read through the 6 pages of responses, but can it be considered that the Jets don't take a high number of pure shots because we have multiple elite shooters who opt for quality chances instead of just peppering goalies with low percentage opportunities? Below, find the top 10 in terms of shooting percentage for all of last season (so a reasonable sample size). You will see the likes of Oshie, Crosby, Malkin etc in the list, but will also see 3 (three!) Jets in the top 8. While I do agree with many points in the article, I think the Jets maintaining their current 9.8% shooting is completely sustainable since we all know Maurice plays his top two lines a ton and these are (obviously) our best shooters. Maintaining the ~.927 save percentage may be the bigger question mark.

1 Oshie, T.J. WSH
23.1%
2 Byron, Paul MTL
22.9%
3 Anisimov, Artem CHI
21.0%
4 Scheifele, Mark WPG
20.0%
5 Johansson, Marcus WSH
18.6%
5 Rakell, Rickard ANA
18.6%
7 Lee, Anders NYI
17.8%
8 Little, Bryan WPG
17.6%
8 Laine, Patrik WPG
17.6%
10 Crosby, Sidney PIT
17.3%
Malkin, Evgeni PIT
17.3%
10 Marchand, Brad BOS
17.3%

As always, GJG!

 

[pucka lucka]

What's driving it is they willingly give up possession of the puck way more than they should. They give away the puck on line changes WAY more than they need to. They chip the puck into the corner WAY more than they should be. COrner battles!

Ehlers seems to be the only player capable or maybe renegade enough to not throw away the puck on a line change every single time.

 

[Jimby]

It could be as simple as looking Scheifele having dramatically increased his takeaways and hitting this year perhaps out of necessity considering his partner Wheeler has turned into a turnover machine who doesn't hit and can't take the puck away from anyone. Doesn't leave a lot of time for shot attempts.

They are spending an enormous amount of time in their own end this year. Tough to generate shots that far from the net. I think it's that simple.

That image was intended in a general sense for people who are trying to rationalize the early results. I do think they are unsustainable if it continues down this path. I do think they can fix things though. I have more faith in that than the sustainability of us "Avalanching" the league for 82 games.