I know you’re not much of a prospects guy, but have you been watching the world juniors? If so who have you been impressed by the most? I’ve really enjoyed watching Elmer Soderblom and Brad Lambert. Never even heard of them before the tournament
Hope I'm not too late! What do you think Shea Theodore's next few seasons will look like? Will he get as much PP time as Pietrangelo, and will he maintain, or even exceed, the kind of impact he had last season?
As you might know if you read my piece, I have a tough time getting a hold on him. Certainly in the past two seasons he's been one of the league's best playdrivers. I expect with Pietrangelo in the picture he'll keep getting sheltered. The real test will be his on-ice finishing numbers. If the pattern continues next year I think we can conclude that something is definitely going on there.
Corey Perry showed he had something left in the tank in the playoffs. But his regular season numbers were quite poor. Is there a chance he contributes something meaningful to the Habs?
I'd say so. His finishing numbers weren't exactly stellar but his playdriving hasn't really declined and in fact he's even gotten a little better defensively. I like the pickup.
What do you think of the devils going into this season? Will any players surprise us (good or bad)? What are your thoughts on Nico Hischier and his ability to emerge as an elite player?
I'm higher on them than most people but there's a lot of wild cards there. Blackwood and Crawford is a tandem I have ranked 7th in the league. But goalies are goalies. Murray is an excellent top pairing defensive defenceman. But injuries. Hughes should take a leap. But how much better will his shot be and who will he play with. I don't have them as last in the division but I could see them there.
I tend to care more about “predictive” stats than the “descriptive” WAR models we’ve been seeing a lot of, especially when it comes to evaluating defencemen.
How much should I take out of an EV GAR metric when I know the target variable (goals) is super volatile in nature?
Great question. That is a very valid concern about these stats in particular and I share your issues when using them to discuss defencemen.
EV GAR is based entirely on goals for, as you mentioned. This makes it extremely volatile to luck - tests that I've done on it found a 0.8+ R^2 with on-ice shooting percentage. Considering that it comprises the majority of WAR (WAR and oiSh% had a ~0.6 r^2) that is cause for concern. This is magnified with defencemen, who obviously don't have much to do with whether the chances go in or not.
This is one of the reasons why I've moved away from using GAR. The model I'm using now, built by Patrick Bacon, targets two things offensively - xGF driving and finishing about replacement. So while luck can play a factor in a part of that (finishing), at the very least the luck is limited to individual shooting outcomes rather than on-ice shooting outcomes. Patrick also uses a prior, which significantly improves the repeatability of his play-driving metrics.
On my cards (and in my writing), I remove finishing entirely for defencemen from these stats because it's basically random for them. So when you see me use WAR for defencemen, it is based purely on scoring chance impacts and not goals.
To stop myself from rambling any more - I'm increasingly finding the best results come from prior-informed models that remove teammate finishing from the equation, and in the case of defencemen remove finishing altogether.
That would be a nice fit, I assume you're thinking of it as a Niskanen replacement. I'm not ready to call him cooked after one partial season after a decade of consistency but I guess we'll have to see.
Do you think if Barzal played on the Lightning he'd have better results than Point? Seems like Barzal doesn't get the credit he deserves because he plays in an Islanders system that doesn't maximize his talents.
I hear that a lot from people but I'm very skeptical. Watching him play it seems pretty abundantly evident that Barzal controls what happens in the offensive zone a lot more than the Islanders' system does - you don't see Ross Johnston circling the zone with the puck. Barzal's main problems are that he doesn't have a very good shot and his teammates for whatever reason don't put away the chances he gives them. Dazzling talent but he's not in the top tier for me.
Dach seems like a sky's the limit kind of guy. Nylander's underlying numbers aren't really impressive and prospect people seem a bit down on him so maybe "low top six"?
It'd be nice to see Kase put up numbers with Pastrnak out but I'll say Bjorkstrand will explode. Jack Hughes could be another one, honestly. At the very least I expect him to far eclipse his numbers from last year.
Yea. Lets see if coach Dave galaxy brains this lineup. How do you assess the chances of regression across the Rangers lineup? We had some guys with career years (fox, panarin, zib). How worrying should that be?
Worrying. I have them out of the playoffs even assuming that Shesterkin can repeat and scale up his rookie results. Fox is one thing because his results were all underlying numbers-based, but Panarin and Zibanejad are in for regular old on-ice shooting regression.
Yea. I hope Igor is durable enough, because that starting role is difficult. Also Georgiev has had a very weird career so far. last year he was atrocious at the 5v5 but made it up at 5v4 for the most part. Yea, Fox's underlying numbers seem repeatable. But I assume there has to be some regression? Can you imagine the defense last year without him? Zib also had a weird year with elite shooting, but horrible defensive metrics. The other two wildcards are Kreider and Buch. The latter looks like he is likely gone unfortunately.
Hi I am a high school senior who is a fan of hockey and am starting to become more and more interested in the world of statistics (I am taking an AP Statistics class right now and really like it). I was wondering where do you think someone who has a basic understanding of statistics should start to dive in to hockey analytics? I also am new here and really like your articles (I am a fan of the New York Rangers and liked your Kaapo Kakko article).
It depends what you're looking to explore - to understand what the stats mean, places like the Athletic have really great explainers (like Charlie O'Connor's Analytics Primers). If you're more into the nuts and bolts hard stats stuff, the archives of websites like HockeyGraphs are probably the way to go.
how frustrated should i be that kulak and mete are now projected to be either on the bottom pair or the taxi squad, and theyre replaced by edmundson and chiarot?
I would say very frustrated. The Habs seem to be doubling down what Chiarot managed to achieve in the first half of the season and neglecting how poor he was down the stretch and in the playoffs. Maybe Edmundson can do something with Petry but Kulak and Mete definitely have a better track record recently.
As a Habs fan, I am very frustrated by what they're doing to Mete. He could've been signed to a long term dirt cheap contract this year. (like 5x2M$) instead he's likely gone within a year and they overpaid for Edmundson. I hope Mete gets time on the PP at least.
Chara can play. Although he was brutal in the playoffs. The only issue I have with Chara as a signing for the Bruins would be the temptation to play him up in the lineup again, when he really should be sheltered heavily. But if they're willing to manage his minutes it'd be a nice add.
It really is up to them. Their moves this summer did not give me much faith that their player identification at the pro level is strong at all, but prospect people seem optimistic about their pool. In theory they should be revving up for stage 2 of the rebuild but that's not where their summer acquisitions will get them.
How do you grade Jack Roslovic and where do you see him playing after his trade request? Would he be a fit on the Rangers?
What impact do you think Smith could have on the Bruins? Where do you think in the line up would be his optimal placement?
I know you’re not much of a prospects guy, but have you been watching the world juniors? If so who have you been impressed by the most? I’ve really enjoyed watching Elmer Soderblom and Brad Lambert. Never even heard of them before the tournament
Which player without a contract would you sign? Coyotes only have 2 defenders signed for in 2 years, who should they keep from their defensive body?
Any indicators under-the-hood that Pavel Zacha is ready to bust loose, or.....probably is what it is 4 full seasons in?
Hope I'm not too late! What do you think Shea Theodore's next few seasons will look like? Will he get as much PP time as Pietrangelo, and will he maintain, or even exceed, the kind of impact he had last season?
As you might know if you read my piece, I have a tough time getting a hold on him. Certainly in the past two seasons he's been one of the league's best playdrivers. I expect with Pietrangelo in the picture he'll keep getting sheltered. The real test will be his on-ice finishing numbers. If the pattern continues next year I think we can conclude that something is definitely going on there.
As a fairly brand new subscriber I missed that piece, just found it and will read it now!
That was an incredible read. Absolutely stunning and fascinating. I've gotta go back and read a lot more of your articles. Thanks for this!
Corey Perry showed he had something left in the tank in the playoffs. But his regular season numbers were quite poor. Is there a chance he contributes something meaningful to the Habs?
I'd say so. His finishing numbers weren't exactly stellar but his playdriving hasn't really declined and in fact he's even gotten a little better defensively. I like the pickup.
What do you think of the devils going into this season? Will any players surprise us (good or bad)? What are your thoughts on Nico Hischier and his ability to emerge as an elite player?
I'm higher on them than most people but there's a lot of wild cards there. Blackwood and Crawford is a tandem I have ranked 7th in the league. But goalies are goalies. Murray is an excellent top pairing defensive defenceman. But injuries. Hughes should take a leap. But how much better will his shot be and who will he play with. I don't have them as last in the division but I could see them there.
Thanks! Everyone in NJ pretty much hit rock bottom last year so any sign of not being dead last in the division is progress.
Would also love to hear your thoughts on Nico Hischier - do the numbers indicate he’s a 40/50 point player or does have a higher ceiling?
I tend to care more about “predictive” stats than the “descriptive” WAR models we’ve been seeing a lot of, especially when it comes to evaluating defencemen.
How much should I take out of an EV GAR metric when I know the target variable (goals) is super volatile in nature?
Great question. That is a very valid concern about these stats in particular and I share your issues when using them to discuss defencemen.
EV GAR is based entirely on goals for, as you mentioned. This makes it extremely volatile to luck - tests that I've done on it found a 0.8+ R^2 with on-ice shooting percentage. Considering that it comprises the majority of WAR (WAR and oiSh% had a ~0.6 r^2) that is cause for concern. This is magnified with defencemen, who obviously don't have much to do with whether the chances go in or not.
This is one of the reasons why I've moved away from using GAR. The model I'm using now, built by Patrick Bacon, targets two things offensively - xGF driving and finishing about replacement. So while luck can play a factor in a part of that (finishing), at the very least the luck is limited to individual shooting outcomes rather than on-ice shooting outcomes. Patrick also uses a prior, which significantly improves the repeatability of his play-driving metrics.
On my cards (and in my writing), I remove finishing entirely for defencemen from these stats because it's basically random for them. So when you see me use WAR for defencemen, it is based purely on scoring chance impacts and not goals.
To stop myself from rambling any more - I'm increasingly finding the best results come from prior-informed models that remove teammate finishing from the equation, and in the case of defencemen remove finishing altogether.
** I should clarify that it is EVO GAR that had a ~0.8 r^2 with oiSh%.
Do you think the Flyers should target Nik Hjalmarsson at the deadline or is he cooked?
That would be a nice fit, I assume you're thinking of it as a Niskanen replacement. I'm not ready to call him cooked after one partial season after a decade of consistency but I guess we'll have to see.
Do you think if Barzal played on the Lightning he'd have better results than Point? Seems like Barzal doesn't get the credit he deserves because he plays in an Islanders system that doesn't maximize his talents.
I hear that a lot from people but I'm very skeptical. Watching him play it seems pretty abundantly evident that Barzal controls what happens in the offensive zone a lot more than the Islanders' system does - you don't see Ross Johnston circling the zone with the puck. Barzal's main problems are that he doesn't have a very good shot and his teammates for whatever reason don't put away the chances he gives them. Dazzling talent but he's not in the top tier for me.
Kirby Dach and Alex Nylander: what is their realistic offensive potential? Top liner? Top 6?
Dach seems like a sky's the limit kind of guy. Nylander's underlying numbers aren't really impressive and prospect people seem a bit down on him so maybe "low top six"?
Break out candidate for the year?
It'd be nice to see Kase put up numbers with Pastrnak out but I'll say Bjorkstrand will explode. Jack Hughes could be another one, honestly. At the very least I expect him to far eclipse his numbers from last year.
Would a Rangers left sided defense of Johnson/lindgren/hajek be the worst ever iced?
Johnson and Hajek were two of the worst defencemen in the league last season. If Hajek doesn't take a leap they could be in the running.
Yea. Lets see if coach Dave galaxy brains this lineup. How do you assess the chances of regression across the Rangers lineup? We had some guys with career years (fox, panarin, zib). How worrying should that be?
Worrying. I have them out of the playoffs even assuming that Shesterkin can repeat and scale up his rookie results. Fox is one thing because his results were all underlying numbers-based, but Panarin and Zibanejad are in for regular old on-ice shooting regression.
Yea. I hope Igor is durable enough, because that starting role is difficult. Also Georgiev has had a very weird career so far. last year he was atrocious at the 5v5 but made it up at 5v4 for the most part. Yea, Fox's underlying numbers seem repeatable. But I assume there has to be some regression? Can you imagine the defense last year without him? Zib also had a weird year with elite shooting, but horrible defensive metrics. The other two wildcards are Kreider and Buch. The latter looks like he is likely gone unfortunately.
Would you recommend taking a statistics course before learning a language to get started in hockey analytics? or something else
If you're looking to get into the model-making side of things there are a lot of resources on how to use stuff like R to analyze public data
I'm not J but there are books/course which do both. Covering stats/R at the same time.
Hi I am a high school senior who is a fan of hockey and am starting to become more and more interested in the world of statistics (I am taking an AP Statistics class right now and really like it). I was wondering where do you think someone who has a basic understanding of statistics should start to dive in to hockey analytics? I also am new here and really like your articles (I am a fan of the New York Rangers and liked your Kaapo Kakko article).
It depends what you're looking to explore - to understand what the stats mean, places like the Athletic have really great explainers (like Charlie O'Connor's Analytics Primers). If you're more into the nuts and bolts hard stats stuff, the archives of websites like HockeyGraphs are probably the way to go.
This website is very helpful too: http://metahockey.com/publications/articles/
how frustrated should i be that kulak and mete are now projected to be either on the bottom pair or the taxi squad, and theyre replaced by edmundson and chiarot?
I would say very frustrated. The Habs seem to be doubling down what Chiarot managed to achieve in the first half of the season and neglecting how poor he was down the stretch and in the playoffs. Maybe Edmundson can do something with Petry but Kulak and Mete definitely have a better track record recently.
As a Habs fan, I am very frustrated by what they're doing to Mete. He could've been signed to a long term dirt cheap contract this year. (like 5x2M$) instead he's likely gone within a year and they overpaid for Edmundson. I hope Mete gets time on the PP at least.
Bruins fan here. Is Chara still a viable option? Or would boston be better suited picking up Ben Hutton?
Chara can play. Although he was brutal in the playoffs. The only issue I have with Chara as a signing for the Bruins would be the temptation to play him up in the lineup again, when he really should be sheltered heavily. But if they're willing to manage his minutes it'd be a nice add.
When do you see the Senators becoming legit contenders?
It really is up to them. Their moves this summer did not give me much faith that their player identification at the pro level is strong at all, but prospect people seem optimistic about their pool. In theory they should be revving up for stage 2 of the rebuild but that's not where their summer acquisitions will get them.