2016-2017 NHL Thread

[princedpw]

https://www.fearthefin.com/2017/8/2/16082098/nhls-best-players-under-age-25-how-did-the-fear-the-fin-writing-staff-vote

The ballot from the Sharks SB Nation afiliate:

Connor McDavid (C, Edmonton Oilers)
Auston Matthews (C, Toronto Maple Leafs)
Nikita Kucherov (RW, Tampa Bay Lightning)
Matt Murray (G, Pittsburgh Penguins)
Mark Scheifele (C, Winnipeg Jets)
Patrik Laine (RW, Winnipeg Jets)
David Pastrnak (RW, Boston Bruins)
Johnny Gaudreau (LW, Calgary Flames)
Jack Eichel (C, Buffalo Sabres)
Leon Draisaitl (C, Edmonton Oilers)
Aleksander Barkov (C, Florida Panthers)
Mitch Marner (RW, Toronto Maple Leafs)
William Nylander (RW, Toronto Maple Leafs)
Filip Forsberg (LW, Nashville Predators)
Zach Werenski (D, Columbus Blue Jackets)
Jacob Trouba (D, Winnipeg Jets)
Hampus Lindholm (D, Anaheim Ducks)
Jonathan Drouin (LW, Montreal Canadiens)
Nikolaj Ehlers (LW, Winnipeg Jets)
Jonathan Huberdeau (LW, Florida Panthers)
Alex Galchenyuk (C, Montreal Canadiens)
Seth Jones (D, Columbus Blue Jackets)
Aaron Ekblad (D, Florida Panthers)
Brandon Saad (LW, Chicago Blackhawks)
Sean Monahan (C, Calgary Flames)

They have Nylander and Marner back to back in the Top 15.  Twitter-sphere is going to have a fit (most of who I follow on the analytics side didn't think those two should be in the Top 25) if they end up that high in the overall list.

In general, I feel these lists seem biased against defensemen with Dougie Hamilton being left off and this list having no defenseman in the top 14.  GMs seem to value defensemen much more highly than wingers (see the Hall for Larsson trade, though, of course, that may be an outlier).

I mean amongst NHL defensemen, Hamilton finished 9th in points, 5th in shots and I think I recall him being top-5 in the NHL in corsi-related stats, which suggests strong defensive abilities (I don't know how else to measure it/not going to bother).  Werenski is a 20-year-old rookie who was 13th in defensemen in points (don't his corsi, but +17 ... yeah, I know +/- ... again, I'm too lazy for real stats).

On the other hand, Marner and Nylander are wingers (at the moment), tied for 32nd amongst forwards.  Nylander is 20 too.  I love those guys but Werenski's rookie season was far more surprising to me.  Forsberg is a 22-year-old who tied for 43rd amongst forwards in points, albeit with 31 goals.

I dunno, I'd push more defensemen further up the list.  Top defensemen like Pronger, Chara, Lidstrom, Neidermeyer have played almost half the game in the playoffs and have had just as much to say about their teams playoffs success over the years as top forwards IMHO.

 

[princedpw]

oh, and I forgot about Drouin -- the guy is 22 and scored 53 points (tied for 68th amongst forwards, albeit in only 73 games).  How does that have more impact than what Douglie Hamilton did (50 points in 81 games + the corsi monster thing)?  I don't really get it.

 

[herman]

I don't really get it.

Every ranking including all three positions will always favour forwards because scoring is the easiest thing to measure.

 

[herman]

17. Mitch Marner
https://www.sbnation.com/nhl/2017/8/3/16075194/nhl-top-25-under-25-rankings-2017-mitch-marner-toronto-maple-leafs

Woohoo!

Highest rank: No. 3
Lowest rank: Not ranked

Lol, wth

 

[Coco-puffs]

I hope Montreal got David Johnson.

Why do you say that?  Seems he riled up a bunch of other analytics people on Twitter quite often, but I thought his counter-points to most of what people were saying were refreshing even if I didn't always agree with him.

If you are thinking him going to Montreal would hurt the team because his opinion was counter to many others on Twitter, I don't think that would be true.

They dumped their analytics guy for the Weber tirade, so I thought having someone who is an analyst that views the game similar to the club would be right in line with their apparent game plan.

His counterpoints were almost always a) goal-metrics based; or b) didn't see the need to apply statistical rigour. It's old school hockey in mathematical clothing.

Now that he's in Calgary, he can hate on Klefbom properly.

Well, goal-based metrics still have their use.  Shot attempt based metrics have proven statistically to have the provide the best PREDICTIVE use.  ie, Someone like Rielly who was around 50% CF last year SHOULD see his goal-based metrics improve.  However, using goal based metrics to ANALYZE the past is not without merit.  Goals-based metrics are not as predictive as shot based metrics because of luck, lower number of events (sampling), etc, but its still very useful to ANALYZE past performance to some degree.  Rielly had a terrible GF% and analyzing WHY is important.  Just saying he'll be fine because he had a 50% CF doesn't solve the problem of why he has historically underperformed in GF% compared to his CF%.  There may be issues that need fixing there.  Similarily, Shea Weber has almost ALWAYS had a GF% above his CF%.  Again, CF% is generally more predictive for future performance but at some point you can't just ignore the RESULTS (ie GF%).

As for the statistical rigour... all of his stuff has been scrubbed so I can't pull examples but he's a data analyst so I don't see how he wouldn't believe in statistical rigour.  His main complaint was the statistics work people were doing was it had more to do with BAD data.  You can develop some of the best models in the world, but things fall apart with bad data and that was the crux of his argument.  I think he went about it the wrong way often and dismissing someone's work on the premise of bad data without looking through it in more detail doesn't put him in the best light- but he had a point.

 

[CarltonTheBear]

17. Mitch Marner
https://www.sbnation.com/nhl/2017/8/3/16075194/nhl-top-25-under-25-rankings-2017-mitch-marner-toronto-maple-leafs

Woohoo!

Highest rank: No. 3
Lowest rank: Not ranked

Lol, wth

That kinda shows the flaw to these rankings which is that they let the voter basically decide the criteria. Person A probably went heavy on future potential while Person B focused more on current impact. Still, with enough votes it all balances out nicely I think.

Although 3 is still pretty darn high of course. He'd be #3 on my Leafs-only ranking.

 

[Kin]

That kinda shows the flaw to these rankings which is that they let the voter basically decide the criteria. Person A probably went heavy on future potential while Person B focused more on current impact. Still, with enough votes it all balances out nicely I think.

Although 3 is still pretty darn high of course. He'd be #3 on my Leafs-only ranking.

Honestly though if there's someone out there arguing Marner over one of McDavid, Matthews or Eichel then it seems like the real problem is the people you've chosen to help make the list regardless of criteria.

 

[herman]

Well, goal-based metrics still have their use.  Shot attempt based metrics have proven statistically to have the provide the best PREDICTIVE use.  ie, Someone like Rielly who was around 50% CF last year SHOULD see his goal-based metrics improve.  However, using goal based metrics to ANALYZE the past is not without merit.  Goals-based metrics are not as predictive as shot based metrics because of luck, lower number of events (sampling), etc, but its still very useful to ANALYZE past performance to some degree.  Rielly had a terrible GF% and analyzing WHY is important.  Just saying he'll be fine because he had a 50% CF doesn't solve the problem of why he has historically underperformed in GF% compared to his CF%.  There may be issues that need fixing there.  Similarily, Shea Weber has almost ALWAYS had a GF% above his CF%.  Again, CF% is generally more predictive for future performance but at some point you can't just ignore the RESULTS (ie GF%).

As for the statistical rigour... all of his stuff has been scrubbed so I can't pull examples but he's a data analyst so I don't see how he wouldn't believe in statistical rigour.  His main complaint was the statistics work people were doing was it had more to do with BAD data.  You can develop some of the best models in the world, but things fall apart with bad data and that was the crux of his argument.  I think he went about it the wrong way often and dismissing someone's work on the premise of bad data without looking through it in more detail doesn't put him in the best light- but he had a point.

I like the half step between shot metrics and goal metrics for future casting (xG) and adding context to what made the goals come out the way they did. There are a lot of team structure, linemate, competition, deployment, and luck factors embedded in goal-metrics that obfuscate individual measures.

Honestly, I didn't follow Johnson's work enough to say for certain one way or another whether he had valid points, other than it looked like he was of the same hockey philosophy as David Staples. Maybe I'm throwing the baby out with the bathwater, but I never felt like I learned anything from his analyses.

Statistical rigour is more about Johnson's criticism of statistical models, where his only reason appeared to be goal metrics not lining up with the models'. He never put in the work to truly dissect a model's shortcomings and it made for a frustrating conversation for the people who put in their due diligence.

All that being said, he ran a very useful set of sites and really put a lot of the public metrics into one accessible space for everyone to use, so I'm happy he was able to parlay that sacrifice into a more profitable venture.

 

[Coco-puffs]

Honestly, I didn't follow Johnson's work enough to say for certain one way or another whether he had valid points, other than it looked like he was of the same hockey philosophy as David Staples. Maybe I'm throwing the baby out with the bathwater, but I never felt like I learned anything from his analyses.

I don't think his philosophy was anything like David Staples.  He wasn't "old-school" in his thinking, he just didn't agree with everything the analytics community was putting out.  Its not that black and white.

 

[herman]

I don't think his philosophy was anything like David Staples.  He wasn't "old-school" in his thinking, he just didn't agree with everything the analytics community was putting out.  Its not that black and white.

I'm not sure what else to call being in love with Polak's and Russell's games. I'm open to suggestions/elaborations.

 

[CarltonTheBear]

I don't think his philosophy was anything like David Staples.

I don't think that's actually an unfair comparison. I don't know how familiar you are with Staples' writing, but he does incorporate analytics a fair bit into his work. They just generally go against the type of work 'hockey twitter' supports, like a preference of goals over twitter or the use of proprietary data. There's definitely some overlap between him and Johnson.

 

[herman]

I don't think his philosophy was anything like David Staples.

I don't think that's actually an unfair comparison. I don't know how familiar you are with Staples' writing, but he does incorporate analytics a fair bit into his work. They just generally go against the type of work 'hockey twitter' supports, like a preference of goals over twitter or the use of proprietary data. There's definitely some overlap between him and Johnson.

Staples is very much into analytics like Neilson's Scoring Chances, which are similar to xG, but subjectively attributed to an affected subset of players on the ice. Nouveaux Old School?

It's the kind of stat that gets Kris Russell 4M/4.

Edit: updated to a more thorough link on Scoring Chances.

 

[Coco-puffs]

I don't think his philosophy was anything like David Staples.  He wasn't "old-school" in his thinking, he just didn't agree with everything the analytics community was putting out.  Its not that black and white.

I'm not sure what else to call being in love with Polak's and Russell's games. I'm open to suggestions/elaborations.

Again, I'm going from memory here since his articles have been scrubbed:

He wasn't in love with Polak.  Just said that everyone crying for an upgrade on defense last season and pointing the finger at Polak should look at more than just his Corsi and his penchant for throwing the puck off the glass and out.  David said Rielly's results were the issue, not Polak's.  He didn't say we should put Rielly on waivers or anything, but defensively speaking Rielly was the one getting poor results.  His underlying stats were decent, and that is nice in that it says things should improve for Rielly, but ultimately, Rielly's results were poor.

Lets compare their seasons:

- CF%:  Rielly 50.7% Polak 47.7%    Yes, Polak doesn't look good here.  No question.  But lets delve a little further.

- Quality of Competition (Percentage of time spent against: Elite/Middle/Grinder):  Rielly 42/39/19  Polak 24/43/33
I point this out because without a doubt Rielly faced stiffer competition and that should be considered when we analyze things.

- DFF% (Dangerous Fenwick, https://oilersnerdalert.wordpress.com/2015/10/30/explaining-dangerous-fenwick/)
Rielly:  50.4%  Polak:  50.4%
Interesting, when you account for shot quality Polak comes out looking decent.  WHO KNEW!

- GF/GA:  Rielly:  49/67  Polak:  40/33
Ah, those results.  Rielly had a shitty year.  Polak did quite well.  I understand the argument that this isn't predictive or repeatable- we shouldn't expect Polak to get better results than Rielly based on GF% but in terms of RESULTS, Polak was fine.  And that DFF% indicates Polak wasn't hurting us in terms of allowing more dangerous opportunities against that what was being generated while he was on the ice. 

All stats from http://puckiq.com/

(interestingly, if you look at the stats and filter by competition you'll find that Rielly had decent GA/60 against elite competetion (2.00 GA/60) and then fell apart with Middle/Grinders (3.9 and 3.6 respectively).  Weird.  Underlying numbers were average against Elite/Middle (around 50% CF%) and exceptional against Grinders (54.5%).  So was it bad luck?  I'm sure quite a bit of it was and we should look forward to an improvement in his results.  Then again, when I watch games I do see correctable things in Rielly's game defensively:  Not picking up the forwards stick or helping the goalie to see when point shots are coming in.  I remember alot of goals coming with Rielly on the ice where the player in front tipped the point shot or the goalie had no chance to see it because he wasn't helping clear the net/improve the goalies vision of the puck)

 

[Coco-puffs]

I don't think his philosophy was anything like David Staples.

I don't think that's actually an unfair comparison. I don't know how familiar you are with Staples' writing, but he does incorporate analytics a fair bit into his work. They just generally go against the type of work 'hockey twitter' supports, like a preference of goals over twitter or the use of proprietary data. There's definitely some overlap between him and Johnson.

No, I'm not that familiar with his writings tbh  My interpretation of it based on people slamming him on twitter often is that he was more of the old-school type. 

 

[herman]

He wasn't in love with Polak.  Just said that everyone crying for an upgrade on defense last season and pointing the finger at Polak should look at more than just his Corsi and his penchant for throwing the puck off the glass and out.  David said Rielly's results were the issue, not Polak's.  He didn't say we should put Rielly on waivers or anything, but defensively speaking Rielly was the one getting poor results.  His underlying stats were decent, and that is nice in that it says things should improve for Rielly, but ultimately, Rielly's results were poor.

Lets compare their seasons:

- CF%:  Rielly 50.7% Polak 47.7%    Yes, Polak doesn't look good here.  No question.  But lets delve a little further.

- Quality of Competition (Percentage of time spent against: Elite/Middle/Grinder):  Rielly 42/39/19  Polak 24/43/33
I point this out because without a doubt Rielly faced stiffer competition and that should be considered when we analyze things.

- DFF% (Dangerous Fenwick, https://oilersnerdalert.wordpress.com/2015/10/30/explaining-dangerous-fenwick/)
Rielly:  50.4%  Polak:  50.4%
Interesting, when you account for shot quality Polak comes out looking decent.  WHO KNEW!

- GF/GA:  Rielly:  49/67  Polak:  40/33
Ah, those results.  Rielly had a shitty year.  Polak did quite well.  I understand the argument that this isn't predictive or repeatable- we shouldn't expect Polak to get better results than Rielly based on GF% but in terms of RESULTS, Polak was fine.  And that DFF% indicates Polak wasn't hurting us in terms of allowing more dangerous opportunities against that what was being generated while he was on the ice. 

All stats from http://puckiq.com/

(interestingly, if you look at the stats and filter by competition you'll find that Rielly had decent GA/60 against elite competetion (2.00 GA/60) and then fell apart with Middle/Grinders (3.9 and 3.6 respectively).  Weird.  Underlying numbers were average against Elite/Middle (around 50% CF%) and exceptional against Grinders (54.5%).  So was it bad luck?  I'm sure quite a bit of it was and we should look forward to an improvement in his results.  Then again, when I watch games I do see correctable things in Rielly's game defensively:  Not picking up the forwards stick or helping the goalie to see when point shots are coming in.  I remember alot of goals coming with Rielly on the ice where the player in front tipped the point shot or the goalie had no chance to see it because he wasn't helping clear the net/improve the goalies vision of the puck)

Thanks for putting this together.

- Quality of Competition (Percentage of time spent against: Elite/Middle/Grinder):  Rielly 42/39/19  Polak 24/43/33

(interestingly, if you look at the stats and filter by competition you'll find that Rielly had decent GA/60 against elite competetion (2.00 GA/60) and then fell apart with Middle/Grinders (3.9 and 3.6 respectively).  Weird.  Underlying numbers were average against Elite/Middle (around 50% CF%) and exceptional against Grinders (54.5%).

For this weird disconnect, I'd point to deployment with forward lines. When Rielly was up against Elites (with Zaitsev, adjusting from the KHL), he was with Komarov - Kadri - Brown for the most part.

All of Polak's good DFF% was against the grinders. Any other level of competition and he was sub 50%.

 

[herman]

16. William Nylander
https://www.sbnation.com/nhl/2017/8/4/16083344/nhl-best-young-players-top-25-under-25-william-nylander-toronto-maple-leafs

Highest rank: No. 4
Lowest rank: Not ranked

Lol again

Averaged rank was higher than Marner's but his peak rank was behind Marner's. It looks like someone's ranking was:
McDavid
Matthews
Marner
Nylander
Whatever

 

[Bender]

Wonder what Matthews lowest rank was.

Sent from my SM-G935W8 using Tapatalk

 

[herman]

Wonder what Matthews lowest rank was.

We'll find out in 3 weeks! (unless it leaks)

 

[Bill_Berg]

Wonder what Matthews lowest rank was.

Sent from my SM-G935W8 using Tapatalk

I wonder what McDavid's was. Anyone rank him not 1st? Would like to know who they did rank 1st if so.


Matthews.

But probably everyone ranked him 1st.

 

[Frank E]

Was just reading about how NHL players aren't really making much noise about no Olympic hockey...so I was thinking, why?

1.  The players don't want to show their cards for the next CBA negotiation...don't want to add to the league's hand, so play it cool.
2.  The players liked the World Cup tourney:  they got a split of the revenue, and the schedule is better for them too.
3.  ?