Melvin's 2021-22 NHL Team Projections

[MS]

Are you trying to say this years forward group is not that much better, and won't score more? If that's the case I strongly disagree.

Maybe you are making the classic "pessimistic fan" error.

Virtanen was playing in a second line role most of the time. Garland is our second line RW now, which is a huge upgrade for our second line. That leaves Hoglander, Podkolzin, and Dickinson as our possible 3rd line. That is a vastly better 3rd line. There shouldn't be any doubt that our forward group is much deeper than the 19/20 team.

The forward group is probably slightly better. There is more scoring potential but not by as much as you're saying - again, guys like Miller and Pearson will produce nowhere near what they did and you won't have Adam Gaudette scoring at a 47-point pace as essentially the #4C - and there were also very few injuries in 19-20.

Blueline is far worse. Goaltending is worse.

 

[MS]

I posted my comparison to the 19-20 team in the other thread. I think it's a reasonable comparison. Biggest thing is that team experienced unusual levels of good health, and then once the injuries finally happened the season was cut short. Also people expecting Demko to be as good as Markstrom was that year are IMO dreaming.

For the first 60 games of that season basically nobody important got hurt. Top 6 forwards/top 5 defenders/goaltending had basically a 10 game injury to Boeser and that's it. As you say, that is extremely unlikely to repeat.

 

[Melvin]

Here are some breakdowns since some people were asking.

Overall Offense

1	VGK
2	COL
3	CAR
4	TOR
5	EDM
6	PIT
7	FLA
8	NYR
9	S.J
10	N.J
11	VAN
12	T.B
13	OTT
14	CGY
15	SEA
16	L.A
17	PHI
18	WSH
19	NSH
20	NYI
21	BOS
22	DAL
23	STL
24	WPG
25	MIN
26	CHI
27	MTL
28	ARI
29	CBJ
30	ANA
31	BUF
32	DET

[TBODY] [/TBODY]

Overall Defense (Goaltending Not Considered)

	Team
1	MIN
2	COL
3	T.B
4	DAL
5	SEA
6	MTL
7	NYI
8	BOS
9	ARI
10	CBJ
11	TOR
12	NSH
13	VGK
14	CGY
15	PHI
16	OTT
17	FLA
18	WSH
19	PIT
20	N.J
21	STL
22	BUF
23	CAR
24	NYR
25	WPG
26	CHI
27	EDM
28	ANA
29	DET
30	L.A
31	VAN
32	S.J

[TBODY] [/TBODY]

Overall Defense (with Goaltending)

	Team
1	SEA
2	TOR
3	T.B
4	NYI
5	COL
6	DAL
7	VGK
8	NSH
9	BOS
10	MTL
11	WPG
12	MIN
13	STL
14	PIT
15	CGY
16	CHI
17	DET
18	NYR
19	N.J
20	CBJ
21	WSH
22	FLA
23	CAR
24	ANA
25	VAN
26	L.A
27	S.J
28	ARI
29	OTT
30	BUF
31	PHI
32	EDM

[TBODY] [/TBODY]

Goal Differential not including Goaltending

	Team
1	COL
2	VGK
3	TOR
4	CAR
5	T.B
6	SEA
7	MIN
8	FLA
9	PIT
10	DAL
11	EDM
12	CGY
13	NYI
14	OTT
15	N.J
16	MTL
17	BOS
18	PHI
19	NSH
20	NYR
21	WSH
22	ARI
23	CBJ
24	STL
25	S.J
26	VAN
27	L.A
28	BUF
29	WPG
30	CHI
31	ANA
32	DET

[TBODY] [/TBODY]

 

[im gangster]

For the first 60 games of that season basically nobody important got hurt. Top 6 forwards/top 5 defenders/goaltending had basically a 10 game injury to Boeser and that's it. As you say, that is extremely unlikely to repeat.

Never let it be said that covid only taketh away. Covid absolutely saved the 19/20 Vancouver Canucks’ season, and let them have that positive experience together in the gifted handout playoffs.

 

[MS]

Never let it be said that covid only taketh away. Covid absolutely saved the 19/20 Vancouver Canucks’ season, and let them have that positive experience together in the gifted handout playoffs.

The fact that COVID is constantly used as an excuse for Benning and his failures when in fact it's probably the best thing that ever happened to him continues to be baffling.

1) it saved their season when they were crashing out of the playoffs.
2) it created a totally artificial non-playoffs 'playoffs' situation in the bubble where the team managed to succeed.
3) the 2021 outbreak somehow is listed as a reason for the most recent season's struggles when it happened after the season was already essentially over and when the team actually had a better record after the outbreak.

 

[im gangster]

The fact that COVID is constantly used as an excuse for Benning and his failures when in fact it's probably the best thing that ever happened to him continues to be baffling.

1) it saved their season when they were crashing out of the playoffs.
2) it created a totally artificial non-playoffs 'playoffs' situation in the bubble where the team managed to succeed.
3) the 2021 outbreak somehow is listed as a reason for the most recent season's struggles when it happened after the season was already essentially over and when the team actually had a better record after the outbreak.

I can’t credit @Jyrki21 enough for encapsulating it so well in art.

 

[Paulinbc]

A troll could be a canucks fan still, or disguised as one. A troll is one who disrupts good faith debates.

Internet troll - Wikipedia

Debate? This is an internet chat room, I don’t think he signed up for any debate.

 

[AwesomeInTheory]

The fact that COVID is constantly used as an excuse for Benning and his failures when in fact it's probably the best thing that ever happened to him continues to be baffling.

1) it saved their season when they were crashing out of the playoffs.
2) it created a totally artificial non-playoffs 'playoffs' situation in the bubble where the team managed to succeed.
3) the 2021 outbreak somehow is listed as a reason for the most recent season's struggles when it happened after the season was already essentially over and when the team actually had a better record after the outbreak.

4) It's reinforced the "aNyThInG cAn HaPpEn" narrative thanks to the Canucks getting god tier goaltending from both Markstrom and Demko, which fans have bought in to as being acceptable/desirable.

 

[Paulinbc]

Here are some breakdowns since some people were asking.

Overall Offense

1	VGK
2	COL
3	CAR
4	TOR
5	EDM
6	PIT
7	FLA
8	NYR
9	S.J
10	N.J
11	VAN
12	T.B
13	OTT
14	CGY
15	SEA
16	L.A
17	PHI
18	WSH
19	NSH
20	NYI
21	BOS
22	DAL
23	STL
24	WPG
25	MIN
26	CHI
27	MTL
28	ARI
29	CBJ
30	ANA
31	BUF
32	DET

[TBODY] [/TBODY]

Overall Defense (Goaltending Not Considered)

	Team
1	MIN
2	COL
3	T.B
4	DAL
5	SEA
6	MTL
7	NYI
8	BOS
9	ARI
10	CBJ
11	TOR
12	NSH
13	VGK
14	CGY
15	PHI
16	OTT
17	FLA
18	WSH
19	PIT
20	N.J
21	STL
22	BUF
23	CAR
24	NYR
25	WPG
26	CHI
27	EDM
28	ANA
29	DET
30	L.A
31	VAN
32	S.J

[TBODY] [/TBODY]

Overall Defense (with Goaltending)

	Team
1	SEA
2	TOR
3	T.B
4	NYI
5	COL
6	DAL
7	VGK
8	NSH
9	BOS
10	MTL
11	WPG
12	MIN
13	STL
14	PIT
15	CGY
16	CHI
17	DET
18	NYR
19	N.J
20	CBJ
21	WSH
22	FLA
23	CAR
24	ANA
25	VAN
26	L.A
27	S.J
28	ARI
29	OTT
30	BUF
31	PHI
32	EDM

[TBODY] [/TBODY]

Goal Differential not including Goaltending

	Team
1	COL
2	VGK
3	TOR
4	CAR
5	T.B
6	SEA
7	MIN
8	FLA
9	PIT
10	DAL
11	EDM
12	CGY
13	NYI
14	OTT
15	N.J
16	MTL
17	BOS
18	PHI
19	NSH
20	NYR
21	WSH
22	ARI
23	CBJ
24	STL
25	S.J
26	VAN
27	L.A
28	BUF
29	WPG
30	CHI
31	ANA
32	DET

[TBODY] [/TBODY]

Wow. I must have had OEL rated much too high. 31? Yikes!!

 

[racerjoe]

Debate? This is an internet chat room, I don’t think he signed up for any debate.

Debate conversation whatever it semantics

 

[Intoewsables]

First of all, I love this and it's super neat. I'm pretty surprised to see Seattle ranked that high even though I think they'll be a fairly good team... but yeah, you covered that already.

How important is goaltending in your model? I'm a big believer in goalies being voodoo for the most part, so I'm curious how much things would change if goaltending was weighted less (if not already low), or removed from the equation entirely.

On a side note, can I ask how/where you're scraping this data (from)? I was trying to do something similar a while back until I realized I couldn't separate the events into ES/PP/PK. Are you using the NHL API?

 

[Peen]

Having a look back at the 20/21 projections, Melvin got 3/4 playoff teams in each division once the season simulation was run.

Incorrect projections: Dallas, Calgary, Arizona, NJ.

NHL Projections 2021

 

[jd22]

Melvin, thanks for putting in the work and explaining how you are looking at numbers, and why. It is great to see objective analysis and data presentation. In a world driven by metrics, your analyses hold far more water than opinions with minimal to no context.

 

[Dala]

Arizona have the ninth best defence overall? Any ideas how that came out of the data? (garbage in, garbage out?)

EDIT: ok, already covered by the bogus #1 PK obviously.

 

[Melvin]

Arizona have the ninth best defence overall? Any ideas how that came out of the data? (garbage in, garbage out?)

EDIT: ok, already covered by the bogus #1 PK obviously.

Jay Beagle is a good penalty killer. I hate the player, but, well, it’s true.

 

[Melvin]

First of all, I love this and it's super neat. I'm pretty surprised to see Seattle ranked that high even though I think they'll be a fairly good team... but yeah, you covered that already.

How important is goaltending in your model? I'm a big believer in goalies being voodoo for the most part, so I'm curious how much things would change if goaltending was weighted less (if not already low), or removed from the equation entirely.

On a side note, can I ask how/where you're scraping this data (from)? I was trying to do something similar a while back until I realized I couldn't separate the events into ES/PP/PK. Are you using the NHL API?

That the API doesn't include the game state is seriously the dumbest thing. It drives me f***ing insane. I have my own scrapers that I wrote in python using BeautifulSoup that pull from the NHL Shift Charts and Game Summaries but they are ugly AF and I want to retire them. There is another seemingly undocumented API here: https://api.nhle.com/stats/rest/en/shiftcharts?cayenneExp=gameId=2020020212. This one will get you shift charts but I've only started playing with it and couldn't tell you much about it.

For this project I was able to get the components I needed from evolving-hockey by giving them some money and downloading the data as CSV's; that was easier than trying to get all my scrapers up and running again after my move.

 

[Intoewsables]

That the API doesn't include the game state is seriously the dumbest thing. It drives me f***ing insane. I have my own scrapers that I wrote in python using BeautifulSoup that pull from the NHL Shift Charts and Game Summaries but they are ugly AF and I want to retire them. There is another seemingly undocumented API here: https://api.nhle.com/stats/rest/en/shiftcharts?cayenneExp=gameId=2020020212. This one will get you shift charts but I've only started playing with it and couldn't tell you much about it.

For this project I was able to get the components I needed from evolving-hockey by giving them some money and downloading the data as CSV's; that was easier than trying to get all my scrapers up and running again after my move.

Cool, I'll look into evolving-hockey and see if their data fits what I'm looking for. I've tried playing around with the shift chart data you mentioned but I remember it being pretty limited as well. Thanks!

 

[Melvin]

Alright, so I figured out what was going on with Arizona's PK and fixed it. They are now 11th rather than 1st. It's actually pretty funny. Basically, due to their current roster my code wasn't able to figure out which defenders were going to kill penalties for them, as only Stralman and Chychrun had done it before. Is Gostisbehere going to kill penalties? What are they doing? And due to lack of foresight it was essentially filling that void with 0's that caused their PKing to look amazing. so I am now filling in voids with league average (which is probably still too generous but better than 0) which takes their PK down to 11th place. This also affected Columbus who had a similar thing and were rated #2 and are now #8.

There were a few roster updates as well, with Marcus Johansson going to Seattle and Brendan Perlini going to Edmonton.

I won't post new rankings yet but will post a weekly update or something as long as the thread is reasonably active. If the thread dies I'll just post one final ranking before the season starts and not keep bumping the thread.

 

[Dala]

Perhaps more fair to fill in 10th percentile worst values instead of the average?

 

[BWJM]

Seattle... 4th ruined it for me.

 

[Melvin]

Perhaps more fair to fill in 10th percentile worst values instead of the average?

yeah that’s true. But also I assume they will add more defenders at some point? So it may become moot.

 

[bertuzzi2bure]

lmao this guys projections are dog $hit

seattle top 5? what?

 

[Melvin]

If anyone wants to give me some constructive suggestions on how to “fix” the Seattle “problem” since according to the prescient experts on this forum it’s impossible for them to be good, I am completely open to suggestions. I doubt they finish that high as well but I am not going to just manually move them down because of feelings.

 

[Dala]

Instead of direct (?) scaling perhaps assume values gets worse with increased toi/role? Not sure I understand your model though.

Decrease could be something like scale by (avg current toi) / (average of destination toi)? Or the opposite given which one is larger etc.

 

[Melvin]

Instead of direct (?) scaling perhaps assume values gets worse with increased toi/role? Not sure I understand your model though.

Decrease could be something like scale by (avg current toi) / (average of destination toi)?

Ah, the problem here is that the average of the destination is the very thing I'm trying to project, so it's complicated.

Like even if I could say with some certainty that Schwartz (to pick a random example) will be 5% worse than SEA's average...what is SEA's average? That is what I'm trying to calculate!