Melvin's 2021-22 NHL Team Projections

[RobertKron]

I have a pretty good idea of how good the other teams are, but I cannot give a precise ranking for each team without studying each teams current roster, and I don't want to do that. Most people don't want to do that.

You can't give a precise ranking for the Canucks without a precise ranking for the other teams, since the Canucks' position depends on the other teams.

 

[Bojack Horvatman]

Great work Melvin. I have a couple questions though.

1) Last year for an 82 game season, the differentials for top teams were lower than they should be. This year they are more normal. Did you do anything different this year that would effect that?

2) Last year league scoring was much lower in your predictions. A handful of teams had more than 200 gf, and the leader had 220. When 2019 24 teams had more than 220 goals. What do you think the reason for this difference was?

 

[logan5]

You can't give a precise ranking for the Canucks without a precise ranking for the other teams, since the Canucks' position depends on the other teams.

Oh shit. You got me.

I mean it's just a f***in' hockey forum. We aren't designing a space station here.

 

[Prometheus]

Joining the chorus to say thank you @Melvin for all this very hard and informative work.

 

[Paulinbc]

Oh shit. You got me.

I mean it's just a f***in' hockey forum. We aren't designing a space station here.

I’m actually designing a space station.

 

[RobertKron]

Oh shit. You got me.

I mean it's just a f***in' hockey forum. We aren't designing a space station here.

No shit. But if you know you can't do a thing, maybe don't act like you're doing the thing.

 

[AwesomeInTheory]

I think Elias Pettersson will get 200 points, the Calgary flames will get 84 regular season wins to finish last in the division, first in the league but second in the conference, Conor McDavid will regress to a 20 point forward and Auston Matthews will win the Norris trophy.

Please treat my opinion as being as valid as everyone else's, thanks and namaste.

E: This was simulated by playing EA NHL on Rookie mode, where Pettersson scored 300 points and then my save file got corrupted, resulting in the Calgary chicanery. Accounting for digital/real world differences, I adjusted the weighting slightly, which is how I arrived at the conclusion Petey is going to hit 200 points, which I think is reasonable.

We also cannot rule out similar chicanery happening in the real world with Calgary. While it is true demonic possession has never occurred in the NHL, it is possible that it might happen. It's also equally plausible that Treliving may have knocked over an ancient idol of a long lost Sumerian god, cursing the Flames. Just because it hasn't happened doesn't mean that it can't happen.

 

[HockeyNightInAsia]

I don't know if I have a concrete explanation. Crosby/Malkin have not been caved in defensively the way Ovechkin has despite being the same age, they are better at drawing penalties, and Jarry/DeSmith project pretty well despite Jarry having a poor year last year.

Thanks Melvin. One caveat is that WSH has come out quite poor analytically past couple of seasons already, and I believe they have found a way to "over perform the models", probably in a similar manner as the Isles (man they are always the anti-analytics LOL). So, just personally, I would make my own overlay (for my fantasy hockey purposes anyway LOL). Jarry is also interesting case.....

 

[Melvin]

Vegas losing Tuch? IMO he was one of the drivers of that team. Lots of pressure on Lehner this year too.
Interesting.

He had surgery and is expected to miss six months. I originally included him but now I think it makes more sense to assume they don't have him, since by the time he comes back all the rosters will be different anyway.

 

[Melvin]

Thanks Melvin. One caveat is that WSH has come out quite poor analytically past couple of seasons already, and I believe they have found a way to "over perform the models", probably in a similar manner as the Isles (man they are always the anti-analytics LOL). So, just personally, I would make my own overlay (for my fantasy hockey purposes anyway LOL). Jarry is also interesting case.....

So basically the opposite of the stupid Coyotes who always do worse than the models.

 

[Melvin]

You are making the classic 'optimistic fan' error in these sorts of projects where you're assuming everything that went great in prior years holds steady and everything that went wrong will be fixed by new moves. And then projecting exponential improvement on all young players, which doesn't happen.

As an example the 19-20 team had JT Miller and Tanner Pearson having massive outlier career seasons and combining for 140 points/82 GP. If they get to 100 this year, you'd take it. Alex Edler scored at an ES pace that year that none of our defenders are likely to match.

Likewise, Podkolzin and Hoglander are unlikely to contribute more than Virtanen and Leivo/Toffoli contributed that year, but you're counting them as 'major additions'.

People projected 'breakout to superstar!' in 19-20 for Pettersson. Didn't happen. Projected it again in 20-21. Didn't happen.

Our defense is very, very bad. The only way you can rationalize it as good is by point-counting with the assumption that 4 different guys all somehow play on the #1 PP unit.

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.

 

[Melvin]

Like, I'm not expecting everyone to put in the kind of effort I did, but if you're going to make a statement like "The Canucks have one of the best forward groups in the NHL," (whatever that means,) I am going to assume you did the research to support such a statement and have a good grasp of which teams have good forward groups and which ones don't. It would be silly to say something like that otherwise, wouldn't it?

I am the wealthiest person in the world! (I have no knowledge of anyone's wealth except mine.)

 

[logan5]

No shit. But if you know you can't do a thing, maybe don't act like you're doing the thing.

Lol. Quit being so dramatic.

While appreciate the work that the op put in, the fact is, even if you had a 100 man team of hockey experts, their predictions would not be any more accurate than somebody here on HF Boards using a simple method.

The guy asked me to rank the Canucks, so I did using a simpler method. My predicted ranking might end up being closer to reality.

It's just a hockey forum that is suppose to be for fun. Who cares how I came up with my conclusions or acting like I'm "doing the thing".

 

[AwesomeInTheory]

I would sincerely like it if these people would tel me based on their own critical analysis where they have the Canucks ranked. Is it 20? Is it 10? Top 5? At least I’m willing to put it out there.

I'm mystified as to why logan is getting grief over his usual effort level when people specifically asked for responses that aren't complete asspulls.

 

[RobertKron]

Lol. Quit being so dramatic.

While appreciate the work that the op put in, the fact is, even if you had a 100 man team of hockey experts, their predictions would not be any more accurate than somebody here on HF Boards using a simple method.

The guy asked me to rank the Canucks, so I did using a simpler method. My predicted ranking might end up being closer to reality.

It's just a hockey forum that is suppose to be for fun. Who cares how I came up with my conclusions or acting like I'm "doing the thing".

lol

 

[VanillaCoke]

Dont let the same troll derail yet another thread. Let him have his genius opinions.

Also clearly a few posters didnt even read the OP, which is pathetic and embarrassing given their responses. Atleast try to understand that which you're commenting on.

 

[logan5]

Like, I'm not expecting everyone to put in the kind of effort I did, but if you're going to make a statement like "The Canucks have one of the best forward groups in the NHL," (whatever that means,) I am going to assume you did the research to support such a statement and have a good grasp of which teams have good forward groups and which ones don't. It would be silly to say something like that otherwise, wouldn't it?

I am the wealthiest person in the world! (I have no knowledge of anyone's wealth except mine.)

I do have knowledge of the other teams. Based on that knowledge, imo the Canucks have a top 5 forward group. I know somebody is going to list off some teams that they think have a better group, but the Canucks are a group that is still maturing, so players like Pettersson will be better, in addition to Garland.

In 19/20 the Canucks were 8th in the league in goal scoring, so it's not a huge stretch to say top 5.

 

[Sinistril]

"All models are wrong, some are useful"

That's true, but most aren't. Funny enough, I did an extremely lazy model with only one assumption before the season last year for the North division. I looked at how my myopic model finished versus yours (from your last update before the season started last year).

Teams	MyExpected	Melvin Expected	Actual	MyDiff	Melvin Diff
Calgary	-4.666666666666	16	-5	0.333333333333996	21
Edmonton	-28.5	-1	29	57.5	30
Montreal	41.333333333333	14	-7	48.333333333333	21
Ottawa	-28.5	-18	-34	5.5	16
Toronto	34.666666666667	3	38	3.333333333333	35
Vancouver	-64.333333333334	-19	-39	25.333333333334	20
Winnipeg	50	6	18	32	12

[TBODY]

[/TBODY]

As you can see, mine predicted 3 of the teams nearly spot on, had much wider errors than you for 3 of the others, and a slightly wider error for Vancouver than you. Whereas yours were pretty consistent in the error magnitudes (relative to mine)... but the errors happened in both directions. Which model was more useful? I think both were junk at the end of the day. We did both project that Vancouver was going to be garbage last year though... so there's that.

 

[logan5]

Let him have his genius opinions..

That's not too much to ask is it?

 

[im gangster]

Ahhhh forget it

 

[Melvin]

That's true, but most aren't. Funny enough, I did an extremely lazy model with only one assumption before the season last year for the North division. I looked at how my myopic model finished versus yours (from your last update before the season started last year).

Teams	MyExpected	Melvin Expected	Actual	MyDiff	Melvin Diff
Calgary	-4.666666666666	16	-5	0.333333333333996	21
Edmonton	-28.5	-1	29	57.5	30
Montreal	41.333333333333	14	-7	48.333333333333	21
Ottawa	-28.5	-18	-34	5.5	16
Toronto	34.666666666667	3	38	3.333333333333	35
Vancouver	-64.333333333334	-19	-39	25.333333333334	20
Winnipeg	50	6	18	32	12

[TBODY] [/TBODY]

As you can see, mine predicted 3 of the teams nearly spot on, had much wider errors than you for 3 of the others, and a slightly wider error for Vancouver than you. Whereas yours were pretty consistent in the error magnitudes (relative to mine)... but the errors happened in both directions. Which model was more useful? I think both were junk at the end of the day. We did both project that Vancouver was going to be garbage last year though... so there's that.

Yours looks like it’s just missing some regression. Those are some wild numbers. You can see that mine projects +38 at the absolute top end even though I can almost guarantee some teams will finish above that. Models should not really be able to project outliers like this, you know?

agree my model from last year was pretty junk. I addressed this in the OP. There were several bugs caused by my fingers being too fat, primarily. That’s the point of the thread though, I’m curious to see where it misses and what kind of assumptions I made that were bad.

 

[Paulinbc]

Dont let the same troll derail yet another thread. Let him have his genius opinions.

Also clearly a few posters didnt even read the OP, which is pathetic and embarrassing given their responses. Atleast try to understand that which you're commenting on.

This is a Canucks fan site. A troll would be someone else coming in from the Edmonton forum or the Seattle forum and saying the Canucks suck. This guy is coming in here in support of the Canucks.

He may have different opinions, but he is not a troll.

 

[Paulinbc]

That's true, but most aren't. Funny enough, I did an extremely lazy model with only one assumption before the season last year for the North division. I looked at how my myopic model finished versus yours (from your last update before the season started last year).

Teams	MyExpected	Melvin Expected	Actual	MyDiff	Melvin Diff
Calgary	-4.666666666666	16	-5	0.333333333333996	21
Edmonton	-28.5	-1	29	57.5	30
Montreal	41.333333333333	14	-7	48.333333333333	21
Ottawa	-28.5	-18	-34	5.5	16
Toronto	34.666666666667	3	38	3.333333333333	35
Vancouver	-64.333333333334	-19	-39	25.333333333334	20
Winnipeg	50	6	18	32	12

[TBODY] [/TBODY]

As you can see, mine predicted 3 of the teams nearly spot on, had much wider errors than you for 3 of the others, and a slightly wider error for Vancouver than you. Whereas yours were pretty consistent in the error magnitudes (relative to mine)... but the errors happened in both directions. Which model was more useful? I think both were junk at the end of the day. We did both project that Vancouver was going to be garbage last year though... so there's that.

0.33 does not round up to 0.34, or 0.3996. I'm not sure I can trust the rest of this table.

 

[Sinistril]

0.33 does not round up to 0.34, or 0.3996. I'm not sure I can trust the rest of this table.

Blame the Linux excel ripoff for any calculation errors like that hah. Literally just put =abs(I14-K14)

 

[racerjoe]

This is a Canucks fan site. A troll would be someone else coming in from the Edmonton forum or the Seattle forum and saying the Canucks suck. This guy is coming in here in support of the Canucks.

He may have different opinions, but he is not a troll.

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

Internet troll - Wikipedia