Do you think Ovechkin's legacy will improve over time?

[Beau Knows]

For me Ovechkin is already the best goal scorer of all-time. Him breaking the record vs missing it by 1 goal or something wouldn't change anything, other than it just being neat.

It's not like the period of time he played is going to be lost to time and all that will remain is the number of goals he scored. There will be mountains of videos of him playing and articles that illustrate his standing in the hockey world during his era.

Likewise, if this higher scoring era we've been in for the last few years continues or even opens up further, and someone comes along and surpassed whatever goal totals Ovechkin ends up with, while not being as dominant against his peers as Ovechkin was. I think people will understand that and take into account that Ovechkin did it in a tougher scoring environment and that he would still be the best goal scorer ever.

 

[The Panther]

Peak and prime seasons are definitely the most important when it comes to ranking players in general and ranking goal scorers, but I think some knowledgeable fans can be become too dismissive of career milestones (just as casual fans and media types tend to make too big a deal of such things).

Does adding 25-35 goal seasons at the end of his career make Ovechkin a better or greater goal scorer than guys like Gretzky, Lemieux, Hull, Richard etc? No, because such seasons pale in comparison to what those players did in their best seasons and over their primes as far as goal scoring is concerned. But that doesn't mean the value of 25-35 goal seasons late in a player's career is zero, either (even if we are talking about a guy who's led the League in goals nine times). Hockey is a grind. Players age and lose their abilities. It can be a challenge and a testament to a player's goal scoring smarts and devotion to achieving a goal to bang in those last 70 or so goals and get the record.

I don't disagree with anything you wrote, and I was not dismissing productive longevity. I am saying that hitting a certain , specific number of (whatever) doesn't overnight lead to a higher ranking.

 

[DitchMarner]

I don't disagree with anything you wrote, and I was not dismissing productive longevity. I am saying that hitting a certain , specific number of (whatever) doesn't overnight lead to a higher ranking.

I agree with that.

I think there's some value in breaking the goal record (some may think there really isn't), but it's not the sort of thing that should decide if he's the greatest goal scorer ever, nor should a few dozen additional goals late in his career move him up the all-time list substantially.

 

[Crosby2010]

If secondary assists were just as good as goals you might have a point.

So a point is a point. I never get the idea that a secondary assist isn't important or crucial to the play. We are talking about Crosby here. This is a guy who for his whole career has been the one that has driven the play and been the guy who has carried every line he's been on. Or are we thinking here that Chris Kunitz really was a world class sniper on his own? When you watch both play especially today, you still realize that Crosby is the one who runs the show and makes the plays. Ovechkin doesn't do this anymore and it shows in their stats. This doesn't mean Ovechkin hasn't had a great career, as he is 39 years old now, but it just means Crosby is easily the better player now and he has been the better player since 2010 while they both were more or less 1a) 1b) from 2005-'10

 

[Midnight Judges]

So a point is a point. I never get the idea that a secondary assist isn't important or crucial to the play. We are talking about Crosby here. This is a guy who for his whole career has been the one that has driven the play and been the guy who has carried every line he's been on. Or are we thinking here that Chris Kunitz really was a world class sniper on his own? When you watch both play especially today, you still realize that Crosby is the one who runs the show and makes the plays. Ovechkin doesn't do this anymore and it shows in their stats. This doesn't mean Ovechkin hasn't had a great career, as he is 39 years old now, but it just means Crosby is easily the better player now and he has been the better player since 2010 while they both were more or less 1a) 1b) from 2005-'10

It's pretty obvious that the guy who passed to the guy who passed to the goal scorer is less likely to be germane to the play than the guy who passed to the goal scorer or the goal scorer. The eye test should show this for any reasonable hockey fan. The secondary pass is simply less likely to be defensed than the primary pass or the goal.

There are statistical ways of showing this as well. Secondary assists are a volatile and unreliable statistic - far more volatile than primary assists or goals. That's because they are more arbitrary, less indicative, and more dependent on teammates.

Take Crosby's secondary assist totals vs his primary assists totals for example. Let's look at 5 consecutive healthy seasons from 2014 to 2018 (age 26 to 30). This is basically the same player consistently doing the same things. Penguins fans at the time bragged incessantly about how consistent Crosby was.

Primary assists:
34
31
28
28
33

^^^Very consistent. These numbers make sense in that they are roughly aligned with the general value of Sidney Crosby in those seasons - totally in line with what you would expect from an elite player in their prime playing very consistent hockey.

Secondary Assists:
34
25
21
21
17
27

Huh. It's all over the place. Sidney Crosby was...wildly inconsistent? Did Sidney Crosby mysteriously forget how to secondary assist in 2017? Was he half as effective at it relative to 2014? There doesn't appear to be some sort of injury reason for this, and if there was, it somehow didn't impact his primary assists. His EV and PP TOI appear to be pretty stable in those seasons, and his zone starts are generally geared towards offense - as one would expect.

So what's really going on here? It seems to be statistical noise.

 

[Letsdothis]

Career 5vs5 a2/60 leaders

1 H. Sedin
2. D. Toews
3. N. Kucherov
4. C. Makar
5. C. McDavid
6. J.Benn
7. M. St.Louis
8. R. Getzlaf
9. A. Panarin
10. J. Thornton


Powerplay:

1. N. Bäckström
2. C. McDavid
3. M. Rielly
4. R-N. Hopkins
5. M. Marner
6. C. Giroux
7. N. Kucherov
8. V. Hedman
9. A. Panarin
10. S. Ghostisbhere

While more prone to yearly variation, I don't think lists like that would form for a stat that is merellä statistical variance. On the flip side, the bottom 10 is all no-talent goons.

 

[VanIslander]

The answer to the op's question is: no.

 

[Michael Farkas]

Man, if some variance in a stat means that the stat is just noise and not indicative of play, I have some joyous news to spread about save pct. haha

It's almost worth the trade off to agree with that not-well-thought-out rubbish if it means dealing a critical blow to save pct.

 

[Johnny Engine]

Man, if some variance in a stat means that the stat is just noise and not indicative of play, I have some joyous news to spread about save pct. haha

It's almost worth the trade off to agree with that not-well-thought-out rubbish if it means dealing a critical blow to save pct.

Twitter and Patreon-based stats wonk Micah Blake McCurdy had an absolutely wonderful analogy that he explained a while back that really set my mind at ease with stuff like that.
I had heard of the concept of "overfitting" before in the context of "well, you don't want to keep adding qualifiers to expected goals, you'll end up overfitting" and I'm open-minded enough to just say, okay, better not be overfitting, but without actually understanding the significance of that concept.

The (heavily paraphrased) analogy McCurdy used is that, if someone handed you an exact copy of the entire earth, with every single detail preserved, it would not make a good substitute for a map, because it does not isolate the information you'd want to find out from a map.

Jumping off from there to use my own example, If I wanted to get from St. John's, NL to my parents home in Carbonear, looking out my front door (using the earth, which I have handy to myself because the earth generally can be found underneath my feet) will not tell me which way is Carbonear, or even how to get out of my own neighbourhood and onto the outer ring road (hopefully pointed in the right direction).

So a map will tell me how to get to Carbonear. It will generally not tell me that Topsail Road in St. John's is a horrendous mess of potholes and traffic lights and that taking Team Gushue north will get me to the highway in a less direct but a faster and far more comfortable fashion. It also won't tell me that avoiding the highway altogether and taking a winding path through Holyrood and Brigus is not only enjoyable (if you have time for it), but also pretty gas-friendly because you aren't flooring it through these communities.
So there's the argument for needing both models (a map, hockey statistics) and intelligent observation (local knowledge, a scout's eye), and I think most people are comfortable with that axiom.

But I think the hidden takeaway is that models don't have to act like intelligent observation to do their job, otherwise, I'd only be able to get to Carbonear if I'd already done it before. Models and maps can be made more intelligent - a map that only shows that the cities of St. John's and Carbonear are both on the island of Newfoundland and in fact connected by highway is of limited utility - but you can also reach the point where what you have isn't a model but rather a detailed list of everything that has ever existed, and that's not much good to anyone.

The concept and McCurdy's analogy brings me a kind of peace, and I don't mind secondary assists and save percentage if they just sit there and point towards players that might be worth looking at some time.

 

[Crosby2010]

It's pretty obvious that the guy who passed to the guy who passed to the goal scorer is less likely to be germane to the play than the guy who passed to the goal scorer or the goal scorer. The eye test should show this for any reasonable hockey fan. The secondary pass is simply less likely to be defensed than the primary pass or the goal.

There are statistical ways of showing this as well. Secondary assists are a volatile and unreliable statistic - far more volatile than primary assists or goals. That's because they are more arbitrary, less indicative, and more dependent on teammates.

Take Crosby's secondary assist totals vs his primary assists totals for example. Let's look at 5 consecutive healthy seasons from 2014 to 2018 (age 26 to 30). This is basically the same player consistently doing the same things. Penguins fans at the time bragged incessantly about how consistent Crosby was.

Primary assists:
34
31
28
28
33

^^^Very consistent. These numbers make sense in that they are roughly aligned with the general value of Sidney Crosby in those seasons - totally in line with what you would expect from an elite player in their prime playing very consistent hockey.

Secondary Assists:
34
25
21
21
17
27

Huh. It's all over the place. Sidney Crosby was...wildly inconsistent? Did Sidney Crosby mysteriously forget how to secondary assist in 2017? Was he half as effective at it relative to 2014? There doesn't appear to be some sort of injury reason for this, and if there was, it somehow didn't impact his primary assists. His EV and PP TOI appear to be pretty stable in those seasons, and his zone starts are generally geared towards offense - as one would expect.

So what's really going on here? It seems to be statistical noise.

In 2017 Crosby might have been a little bit busier winning the Rocket Richard trophy against the field in the NHL................which included Ovechkin.

Man, you are really reaching if you are looking at his secondary assists. I don't think those are inconsistent at all, it is what you would expect from an elite player and from secondary assists which in general aren't as central to the goal being scored as the primary assist, but are still important.

Here is the spoiler alert, Ovechkin too could have gotten assists if he wanted. But he didn't. Ovechkin went 5 seasons in a row of getting less than 30 assists. Crosby was just a better overall player, and a more broad and diverse offensive talent. He could always beat you in more ways than just passing and scoring.

 

[rmartin65]

It's pretty obvious that the guy who passed to the guy who passed to the goal scorer is less likely to be germane to the play than the guy who passed to the goal scorer or the goal scorer. The eye test should show this for any reasonable hockey fan. The secondary pass is simply less likely to be defensed than the primary pass or the goal.

There are statistical ways of showing this as well. Secondary assists are a volatile and unreliable statistic - far more volatile than primary assists or goals. That's because they are more arbitrary, less indicative, and more dependent on teammates.

Take Crosby's secondary assist totals vs his primary assists totals for example. Let's look at 5 consecutive healthy seasons from 2014 to 2018 (age 26 to 30). This is basically the same player consistently doing the same things. Penguins fans at the time bragged incessantly about how consistent Crosby was.

Primary assists:
34
31
28
28
33

^^^Very consistent. These numbers make sense in that they are roughly aligned with the general value of Sidney Crosby in those seasons - totally in line with what you would expect from an elite player in their prime playing very consistent hockey.

Secondary Assists:
34
25
21
21
17
27

Huh. It's all over the place. Sidney Crosby was...wildly inconsistent? Did Sidney Crosby mysteriously forget how to secondary assist in 2017? Was he half as effective at it relative to 2014? There doesn't appear to be some sort of injury reason for this, and if there was, it somehow didn't impact his primary assists. His EV and PP TOI appear to be pretty stable in those seasons, and his zone starts are generally geared towards offense - as one would expect.

So what's really going on here? It seems to be statistical noise.

These aren't gotcha questions, just some things I became curious about while reading this post-

What does Ovechkin's secondary assist breakdown look like?

What do Ovechkin's ES goals and PP goals look like?

 

[Midnight Judges]

In 2017 Crosby might have been a little bit busier winning the Rocket Richard trophy against the field in the NHL................which included Ovechkin.

Man, you are really reaching if you are looking at his secondary assists.

In a discussion about secondary assists I looked at secondary assists.

Guilty as charged.

 

[Midnight Judges]

These aren't gotcha questions, just some things I became curious about while reading this post-

What does Ovechkin's secondary assist breakdown look like?

It fluctuates randomly, and with practically zero correlation to Oveckin’s individual quality of play.

What do Ovechkin's ES goals and PP goals look like?

Goals in general are the analog here, not situational subsets (which are inherently more volatile).

Ovechkin’s goal totals per season align quite well with explainable events - overall quality of play, injuries, coaching schemes, etc.

 

[Midnight Judges]

Here is the spoiler alert, Ovechkin too could have gotten assists if he wanted. But he didn't.

His role is to score goals and he’s done it better than anyone in history.

When Ovechkin came to Washington they were a lottery team. They’ve won the 3rd most games in the nhl over the span of his career despite not having any other top 200 players of all time on the roster ( is Backstrom top 300?).

They could have easily deployed him in ways that hurt the team but boosted his point totals. He’s an excellent passer, it’s just that his shot is an all-time generational weapon.

As an aside, the goal scorer cannot get credit for the secondary assist. As the player with 45% more goals than anyone else since 2005, these likely do add up a bit more so for Ovie than anyone.

 

[Letsdothis]

Let's compare two players:

Player 1:

p1/gp .672
g/gp .331
ev p1/gp .424
pts/gp .889

Player 2:

p1/gp .581
g/gp .199
ev p1/gp .361
pts/gp .907


Player 1 has a ~15.7% higher primary points p/gp, ~66.3% higher g/pg, and ~17.5% higher primary points per game at even strength (excluding EN situations). If we assess substantially higher value to primary points than overall points, and goals than assists, player 1 would clearly be a couple of tiers ahead of player 2.

The two players in question are Daniel and Henrik Sedin from 2007-2018. I don't think anyone would think we need different tiers for these players, offensively or overall.

Henrik Sedin's line actually scored goals at a higher rate than Daniel Sedin's did when they were separated (although the separation is minimal). This is quite inconsistent with the notion that primary points and goals >>> overall points in value.

 

[MadLuke]

Could it just be a bit of proxy for PP being how well the pp is going ?

when you look at crosby even strength scoring per 60 rates, I am not sure if secondary assist is particularly noisy:

Look how many season are in that 0.54-0.74 range (average +/- 0.1)

Season	Team	GP	TOI	TOI/GP	Goals/60	Total Assists/60	First Assists/60	Second Assists/60	Total Points/60
20072008​	PIT	53​	793.4667​	14.97107​	1.36​	1.97​	1.51​	0.45​	3.33​
20082009​	PIT	77​	1201.917​	15.60931​	1.3​	1.8​	1.3​	0.5​	3.1​
20092010​	PIT	81​	1298.233​	16.02757​	1.66​	1.66​	1.11​	0.55​	3.33​
20102011​	PIT	41​	656.4​	16.00976​	1.92​	2.29​	1.55​	0.73​	4.2​
20112012​	PIT	22​	318.0167​	14.4553​	1.13​	3.77​	2.45​	1.32​	4.91​
20122013​	PIT	36​	576.7​	16.01944​	1.25​	2.81​	1.98​	0.83​	4.06​
20132014​	PIT	80​	1373.367​	17.16708​	1.09​	1.79​	1.05​	0.74​	2.88​
20142015​	PIT	77​	1230.9​	15.98571​	0.88​	1.71​	1.02​	0.68​	2.58​
20152016​	PIT	80​	1308.8​	16.36​	1.19​	1.6​	0.96​	0.64​	2.8​
20162017​	PIT	75​	1210.083​	16.13444​	1.49​	1.69​	1.09​	0.6​	3.17​
20172018​	PIT	82​	1376.133​	16.78211​	0.87​	1.35​	0.78​	0.57​	2.22​
20182019​	PIT	79​	1358.6​	17.19747​	1.02​	2.03​	1.55​	0.49​	3.05​
20192020​	PIT	41​	658.75​	16.06707​	1.09​	1.64​	1.18​	0.46​	2.73​
20202021​	PIT	55​	945.85​	17.19727​	1.14​	1.33​	0.82​	0.51​	2.47​
20212022​	PIT	69​	1136.383​	16.46932​	1.21​	1.64​	1.06​	0.58​	2.85​
20222023​	PIT	82​	1327.167​	16.18496​	1.09​	1.9​	1.18​	0.72​	2.98​
20232024​	PIT	82​	1336.7​	16.30122​	1.44​	1.75​	1.21​	0.54​	3.19​

	goal	assist	a1	a2	pts
Average	1.24​	1.93​	1.28​	0.64​	3.17​
average-deviation	0.20​	0.38​	0.31​	0.14​	0.47​
Ratio	0.16​	0.20​	0.24​	0.21​	0.15​

ratio with career average:

Column1	goal	assist	a1	a2	pts
20072008​	109%​	102%​	118%​	70%​	105%​
20082009​	105%​	93%​	101%​	78%​	98%​
20092010​	134%​	86%​	87%​	86%​	105%​
20102011​	154%​	119%​	121%​	114%​	133%​
20112012​	91%​	196%​	191%​	206%​	155%​
20122013​	101%​	146%​	154%​	129%​	128%​
20132014​	88%​	93%​	82%​	115%​	91%​
20142015​	71%​	89%​	80%​	106%​	81%​
20152016​	96%​	83%​	75%​	100%​	88%​
20162017​	120%​	88%​	85%​	93%​	100%​
20172018​	70%​	70%​	61%​	89%​	70%​
20182019​	82%​	105%​	121%​	76%​	96%​
20192020​	88%​	85%​	92%​	72%​	86%​
20202021​	92%​	69%​	64%​	79%​	78%​
20212022​	97%​	85%​	83%​	90%​	90%​
20222023​	88%​	99%​	92%​	112%​	94%​
20232024​	116%​	91%​	94%​	84%​	101%​

When talking Power play secondary assist, that easy in our head to figure how much where you play on which power play will determine this, those seem indeed all over the place (which make sense considering where-what he do on the PP some season, a Markov blueline quaterback style would maybe be less noisy). While in even strength it is a proxy about how much you drive the play toward goals happening.

Total Points is the less noisy it seem (which make sense would "eat" all the error)

 

[Letsdothis]

Let's compare two players:

Player 1:

p1/gp .672
g/gp .331
ev p1/gp .424
pts/gp .889

Player 2:

p1/gp .581
g/gp .199
ev p1/gp .361
pts/gp .907


Player 1 has a ~15.7% higher primary points p/gp, ~66.3% higher g/pg, and ~17.5% higher primary points per game at even strength (excluding EN situations). If we assess substantially higher value to primary points than overall points, and goals than assists, player 1 would clearly be a couple of tiers ahead of player 2.

The two players in question are Daniel and Henrik Sedin from 2007-2018. I don't think anyone would think we need different tiers for these players, offensively or overall.

Henrik Sedin's line actually scored goals at a higher rate than Daniel Sedin's did when they were separated (although the separation is minimal). This is quite inconsistent with the notion that primary points and goals >>> overall points in value.

Fixed the timeline: indeed this was a sample size of 11 seasons (from 07-18), not just 17 to 18 as I mistyped. If the dat was from one season only, it would obviously not have much relevance.