By The NumbersHockey Analytics... the Final Frontier. Explore strange new worlds, to seek out new algorithms, to boldly go where no one has gone before.
The reason Crosby's 2010 WAS that impressive is because he dominated every inch of the ice, showing a dedication to defense and better overall play that he hadn't shown yet in his career. It was the most consistently dominant I'd seen any player play in my entire life, points be damned.
When was the last even strength more dominant season than 2010 Crosby ?
These numbers are adjusted to 6.00 even strength goals per game.
ES data is only available for the past 20-25 years, except for select players that played into the 90s (e.g. Trottier). ESG data is available back to 1968. I probably forgot some players from the 70s and 80s, but tried to include the some of the best goal scorers from those eras.
Career Adjusted ES Goals:
PLAYER
ESG
Gretzky
735
Jagr
718
Hull
649
Selanne
622
Esposito
597
Robitaille
590
Shanahan
585
Yzerman
582
Messier
578
Sakic
577
Sundin
569
Dionne
556
Modano
555
Iginla
541
Recchi
524
Lemieux
515
Tkachuk
489
Lafleur
480
Kurri
467
Francis
464
Career Adjusted ES Assists
PLAYER
ESA
Gretzky
1,458
Jagr
958
Francis
940
Messier
927
Oates
909
Yzerman
862
Sakic
859
Recchi
858
Gilmour
810
Dionne
798*
Thornton
745
Esposito
737*
Turgeon
723
Trottier
712
Sundin
711
Coffey
710
Lemieux
707
Bourque
699
Modano
685
Lafleur
673*
* estimated using % of goals at ES each season
Career Adjusted ES Points
PLAYER
ESP
Gretzky
2,193
Jagr
1,676
Messier
1,505
Yzerman
1,444
Sakic
1,436
Francis
1,403
Recchi
1,381
Dionne
1,354*
Esposito
1,334*
Sundin
1,280
Selanne
1,265
Robitaille
1,240
Modano
1,240
Oates
1,227
Lemieux
1,222
Shanahan
1,198
Hull
1,194
Turgeon
1,186
Gilmour
1,178
Lafleur
1,153*
Last edited by Czech Your Math: 09-15-2012 at 06:37 PM.
Wow. Iginla is alot better ES then I expected. The only people he is behind consistently are some of the 5-10 best players to ever play the game. In goals at least. Hmm
We know a higher percentage of overall offense comes from first liners now than in the 1980s. Is this not still true at even strength?
This is true at even strength. The distribution question is an interesting one.
Here are some numbers for a selection of NHL seasons since expansion.
Year
1
2
3
4
5
6
7
8
9
10
11
12
1968
28.8%
25.9%
23.7%
22.1%
19.7%
17.8%
16.6%
14.6%
12.3%
9.4%
7.2%
5.6%
1973
30.0%
27.0%
24.3%
21.7%
20.3%
18.2%
16.3%
13.9%
12.0%
9.7%
7.0%
4.9%
1978
29.1%
23.7%
21.1%
19.2%
17.7%
16.6%
14.7%
13.4%
11.9%
10.0%
8.1%
5.9%
1983
27.9%
24.4%
21.6%
18.8%
17.1%
15.3%
14.1%
12.7%
11.1%
9.7%
8.1%
6.4%
1988
28.8%
24.5%
21.4%
19.5%
17.3%
14.9%
13.9%
12.0%
10.8%
9.1%
7.6%
6.3%
1993
30.4%
24.8%
21.4%
19.8%
17.4%
15.1%
13.4%
11.5%
10.3%
8.3%
7.3%
6.2%
1998
31.9%
24.6%
21.3%
19.3%
17.3%
15.5%
13.2%
11.5%
10.0%
8.8%
6.9%
6.0%
2003
30.9%
26.6%
23.5%
20.0%
17.5%
15.5%
13.5%
12.2%
10.7%
9.0%
7.5%
6.2%
2008
32.6%
27.5%
24.0%
21.1%
17.8%
16.0%
14.2%
12.3%
10.7%
9.2%
7.8%
6.2%
2012
30.5%
26.4%
23.6%
21.0%
18.1%
15.3%
13.6%
11.9%
10.0%
8.4%
6.7%
5.2%
Year
1-3 (First line)
4-6 (Second line)
7-9 (Third line)
10-12 (Fourth line)
1-6 (Top six)
7-12 (Bottom six)
1968
26.1%
19.9%
14.5%
7.4%
23.0%
11.0%
1973
27.1%
20.0%
14.1%
7.2%
23.6%
10.6%
1978
24.7%
17.8%
13.3%
8.0%
21.3%
10.7%
1983
24.7%
17.1%
12.6%
8.0%
20.9%
10.3%
1988
24.9%
17.3%
12.2%
7.7%
21.1%
10.0%
1993
25.6%
17.4%
11.8%
7.3%
21.5%
9.5%
1998
26.0%
17.4%
11.5%
7.2%
21.7%
9.4%
2003
27.0%
17.6%
12.2%
7.5%
22.3%
9.9%
2008
28.0%
18.3%
12.4%
7.8%
23.2%
10.1%
2012
26.8%
18.1%
11.8%
6.8%
22.5%
9.3%
The percentages posted are for the average percentage of team even strength goals for which the Nth highest even strength scoring forward on the team received a point.
So in 1968, the average team had the top EV scorer post a point on 28.8% of even strength goals. This number rose by 1973, dropped and reached a low point in the 1980s, and then rose to a high point in the late 90s and 00s.
Compare Wayne Gretzky's 1982-83 season to Jaromir Jagr's 1997-98 season.
Gretzky scored 133 even strength points in a context where 1. The average team scored 228.9 even strength goals, and 2. The average team leading ES scorer recorded a point on 27.9% of team ES goals.
Jagr scored 64 even strength points in a context where 1. The average team scored 148.9 even strength goals, and 2. The average team leading ES scorer recorded a point on 31.9% of team ES goals.
These simple adjusted points adjust for 1. but not 2.
That's some interesting info and I didn't have access to that. Where did you get the data for those results?
Any theories as to why first liners score a higher proportion of points than in the 80s? It could be that once all of the European/Russian forwards were free to come to the NHL, they did so disproportionately, so that the top line forwards were relatively better compared to the 3rd/4th liners than back in the 80s. It could be that 3rd/4th liners were expected to score more back in the 80s, and are relied on more defensively in more recent years. It could be that with better goalie technique/equipment, it became disproportionately more difficult for 3rd/4th liners to score than for top line forwards. It may be a combination of these factors, but I'm interested to hear other theories and how any/all can be substantiated with data.
Last edited by Czech Your Math: 09-16-2012 at 03:42 AM.
That's some interesting info and I didn't have access to that. Where did you get the data for those results?
Any theories as to why first liners score a higher proportion of points than in the 80s? It could be that once all of the European/Russian forwards were free to come to the NHL, they did so disproportionately, so that the top line forwards were relatively better compared to the 3rd/4th liners than back in the 80s. It could be that 3rd/4th liners were expected to score more back in the 80s, and are relied on more defensively in more recent years. It could be that with better goalie technique/equipment, it became disproportionately more difficult for 3rd/4th liners to score than for top line forwards. It may be a combination of these factors, but I'm interested to hear other theories and how any/all can be substantiated with data.
I think you hit the three possible explanations, or at least the only ones I could think of. It is definitely true that lower lines are coached to play more defensively now that in the 80s. The 80s were pretty unique in NHL history in that lower lines were often used simply to outscored the opponent's lower lines, rather than as primarily checking units. The other two reasons you give are probably true, as well. Good call on the "Europeans coming over tend to be skilled scorers" one - I hadn't seen it posted on the history board yet.
My guess is that the biggest factor is a change in the way lower lines are used by coaches, but I think the other two factors probably contribute, as well.
Now mathematically accounting for these factors - that's the hard part! Clearly more recent first liners should not get extra credit for lower line players having more difficulty scoring or or the way coaches use lower lines. But they probably should get credit for the fact that more of them are skilled Euros.
For now, I think the most accurate estimatation would be to assume the greater difficulty in first liners from the 80s getting adjusted points is all due to external factors of era, and a second adjustment should be made. Unless somehow you can find a way to tease out any disproportionate impact of Europeans coming over.
My guess is the impact of Europeans is a minor factor though - considering the percentages today resemble percentages from the 70s (correct?) and it's really just the 80s that is the major outlier, right?
The short version of my ramble post - if percentage of offense coming from first liners in the 1970s was similar to today (and I think it was), the effect of Europeans coming over on the percentages is probably minimal. I think coaching, specifically the usage of lower lines in the 80s, is probably the biggest factor
For now, I think the most accurate estimatation would be to assume the greater difficulty in first liners from the 80s getting adjusted points is all due to external factors of era, and a second adjustment should be made. Unless somehow you can find a way to tease out any disproportionate impact of Europeans coming over.
I think it could be examined at least. Europeans & Russians really started coming over around '89 or '90. So one could look at the late 80s vs. the early 90s (through ~'93 or '94, before defense became so emphasized) and see if there was a change in the early 90s as more European/Russian stars entered the NHL.
Quote:
Originally Posted by TheDevilMadeMe
My guess is the impact of Europeans is a minor factor though - considering the percentages today resemble percentages from the 70s (correct?) and it's really just the 80s that is the major outlier, right?
The data Overpass presented looks similar for the 70s vs. the 90s/00s, but if one was doing "extra" adjustments, it should be considered that adjusted points were much easier for top line players in the 70s than in later decades (I assume this would apply to just ES points as well).
That's some interesting info and I didn't have access to that. Where did you get the data for those results?
Any theories as to why first liners score a higher proportion of points than in the 80s? It could be that once all of the European/Russian forwards were free to come to the NHL, they did so disproportionately, so that the top line forwards were relatively better compared to the 3rd/4th liners than back in the 80s. It could be that 3rd/4th liners were expected to score more back in the 80s, and are relied on more defensively in more recent years. It could be that with better goalie technique/equipment, it became disproportionately more difficult for 3rd/4th liners to score than for top line forwards. It may be a combination of these factors, but I'm interested to hear other theories and how any/all can be substantiated with data.
ESG data is from official NHL stats.
ES assists are official from 1987-88. Before that time they are estimates, but I have replaced the estimates for most prominent players with unofficial boxscore data from the HSP. The estimates may add a little error to the results pre-88 but should not make much difference to the top end.
I'm planning to make a full post at some point with distribution stats of this type for all years and both F and D.
The short version of my ramble post - if percentage of offense coming from first liners in the 1970s was similar to today (and I think it was), the effect of Europeans coming over on the percentages is probably minimal. I think coaching, specifically the usage of lower lines in the 80s, is probably the biggest factor
In 1973 there were 16 teams. In 1978 there were 18. In 1983 and 1988 there were 21. In 1993 there were 24 teams, 26 in 1998, and all later years in the grid have 30 teams.
I can't speak for the high number in the early 70s, but the later trend seems to indicate "more teams=more balanced scoring". One guess I might have is that 1973 did not use divisions, while the later years did.
But if you want a great look at how teams in the 80s were all basically the same aside from your top few players, compare the 1986 Oilers and the 1986 Leafs. Take off the top four scoring forwards and the top scoring defenseman from each team, and look at the roster. It's almost statistically identical; Toronto is actually offensively more powerful at that point.
In 1973 there were 16 teams. In 1978 there were 18. In 1983 and 1988 there were 21. In 1993 there were 24 teams, 26 in 1998, and all later years in the grid have 30 teams.
I can't speak for the high number in the early 70s, but the later trend seems to indicate "more teams=more balanced scoring". One guess I might have is that 1973 did not use divisions, while the later years did.
But if you want a great look at how teams in the 80s were all basically the same aside from your top few players, compare the 1986 Oilers and the 1986 Leafs. Take off the top four scoring forwards and the top scoring defenseman from each team, and look at the roster. It's almost statistically identical; Toronto is actually offensively more powerful at that point.
That's not what the data shows. Scoring in the 1980s was more balanced than it has been from the mid 1990s to today.
I'm not sure if the data supports this, but it seems to that expansion should naturally lead to less scoring by the lower lines.
I.e. in a 12 team league, there are 72 players playing on the top two lines. In a 30 team league all of those (theoretically), plus 18 others would be playing on the first line. The second line would consist of many players that would not even have made it in a 12 team league.
Intuitively this would mean that difference between the average level of play on the first and second line should increase.
I'm not sure if the data supports this, but it seems to that expansion should naturally lead to less scoring by the lower lines.
I.e. in a 12 team league, there are 72 players playing on the top two lines. In a 30 team league all of those (theoretically), plus 18 others would be playing on the first line. The second line would consist of many players that would not even have made it in a 12 team league.
Intuitively this would mean that difference between the average level of play on the first and second line should increase.
Perhaps the absorption of the WHA into the NHL is what led to lower lines scoring more in the 80s than in the 70s? It's possible.
The percentages posted are for the average percentage of team even strength goals for which the Nth highest even strength scoring forward on the team received a point.
So in 1968, the average team had the top EV scorer post a point on 28.8% of even strength goals. This number rose by 1973, dropped and reached a low point in the 1980s, and then rose to a high point in the late 90s and 00s.
It would seem using the Nth highest scoring (ES) player on each team makes the metric influenced by league parity. In particular for top (line) players, the more balanced the league is in terms of top talent distribution, the higher the top tiers would be expected to score compared to bottom tiers. For instance, I would expect top players to be lower scoring compared to bottom tiers of players during the early-mid 70s, because I would expect their bottom lines were proportionately closer to those of the powerhouses than the top lines were.
I know it's difficult to filter out the influences of other factors, but do you think this factor could have a significant affect on the results by tier?
Last edited by Czech Your Math: 11-04-2012 at 03:55 AM.
