- Jan 16, 2005
- 9,222
- 17,283
* Updated through 2023
As most of you know, the NHL media votes on all-star teams by position (one space each for left wing, centre, right wing and goalie; two spaces for defense) at the end of each season. Based on the voting results, the NHL announces first-team and second-team all-stars. This tradition dates back to the 1930-31 season.
In August 2007, I did a research project on the NHL’s “third-team all-stars”. Through a combination of direct research and compiling information already available on this site, I assembled the NHL third-team all-stars from 1965-2007 (updated annually thereafter). Over time, I became frustrated that I had done a poor job of systematically organizing and analyzing the data. The purpose of this thread is to assemble all the relevant data in one place. If there are errors in the data, please let me know.
==========
Why bother looking at third-team all-stars at all?
Conventional wisdom suggests that we should look at only the top two players at each position (four for defensemen). This never made sense to me; there are always great performers beyond the players on the first two teams. This project helps highlight players who had very good seasons, but perhaps not the very best.
Why stop at the third team? Why not look at fourth, fifth, etc. team all-stars?
In general, the top three players usually earn the majority of the votes. I don’t find it interesting or informative to spend time evaluating players who are fighting over a few throwaway votes. There may be some value in examining fourth-team all-stars but I don’t have time to assemble this data. If someone wants to do so, I’d encourage them.
What are the biggest weaknesses of looking at all-star spots?
The most obvious weakness is that the level of competition at each position has varied over time. For example, in 1989 Steve Yzerman had an exceptional season – however, he had the misfortune of competing against Mario Lemieux (who scored 85 goals and 199 points) and Wayne Gretzky (who scored 168 points while singlehandedly turning around the Kings franchise). There have been some very mediocre third-, second- and even first-team selections during years with weaker competition (Jim Carey as a first-team all-star in 1996 stands out as a particularly weak selection – though he won the Vezina that year). Also, as with all regular season awards, the playoffs are not taken into account.
How do you “weigh” first-team spots relative to second- and third-team spots?
I considered giving different weightings to the awards (i.e. perhaps a first-team selection is worth five points and second-team selection is worth three) but I decided not to. I want the data to speak for itself. It’s obvious that Scott Niedermayer (3-1-0) had a fairly short but brilliant peak, while Serge Savard had a lengthier but lower level of success (0-1-5). Turning this into a junk statistic by adding weightings adds little value.
Why are you starting at the 1967-68 season?
From a practical standpoint, I needed to start somewhere. I have continuous all-star voting data for each year from 1967-68 to the present, so that season seemed to be the logical starting point. (The 1967-68 season, obviously, was the first year after expansion, but that’s not why it was chosen as the beginning). The tables below include data for some players (ie Gordie Howe, Bobby Hull, etc.) whose careers precede 1967-68.
How have the voting procedures changed over time?
There isn’t much detail about the voting procedures from 1968 to 1981. For many (most?) of the years, the all-star teams, along with other major awards, were voted upon twice – once halfway through the year, and once at the end. Presumably this was to prevent a late-season hot streak from diminishing a steady effort all season long.
The voting procedures have been consistent from 1982 to present (except 1995). Each voter is given three votes for each position (six for defense); first-place votes are worth 5 points, second-place votes are worth 3 points and third-place votes are worth 1 point.
In 1995 (one of the two shortened seasons), each of the two conferences nominated three finalists for each position (six for defensemen); the exact mechanism for this is unclear. A committee of fifteen members of the Professional Hockey Writers’ Association voted on the winners from the six conference representatives (twelve for defensemen). This yielded some strange results because some players finished with zero votes, For example, Joe Nieuwendyk was one of three Western conference representatives at centre but, in the final, official vote, he wasn’t named in the top three on any writers’ ballot. Thus he (in addition to Al MacInnis and Curtis Leschyshyn) were finalists, while earning zero official votes!
What are the weirdest voting results you’ve found?
See above for the weird results from 1995.
In 1977, Marcel Dionne (a centre) was named the first-team all-star centre. He also received the 3rd-most votes as a RW. The actually right-winger with the most votes was the second-best #99 in history, Wilf Paiment. In the following tables I’ve treated Paiment as the third-team RW for 1977.
In 2012, Claude Giroux finished third in all-star voting both at the centre and right wing positions. Giroux is primarily a centre, so let's give him credit for a spot as third-team centre. Marian Hossa gets bumped up to third team right wing.
Also in 2012, James Neal was the first-team all-star right wing, and an extremely close third in left-wing voting. In fact, a single second-place vote (or two third-place votes) would have resulted in Neal officially being an all-star at two positions simultaneously! I consider Neal to be primarily a RW, so the third LW position goes to Scott Hartnell.
See post #22 for a long discussion about the voting debacle in 2013.
What are the significant judgments/assumptions you’ve had to make in compiling the data?
See above.
Where can I see the original post?
http://hfboards.mandatory.com/showthread.php?t=407941
As most of you know, the NHL media votes on all-star teams by position (one space each for left wing, centre, right wing and goalie; two spaces for defense) at the end of each season. Based on the voting results, the NHL announces first-team and second-team all-stars. This tradition dates back to the 1930-31 season.
In August 2007, I did a research project on the NHL’s “third-team all-stars”. Through a combination of direct research and compiling information already available on this site, I assembled the NHL third-team all-stars from 1965-2007 (updated annually thereafter). Over time, I became frustrated that I had done a poor job of systematically organizing and analyzing the data. The purpose of this thread is to assemble all the relevant data in one place. If there are errors in the data, please let me know.
==========
Why bother looking at third-team all-stars at all?
Conventional wisdom suggests that we should look at only the top two players at each position (four for defensemen). This never made sense to me; there are always great performers beyond the players on the first two teams. This project helps highlight players who had very good seasons, but perhaps not the very best.
Why stop at the third team? Why not look at fourth, fifth, etc. team all-stars?
In general, the top three players usually earn the majority of the votes. I don’t find it interesting or informative to spend time evaluating players who are fighting over a few throwaway votes. There may be some value in examining fourth-team all-stars but I don’t have time to assemble this data. If someone wants to do so, I’d encourage them.
What are the biggest weaknesses of looking at all-star spots?
The most obvious weakness is that the level of competition at each position has varied over time. For example, in 1989 Steve Yzerman had an exceptional season – however, he had the misfortune of competing against Mario Lemieux (who scored 85 goals and 199 points) and Wayne Gretzky (who scored 168 points while singlehandedly turning around the Kings franchise). There have been some very mediocre third-, second- and even first-team selections during years with weaker competition (Jim Carey as a first-team all-star in 1996 stands out as a particularly weak selection – though he won the Vezina that year). Also, as with all regular season awards, the playoffs are not taken into account.
How do you “weigh” first-team spots relative to second- and third-team spots?
I considered giving different weightings to the awards (i.e. perhaps a first-team selection is worth five points and second-team selection is worth three) but I decided not to. I want the data to speak for itself. It’s obvious that Scott Niedermayer (3-1-0) had a fairly short but brilliant peak, while Serge Savard had a lengthier but lower level of success (0-1-5). Turning this into a junk statistic by adding weightings adds little value.
Why are you starting at the 1967-68 season?
From a practical standpoint, I needed to start somewhere. I have continuous all-star voting data for each year from 1967-68 to the present, so that season seemed to be the logical starting point. (The 1967-68 season, obviously, was the first year after expansion, but that’s not why it was chosen as the beginning). The tables below include data for some players (ie Gordie Howe, Bobby Hull, etc.) whose careers precede 1967-68.
How have the voting procedures changed over time?
There isn’t much detail about the voting procedures from 1968 to 1981. For many (most?) of the years, the all-star teams, along with other major awards, were voted upon twice – once halfway through the year, and once at the end. Presumably this was to prevent a late-season hot streak from diminishing a steady effort all season long.
The voting procedures have been consistent from 1982 to present (except 1995). Each voter is given three votes for each position (six for defense); first-place votes are worth 5 points, second-place votes are worth 3 points and third-place votes are worth 1 point.
In 1995 (one of the two shortened seasons), each of the two conferences nominated three finalists for each position (six for defensemen); the exact mechanism for this is unclear. A committee of fifteen members of the Professional Hockey Writers’ Association voted on the winners from the six conference representatives (twelve for defensemen). This yielded some strange results because some players finished with zero votes, For example, Joe Nieuwendyk was one of three Western conference representatives at centre but, in the final, official vote, he wasn’t named in the top three on any writers’ ballot. Thus he (in addition to Al MacInnis and Curtis Leschyshyn) were finalists, while earning zero official votes!
What are the weirdest voting results you’ve found?
See above for the weird results from 1995.
In 1977, Marcel Dionne (a centre) was named the first-team all-star centre. He also received the 3rd-most votes as a RW. The actually right-winger with the most votes was the second-best #99 in history, Wilf Paiment. In the following tables I’ve treated Paiment as the third-team RW for 1977.
In 2012, Claude Giroux finished third in all-star voting both at the centre and right wing positions. Giroux is primarily a centre, so let's give him credit for a spot as third-team centre. Marian Hossa gets bumped up to third team right wing.
Also in 2012, James Neal was the first-team all-star right wing, and an extremely close third in left-wing voting. In fact, a single second-place vote (or two third-place votes) would have resulted in Neal officially being an all-star at two positions simultaneously! I consider Neal to be primarily a RW, so the third LW position goes to Scott Hartnell.
See post #22 for a long discussion about the voting debacle in 2013.
What are the significant judgments/assumptions you’ve had to make in compiling the data?
See above.
Where can I see the original post?
http://hfboards.mandatory.com/showthread.php?t=407941
Last edited:
