HabsCast Hockey Blog

I recently did a study of top NHL scorers by year attempting to adjust for era to put each players Goals and Assists on an equal playing field. My methodology was as follows:

This study only involved NHL teams and players.

I studied goals per game (goals/game) and assists per game (assists/game) by season. I could instead have studied raw goals and assists and then produced a rate statistic afterwards but this would be more complicated due to the fact that I used regression models. Thus everyone’s goals and assists would be equalized but games played would not be since earlier eras played between 20 and 50 games per season compared to the modern 80+. This would not have been a terribly difficult thing to compensate for, but I preferred my way and I believe the results would have been extremely similar using either method. In light of the above, only players who played in 50% of the their team’s games were counted. In addition, my database only contained player’s who played at least 20 games in a season. This makes some of the findings on the late teens and early 20s suspect although its a fair sight better than some of the other systems that wildly overrate early player’s statistics

Despite a very large gap in numbers of teams and players in the league, I made the simplifying assumption that the Top 25 is the Top 25. That is, even if, in the 80s, there were 500 players in the database for a season but only 100 for years in the 60s, I studied each years Top 25 in Goals and Assists.

There are any number of reasons for this:

1. If I were to choose a number each season based on percentage, it biased the results toward the modern era by driving up averages for early years and driving down averages for modern years.

2. Less than 25 data points seemed to allow individaul dominance to skew the data too much (for instance Wayne Gretzky’s assists in the mid-80s)

3. If the talent level of the Top 25 offensive players does not remain relatively stable, it calls into question whether a valid comparison can be made.

4. The survey results are extraordinarily consistent, suggesting that this assumption is indeed quite defensible.

5. I did not use Markov Chains or more sophisticated techniques because this was done in my spare time sd simplifying assumptions were necessary or I would have simply given up. It would be a very interesting attempt to quantify league strength from year to year and adjust for this.

Further, I have in mind a rather accurate way to adjust for stength of team and strength of opposition from 1967 onwards, something I will be working on in the future.

The statistics I produced were a player’s projected goals and assists projected over a full 82 game schedule and adjusted for the era(s) in which they played. I have also tacked on an additional stat called Goals Created.

Using a very, very simplified formula for a stat that can be analyzed much more rigorously, I assigned the following credit: play-making and goal-scoring each account for 50% of each goal scored. Each goal scored is thus given a value of .5 (if anything number crunching suggests this is a bit too high). But for assists, we first have to adjust for the fact that there are 5 assists for each 3 goals, roughly speaking (it is normally between 1.65 and 1.75 assist/goal). That means each Assist is multiplied by 3/5 and then divided by 1/2 or in others multiplied by .3

This gives a more than decent approximation for how many goals a player individually accounted for.

Onto some lists

(NOTE: rather than normalize the stats to some hypothetical average I decided to put them in the context of 80s and early 90s scoring for fun. So, yes, the stats are inflated):

Top 10 Goal Scoring Seasons:

Bobby Hull 104 1965-66
Bobby Hull 102 1966-67
Phil Esposito 101 1970-71
Wayne Gretzky 101 1981-82
Brett Hull 101 1990-91
Wayne Gretzky 100 1983-84
Gordie Howe 99 1952-53
Bobby Hull 99 1961-62
Maurice Richard 98 1946-47
Charlie Conacher 97 1934-35

others with 90 goal years: Bill Cook, Lynn Patrick, Jean Beliveau, Bernie Geoffrion, Mario Lemieux, Mike Bossy, Jari Kurri, Pavel Bure, Cam Neely, Jarome Iginla

Top 10 Assist Seasons

Wayne Gretzky 140 1986-87
Wayne Gretzky 139 1985-1986
Wayne Gretzky 138 1982-83
Wayne Gretzky 138 1984-85
Bill Cowley 136 1940-41
Wayne Gretzky 136 1983-84
Ted Lindsay 135 1949-50
Bobby Orr 135 1969-70
Bill Cowley 134 1938-39
Bobby Orr 133 1970-71
Wayne Gretzky 133 1981-82

others with 125 Assist seasons: Georges Boucher, Dick Irvin, Joe Primeau, Elmer Lach, Phil Esposito, Stan Mikita, Gordie Howe, Mario Lemieux, Jaromir Jagr, Peter Forsberg, Joe Thornton

Goals Created (see above for my formula) — I will edit in years later

1. Wayne Gretzky 95
1. Wayne Gretzky 95
3. Wayne Gretzky 90
4. Mario Lemieux 89
4. Mario Lemieux 89
6. Gordie Howe 88
6. Mario Lemieux 88
6. Wayne Gretzky 88
9. Wayne Gretzky 87
10. Bobby Hull 86
10. Phil Esposito 86

others with 80 GC years: Jaromir Jagr, Guy LaFleur, Charlie Conacher, Howie Morenz, Jean Beliveau. (That makes only 10 players in NHL history — Bernie Geoffrion and Jean Ratelle both came close with a 79 GC season each)

[Stats via HFBoards.com]

Digg - StumbleUpon - Reddit - Del.icio.us - Technorati - Email - Etc...

If you enjoyed this post, make sure you subscribe to my RSS feed!