Comparing NHL Greats

One of the biggest challenges in comparing the best modern-day players to those of previous eras is the changes in the the way the game is played. Season length, roster size, and schedule strength have all changed considerably since 1917, and goalies have gotten significantly better - the stats show us that. The stats also show us, for instance, that the average number of goals in a game spiked in the 80s. This all has to be factored in to get an accurate comparison of players between the years. Read on to see the playing field levelled and how the modern era stacks up.

January 2016
Data Visualization
Art-Ross_Actual

Every leading scorer’s stats since 1917. Gretzky’s multiple 200+ point seasons in the 80s show a clear change in gameplay style and are almost incomparable to the 100 point marks of the 2000s.

The game of hockey has changed a lot in the past 100 years. Rosters have gotten bigger, the league has gotten larger, goalies have gotten better, and the rules have changed a great deal. One thing, however, has stayed pretty much constant: the way we tally goals and assists. A goal in the game last night is worth the same at the end of the season as it was in a game forty-five years ago. So why has nobody been able to compete with Wayne Gretzky or Mario Lemieux’s incredible seasons of the expansion era? Well, it’s simple. The league has gotten larger, goalies have gotten better, and the rules have changed a great deal. It’s not exactly fair to compare players between eras.

To account for this, I’ve created a series of adjustments that we can apply to every player’s statistics to even the playing field. I want to see how every player would do if they all theoretically played for the same team (separately, of course) in the average post-lockout (2004-05) season – meaning 82 games on the schedule, 18 players on a roster, 5.659 goals per game, 1.719 assists per goal, and an average save percentage of .911.

My first adjustment accounts for the difference in season length. Back in 1919/20, there were only four NHL teams and they each played twenty-four games. Now, there are thirty teams, each playing a hefty 82 games. Obviously, a shorter season correlates to a smaller point total, even for Joe Malone, who scored 48 points for the Quebec Athletic Club in 1919/20. This equates to players with shorter seasons being given a handicap, increasing their potential point total.

calculations_Schedule

To even everyone out, I’ve divided the average number of games played by each team, post-lockout (not including the shortened 2012/13 season), and divided it by the average number of games played in the stat year.  For example, let’s look at the first season of the expansion era, which was led by Stan Mikita of the Chicago Blackhawks. The season length in 1967/68 was 74 games. Our current 82 games divided by that length gives us a factor of 1.108. As you’d expect, a higher number means a shorter season, giving the player the aforementioned handicap. This would effectively increase his point total from 87 to 93.

Of course, it’s not as simple as just multiplying someone’s point total in their respective season by a schedule factor to bring them up to an 82-game season, no sir, it is not. If that were the case, the boys from the early years would see a huge statistical improvement while guys who played in the nineties wouldn’t see any change at all. The playing field still isn’t equal, but we’re getting there.

We also have to account for the difference in roster sizes. In 1917, teams were allowed a maximum of twelve players. It wasn’t until 1982 that roster sizes were increased to the current size of eighteen skaters and two goalies. They went up and down a few times in between, but the point is, they’ve gotten bigger. Since game length has remained at roughly the same three twenty-minute periods, this means that back in the formative years of the league, when Joe Malone scored 44 goals in one season, he actually had more ice time every game than say, Sidney Crosby does in the era of eighteen man rosters. So the guys playing in pre-1982 seasons had a slight advantage in that they had more time to score.

calculations_Roster

To make up for this, let’s divide the roster size in the stat year and divide it by the current roster size of 18. Let’s look again at 1967/68. The Black Hawks had a 16-man team, which, when divided by 18, gives us a roster factor of .889. Opposite to the schedule changes, a lower number means a small roster, which means more playing time, so a decreased point total is in order. Multiplying Mikita’s new points total of 93 by .889 lowers his total to 83.

The next adjustment address the change in playing style and goalie skill over the four main eras of the league. The Wayne Gretzkys of the high-flying 80s, playing in an era of increased scoring, poor goaltending, and weak expansion teams to prey on, obviously had an easier time scoring than the Peter Forsbergs of the recent dead-puck era, who played in a time of butterfly goaltending and the Trap. Likewise, guys who played in the league’s early years played against statistically dominant defences, but played through the introduction of icing, offsides, and forward passing, which lead to a surge in scoring by the Original Six era. To truly even out the playing circumstances, we divide the average league-wide goals per game (LGPG) in the post-lockout years by the total goals scored throughout the league (not including the player in question’s), in turn divided by the total number of games played in the league in that particular year.

calculations_Era-G

Seeing as the average number of assists per goal has risen dramatically since the 20s, we do the same with assists totals. We first multiple the average league-wide assists per goal, post-lockout (LAPGoal) by the post-lockout LGPG and divide that by the total league-wide assists in the stat year, minus the player’s, divided by the games played that season.

calculations_Era-A

Back to our Stan Mikita example: the league combined for 2476 goals and 4014 assists in 444 games in 1967/68, while Mikita scored 40G and 47A. The post-lockout goals-per-game is 5.657 and assists-per-goal are up at 1.719. All of this, multiplied by Mikita’s previously-adjusted point totals, ups his numbers to 41G and 50A.

What this means is that the guys playing in seasons in which more goals were scored league-wide will have their numbers reduced to more closely coincide with the number of goals scored in the modern, post-2004-2005-lockout era.

Now I know what you’re thinking, “but Sam, didn’t teams like the 1970/71 Boston Bruins and the 1985/86 Edmonton Oilers have a much easier schedule than say, the 1967/68 Chicago Blackhawks and especially the 1919/20 Quebec Athletic Club?” Well to that, I say you’re absolutely right. Teams like the Bruins in the 70s and the Oilers in the 80s had statistically easier schedules than most. That’s because of expansion. You see, expansion teams, on average, weren’t very good. They were new to the league and less established than the Original Six, and teams that joined the NHL out of the WHA were much better than teams that were new to either league. This means that teams playing in the same division as a weaker expansion team had an easier schedule than teams playing in a division without weak expansion teams, or if you get down to it, the expansion teams themselves. We measure this with a stat called Strength of Schedule, or S.O.S.. The average S.O.S. is always 0, and the lower the number, the weaker, on average, the team’s opponents., while the higher the number, the stronger, on average, the team’s opponents.

To really even the playing field and compare players through the years, we have to account for how difficult the player’s respective team’s schedule was. Wayne Gretzky playing for an Oilers team in 1985 had a much easier season than even Jamie Benn playing for the Dallas Stars in 2014/15. So we’ll factor that in too, by converting the average S.O.S. to one and multiplying it by our player’s totals and all of our other adjustments.

Art-Ross_All

This is the result. Wayne Gretzky and Mario Lemieux still dominate, but the numbers are definitely closer. The adjustments are more obvious when we look at just goal-scoring (below).

The results for Art Ross Trophy winners is as follows. The first column is the actual top ten seasons in the league’s history. To the right is the top ten seasons using adjusted points, not accounting for schedule strength. The third column factors in schedule strength for post-1967 seasons. If we include early seasons, Joe Malone’s 49 points in 1920 jump to a monster 211, simply because his Quebec Athletic Club, a new team, had an unheard of S.O.S. number of .90 (it was a tough year for the Bulldogs). Still, there are no modern (post-2005-lockout) players in the top ten. Joe Thornton is closest, though, jumping from 27th to 12th, with 123 points.

Art-Ross_toptens

Doing the same for Rocket Richard Trophy winners (as it’s been called since 1998) gives us these results. Ovechkin cracks in at fourth with 67 goals (close to his actual 65 goal total), but if we factor in the Washington Capitals’ relatively easy schedule in 07/08, his total is bumped down to just 54. Steven Stamkos, on the other hand, is sitting pretty with 62.

Rocket-Richard_toptens

The full visualizations are included here:

Of course, there’s more to it. Equipment has improved – goalie pads are lighter, sticks are curved and more flexible, helmets actually exist. Players are getting bigger, stronger, and faster, making it harder for offensive players to find space. The salary cap exists now – teams have never been more even, and adjusting for strength of schedule only begins to acknowledge this. A team like the Oilers of the 80s wouldn’t, or couldn’t, exist in today’s league. This brings up the skill level of leading scorers’ teammates: the better the team, the more of an advantage each player has when it comes to scoring.

Like with every professional sport, the technology has improved. Tape and footage exist for every game, from every possible angle, and there’s statistics based on pretty much everything. This unprecedented access to each team’s systems and statistics works two ways: offensive players have the book on the goalies they’re up against and the defensive systems of each team they’ll play in the season, meaning they should have an advantage their adjusted numbers lowered to make up for that. At the same time, though, defensive teams have the book on all the top scorers. The goalies know where they shoot and the defenseman know their plays. Maybe these cancel each other out? I’m not sure, but I can also say maybe it doesn’t matter. Maybe this technology doesn’t have any real affect on what’s happening on the ice. An elite goalie will make the save regardless of what they’ve seen in the video room and a natural goal scorer will find a way to put the puck in the net no matter what.

It might not be perfect, but the graphs above, at least, begin to even out the playing field between NHL eras. The question for the past few years has always been, is Ovechkin as good a goal-scorer as the greats like Gretzky, Lemieux, and Hull? Based on the stats, I’d have to say no. But Stamkos comes pretty close.

View my data here.

 


 

Notes and Sources
Designed with by Sam Vickars (www.samvickars.is).
© Copyright 2015, all rights reserved. Trademarks property of their respective owners.

Some data used herein was obtained free of charge from and is copyrighted by The Hockey Summary Project. For more information about The Hockey Summary Project, please visit this link. Other information used was obtained free of charge from and is copyrighted by The Hockey Databank Project. For more information, please visit this link.Other sources include:
Books
National Hockey League Guide (various years)
National Hockey League Official Record Book (1982-83 and 1983-84)
National Hockey League Official Guide & Record Book (1984-85 to present)
The Stanley Cup Records and Statistics (various years)
World Hockey Association Media Guide (various years)
WHA Schedule & Statistics (1974-75)
The Sporting News Hockey Guide (various years)
Official NHL Record Book 1917-64
Total Hockey; Total Sports Publishing (1st edition, 1998; 2nd edition, 2000)
The Encyclopedia of Hockey, by Robert A. Styer; A.S. Barnes (2nd edition, 1973)
The Hockey Encyclopedia, by Stan Fischler and Shirley Walton Fischler; Macmillan (1983)
The Trail of the Stanley Cup (Vol. 1, 2, and 3), by Charles L. ColemanPeriodicals
The Sporting NewsOnline
ESPN.com: http://www.espn.com/nhl/statistics
Hockey Research Association: http://www.hockeyresearch.com/stats
Hockey-Reference.com (Justin Kubatko): http://www.hockey-reference.com
Internet Hockey Database (Ralph Slate): http://www.hockeydb.com
Legends of Hockey.net (Hockey Hall of Fame): http://www.legendsofhockey.net/html/search.htm
LostHockey.com: http://www.losthockey.com
National Hockey League: http://www.nhl.com
The Sports Network: http://www.sportsnetwork.com
USA Today hockey stats archive: http://www.usatoday.com/sports/hockey/archive.htm*Please note I am working on making this interactive, but who’s got time for that these days, right?