Defining value has long been one of the most divisive topics in sports. At the core of the current debate is the 21st century explosion of accessible information and how it is used to measure player value. Each year, the divide is clear among the Professional Hockey Writers Association when it comes to the vote for the Hart Trophy, the NHL’s most valuable player award. The award description is only a dozen words (“to the player adjudged to be the most valuable to his team“), yet the annual interpretations and scrutiny showcase the vast range of views on the matter.

Ferocious, talented and driven, Bobby Clarke won three most valuable player trophies in four years as the inspirational leader of the Philadelphia Flyers. "Bobby Clarke 001" by rchdj10 is licensed under CC BY-ND 2.0

Some writers (and fans) believe points scored – regardless of teammates, playing time, defensive play, or goals vs. assists – are the primary value metric. Some feel the player’s team must make the playoffs, value being linked to securing a certain number of standings points. Some weigh the narrative as to how the player’s impact on their team separated him from the pack. Some are fused to an analytical view, largely removing the narrative. Others simply translate most valuable as the most “outstanding” performance on the ice that year.

While there is no single or correct answer when it comes to value, data-driven breakthroughs on the subject continue to emerge in the NHL landscape. Despite hockey’s speed and fluidity, the work being done to measure individual performance is evolving rapidly. (I’m partial to Dom Luszczyszyn’s Game Score Value Added because its underlying principles are sensible, the writing on its outputs are clear, and the related played cards are easy on the eyes). Isolating performance by contextualizing usage, competition, teammate quality, special teams, zone starts, score effects, and luck, among other factors, has revolutionized hockey analysis. This work should be applauded due to the depth and validity of the research. Data-based findings in hockey are not the only answer to player evaluation, but they have added quality takeaways to a sport glacially slow to accept outside views.

So, to measure careers historically, let’s just use one of these sophisticated models and work back to 1917-18? Unfortunately, we can’t…

The Challenge

Contemporary hockey analysis depends heavily on information that only starting being tracked for the 2007-08 season. We have no idea how many shot attempts Jean Beliveau took, how he performed with or without “Boom Boom” Geoffrion on the ice, or how many even-strength minutes he typically played. The work of many of hockey’s brilliant minds focuses on using recent past results to project future outcomes. Recognizing the limits of data prior to the availability of “fancy stats,” they have logically drawn an analytical line in the sands of NHL time in 2007.

As our mission in the Adjusted Hockey project is to compare every player, we have to develop an approach to measure value that can cover 100-plus NHL seasons. Of course, the further back you go, the less information available. So, even a systematic method will have caveats, the limits of league recordkeeping requiring the use of proxies in lieu of optimal data. Recognizing this, we will have to make assumptions, some of which will have omissions or won’t capture value to the finest possible degree. But if we can evaluate NHL careers in a comprehensive, systematic way using the best available information, we’ll better slice through the fog that clouds the evaluation of players across generations.

What aspects of a career do we need to consider?

Before we consider how we will measure value, in order to create a holistic system, we must consider multiple elements of a player’s career. Consider the following accomplishments and the respective standing of the player in hockey history:

Each player delivered a different form of excellence to their respective teams over the course of their careers. This simple exercise gives us three critical takeaways when it comes to value:

  1. A player can add value by contributing significant value totals over a career.
  2. A player can add value by contributing efficient value over a career.
  3. A player can add value by contributing at a high level at some point in a career.

This is our first breakthrough as to how to evaluate a player – the system needs to consider career value, pace value, and peak value. We’ll need to quantify and weigh each of these factors in a standardized manner. But we’ve established the three main components that will be the foundation of the system.

Career. Pace. Peak.

Ron Francis' sterling career is celebrated for the value he brought as the sum of its parts over 23 seasons. "Ron Francis 001" by rchdj10 is licensed under CC BY-ND 2.0

Before we pop any champagne bottles, many challenging questions immediately arise:

  • How do we capture these distinct types of contributions into a body of work that could last a few games, a few seasons, or span more than two decades?
  • How do we distribute value among forwards, defencemen, and goaltenders?
  • Is the performance of a player’s team relevant in determining his contributions?
  • How will both a player’s offensive and defensive contributions be reflected?
  • How do we account for varying schedule lengths and roster sizes?
  • Is it more valuable to be excellent for a short time, or useful for a long time?

A hockey career is not limited to the regular season, either. Players also perform on the more exclusive stages of the NHL post-season and international competitions. With that said:

  • How will we factor in performance on these platforms outside of the rhythm of regular season play?
  • Given there is only one Stanley Cup, how do we account for results in a 32-team league vs. the NHL’s first half-century featuring 10 teams or less?
  • How do we factor in international play given its wide-ranging formats, player availability, and national talent pools?

We have no shortage of questions to answer!

Fortunately, this is not the first time some of these challenges have been considered. While a hockey methodology this wide-ranging has not been attempted, there have been several groundbreaking projects in sports that have assigned value in a systematic way. Baseball, as with most innovative statistical analysis, led the way. Decades of original work by Bill James, among others, slowly brought the conversation mainstream. James’ efforts launched the concept of measuring the individual contributions of baseball players (a “player contribution” system), culminating in his Win Shares model outlined in 2001’s The New Bill James Historical Baseball Abstract. Eventually, various forms of Wins Above Replacement (WAR) became commonplace in sports.

Shortly after James, a trio of pioneering hockey analysts – Iain Fyffe (Point Allocation, 2002), Tom Awad (Goals Versus Threshold, 2003), and Alan Ryder (Player Contribution, 2003) – developed like-minded methodologies for hockey. While these three credit-assigning systems are not part of popular hockey lexicon, they paved the way for a generation of analysts to push their work forward.

Point Shares

While contemporary analysis helps us understand contemporary hockey, we do not have sufficient inputs to evaluate the NHL’s first 90 years under its lens. So, we need to simplify things. Justin Kubatko has done the groundwork for us. Leveraging a Bill James-style system he created to measure NBA Win Shares, Kubatko created Hockey Reference’s Point Shares system. This methodology will be the root of how we assign value to NHL players by season, its steps designed to work around historical information gaps. While we’ll need to tinker with and expand this system to fit our purposes, it is an ideal starting point for our goal of assigning value across the NHL’s 100-plus seasons.

In this section of the site, Point Shares, we’ll learn about how they work, examining their strengths, limitations, required adjustments, and how we’ll use the concept to create a comprehensive approach to evaluate players across eras.

Adjusted Point Pace data from Adjusted Hockey;
All other data from