View from back of hocket net of blue hockey team scoring goal on red hockey team

As the NHL regular season comes to an end and fans across the country gear up for another year of playoff pools, bracket tinkering, and inevitable Maple Leafs heartbreak, hockey is top of mind. If you opted to skip out on the NHL this season you missed one of the tighter MVP races in recent memory – Auston Matthews gunning for 70 goals, reigning NHL MVP Connor McDavid looking to become just the fourth player in history to tally 100 assists in a season, all while MVP front runner Nathan MacKinnon casually puts up 135+ points.

Goals, assists, and points are the typical stat vernacular when it comes to talkin’ puck, but more recently new terms have come to be the standard when discussing player performance. Terms like Corsi and Fenwick, or xG and GAR are becoming commonplace. Where did these terms come from? How did analytics find its way into a 100+ year old game? Who is using them? Do they work? As you prepare for small talk about that game last night, let’s dive into the intersection of analytics and hockey.

Pre-game warmup: Understanding the context of hockey analytics

Hockey analytics, at its core, is the systematic study of data to gain insights into the various facets of the game. From player performance metrics to team strategies and game dynamics, these analytics place a magnifying glass onto every aspect of hockey with precision and depth. Goals, assists, and points have always served as traditional metrics for assessing player effectiveness; however, as hockey has progressed, so too have the methods for evaluating players, extending beyond simple counting statistics. Nevertheless, it's imperative to recognize that these metrics must be contextualized and not relied upon in isolation to fully comprehend a player's impact.

Analytics in sports are a tool, not a rule - just ask the Toronto Blue Jays. Noone in the NHL shares this sentiment more than perennial “did he just say that?” quote factory and head coach of the Philadelphia Flyers, John Tortorella. When asked about analytics in 2022, he got right to the point, “Don’t ever say that analytics is the end goal for me. Because it’s not… I think … that’s another subject for another day. I have certain analytics that I think are good, and I think most of them are trash. Because I believe in the eye test and stuff.” [via: The Atheltic] Whether it’s called the eye test, gut feeling or just an inkling, hockey will always be more than just the numbers, no matter how advanced they become.

First period: The journey of hockey analytics and the debut of Corsi

To grasp the present, we must first trace the footsteps of the past. The roots of hockey analytics stretch back to the early days of the sport, where simple stats such as goals and assists provided a rudimentary framework for evaluating player and team performance. However, it was not until the turn of the millennium that hockey analytics truly began to take hold, propelled by advancements in technology and a burgeoning community of data enthusiasts.

One pivotal moment in the history of hockey analytics came with the introduction of the Corsi statistic, named after goalie coach Jim Corsi. Corsi quickly gained traction within the analytics community as a potent predictor of team success. A Corsi number is derived by dividing a team's total shot attempts at even strength by the opposing team's total shot attempts. Shot attempts indicate any and all shots directed toward the net, including shots on goal, missed shots, and blocked shots.

On an individual level, a player's Corsi number is found by dividing their team's shot attempts while they’re on the ice by the opponent's shot attempts during the same duration. Essentially, it operates similarly to plus-minus, but instead of goals, it tracks shot attempts.

Put simply, a higher Corsi number indicates a more favourable performance. For instance, during the 2013-14 season, the Stanley Cup Champion Los Angeles Kings boasted a 931 Corsi number (3,888 shot attempts for vs. 2,957 allowed), whereas the Toronto Maple Leafs, who finished sixth in the Atlantic Division that season, had a negative 1064 Corsi (3,259 shot attempts for vs. 4,323 allowed).

Other popular advanced stats include:

Fenwick: Similar to Corsi without counting blocked shots in its calculations. Developed by hockey blogger Matt Fenwick.

Expected Goals (xG): A model-based metric that takes into account over 100 data points to analyze the overall quality of a shot attempt and then predict the probability of a goal. Shot angle, distance, and speed are among the data points taken into consideration to provide a final expected goals (xG) number, which can be used to evaluate if a player is exceeding or failing to meet expectations.

Expected goals (xG) stats for 2021-2022 NHL season Credit:

Goals Above Replacement (GAR): A stat modeled after the popular baseball Wins Above Replacement metric, GAR is an attempt to assign a total value to each player and represent how much that player contributed to their team’s success (or lack thereof). By looking at Even Strength Offense, Even Strength Defense, Power Play Offense, Short Handed Defense, Penalties Taken, and Penalties Drawn among other data, GAR answers how much better (or worse) does this player perform versus a replacement player, a hypothetical baseline player made up of looking at averages in those same stat categories across the league.

Second period: Team strategies and analyzing success

In addition to individual performance, analytics are also used to analyze lineup combinations, offensive and defensive systems, and situational tactics such as power plays and penalty kills, helping teams optimize their game plans for success.

One notable example of successful analytics implementation in hockey can be found in the rise of the Tampa Bay Lightning as a dominant force in the NHL. The Lightning's journey to success is intricately intertwined with their adept use of analytics to inform player acquisitions, lineup decisions, and strategic adjustments.

Despite finishing the 2018-19 season with a record-setting 128 points, the Lightning suffered a stunning first-round playoff exit at the hands of the Columbus Blue Jackets. In the wake of this disappointment, the Lightning's management team, led by General Manager Julien BriseBois, turned to analytics and the team’s Director of Hockey Analytics, Michael Peterson, to identify areas for improvement and optimize their roster for future success.

“I would have a hard time finding an area where (Peterson) or his department are not providing us with some sort of information,” said BriseBois. “Most of the time we’re looking for information, we add it to other information we have from other sources and help paint the picture. They’re not in opposition, it’s more complementary to help us make better decisions. It might challenge our assumptions or it might help us dig deeper to validate assumptions.” (via The Atheltic)

By leveraging advanced analytics, the Lightning identified undervalued players who excelled in key areas such as puck possession, scoring efficiency, and defensive impact. One notable acquisition was Forward Blake Coleman, whose strong underlying metrics made him an ideal fit for the Lightning's style of play. Coleman's addition provided a significant boost to the Lightning's forward depth and bolstered their lineup for the playoff push.

In the 2019-2020 season, the Lightning's analytics-driven strategy paid dividends as they captured the Stanley Cup. From player acquisitions to in-game decision making, analytics played a central role in the Lightning's championship run, solidifying their status as a model of success in the modern NHL.

Interested in working for an NHL team? Check out the NHL Team Jobs Board for opportunities from marketing to sales to development and data.

Third period: The influence of analytics on NHL operations

While hockey analytics have reshaped on-ice performance and player personnel decisions, their influence extends far beyond what happens between the whistles. NHL teams leverage data-driven insights to enhance various aspects of their operations, including fan engagement, ticket sales, and concessions. Here are just a few examples of how analytics are transforming the business side of hockey:

Fan engagement: By analyzing fan demographics, preferences, and behaviour, teams can tailor marketing campaigns, promotions, and content to resonate with their audience, fostering a deeper connection and loyalty among fans. For example, Gravy Analytics offers a powerful platform for monitoring fan engagement using a combination of data analytics and location intelligence. Working with the Anaheim Ducks, Gravy Analytics was able to find trends in where fans were stopping before attending a home game. They found over an entire season, fans were likely to visit Starbucks, Rite Aid and Costco more often than the average consumer. Armed with this new information, Anaheim could now approach these brands to propose potential sponsorships and capitalize on the existing synergy between them.

Fanvue Game Day Analytics by Gravy Analytics Example Gravy Analytics dashboard of game day analytics for the Anaheim Ducks. Via:

NHL broadcasts: In addition to transforming the way teams play and operate, hockey analytics are also revolutionizing the way fans experience the game away from the rink. Announced in March of 2022, the NHL partnered with Amazon Web Services (AWS) to develop new tools to enhance the viewing experience for fans under the name NHL Edge IQ. This includes an in-game on-screen stat overlay predicting the probability of a player winning a face-off, dubbed Faceoff Probability.

Faceoff Probability during a Toronto Maple Leafs NHL hockey game broadcast An example of Faceoff Probability during a broadcast. Credit:

Also offered as part of NHL Edge IQ is Opportunity Analysis, a machine learning based analytics platform that provides insights into the quality of a scoring opportunity–think expected goals (xG) but in real-time. Check out this walkthrough by clicking the image below:

AWS and NHL Unveil Opportunity Analysis | Amazon Web Services

By displaying these probabilities and analytics on screen during live broadcasts, fans gain a deeper understanding of the ebb and flow of the game, as well as the strategic decisions made by coaches and players. By highlighting key faceoff matchups and shot trends, these statistics offer valuable insights into the tactical battles that unfold throughout the game, adding a new layer of depth and analysis to the viewing experience.

Overtime: The future of hockey analytics

With advancements in technology, data science, and artificial intelligence, the possibilities for innovation and discovery in hockey are boundless. From player tracking and biometric sensors to machine learning and predictive modeling, the frontier of hockey analytics has endless opportunities for exploration and advancement.

In the years to come, we can expect to witness even deeper integration of analytics into every facet of the game, from player development and scouting to in-game decision making and fan engagement. As teams continue to harness the power of data to gain a competitive edge, the lines between science and sport will blur, ushering in a new era of excellence in all sports.

So, whether you're a die-hard fan or a casual observer, take note the next time you’re watching a game of all the implementations of analytics, and the next time you’re asked about the game last night you’ll revel in the chance to talk about Connor McDavid’s Goals Above Replacement trend over the last five seasons. [spoiler: you don’t need analytics to tell you he’s very good at hockey, but it sure makes you sound smart.]

Ready to take the ice as a Data Analyst or Data Scientist? Check out Lighthouse Labs’ data programs here.