As the hockey world undergoes the data revolution that has arisen from the advancements in technology in our modern society, we find ourselves with access to new information seemingly every day. How do we keep up with all of the new terms and ideas? Well, the goal of this article is to define and describe some of the most common advanced statistics in my articles, so if there is ever a point of confusion, you can feel free to check out this guide to advanced stats!

The most common metric you will see is Goals Above Replacement (GAR) or Wins Above Replacement (WAR). These metrics are proportional and they measure the overall value of a player in goals or wins, respectively. While GAR is a good measure of overall value, it can be divided into two subcategories to paint a better picture of exactly where each skater provides value: Offensive Goals Above Replacement (Off) and Defensive Goals Above Replacement (Def). Each of those metrics indicates the player’s value on either side of the ice. But we can go even further. Off can be split up into Even-Strength Offense Goals Above Replacement (EVO) and Powerplay Offense Goals Above Replacement (PPO), and Def can be split up into Even-Strength Defense Goals Above Replacement (EVD) and Shorthanded Defense Goals Above Replacement (SHD). These four statistics give us a precise measurement of value in a subsection of the game. All of these metrics can be set on a per-60-minute basis to become a rate statistic instead of a counting statistic.

And better yet, you can add the word “Expected” in front of all of these statistics, and they have a slightly different meaning. When a statistic is an expected outcome statistic, it does not factor in what actually happens on a given shot; it merely factors in the probability of converting the shot. So shooting percentage is the primary difference between expected outcome statistics and observed statistics. So Expected Goals Above Replacement (xGAR), Expected Wins Above Replacement (xWAR), Expected Offensive Goals Above Replacement (xOff), Expected Defensive Goals Above Replacement (xDef), Expected Even-Strength Offense Goals Above Replacement (xEVO), Expected Powerplay Offense Goals Above Replacement (xPPO), Expected Even-Strength Defense Goals Above Replacement (xEVD), and Expected Shorthanded Defense Goals (xSHD) all exist, and they are the expected counterparts of the previously mentioned observed statistics.

I tend to use observed metrics (GAR, Off, Def, etc…) for forwards, and I tend to use expected metrics (xGAR, xOff, xDef, etc…) for defensemen because forwards tend to have more sustainable shooting percentages than defensemen, so the observed metrics are more reliable for forwards than defensemen.

The next most common set of metrics you will see pertains to Regularized Adjusted Plus-Minus, more commonly known as RAPM. And the easy way to remember these is by remembering the three pairs of statistics. Goals For (GF) and Goals Against (GA) portray the offensive and defensive value of a skater as outcome statistics, while Expected Goals For (xGF) and Expected Goals Against (xGA) portray the offensive and defensive value of a skater as expected statistics. Similarly, the main difference between these values is the outcome of a shot is accounted for in GF and GA, while the probability of each outcome of a shot is accounted for in xGF and xGA. Corsi For (CF) and Corsi Against (CA) are measures of the shot attempts for and against, respectively, while the skater is on the ice. Each of these metrics can be put on a per-60-minute basis to become a rate statistic

In terms of transition data, the three statistics you will see are Zone Entries, the number of times a skater enters the offensive zone, Zone Exits, the number of times a skater exits the defensive zone, and Carry-in%, the proportion of carry-ins to dump-ins for a skater.

Diving into shooting talent using MoneyPuck’s expected goals model, we can find Expected Goals (xGoals), which is the expected value of goals for each skater given the probability of each shot and the frequency of shots. This can also be expressed as a rate statistic on a per-60-minute basis. Shooting Talent Above Average is a more precise measure of the shooting ability of a player, and Shooting Talent Adjusted xGoals measures the expected value of goals, except unlike in xGoals, the probabilities are based on the shooting abilities of the skater. Finally, Goals Above Shooting Talent is a measure of randomness in the goal-scoring department.

Let’s not forget about goaltenders! You will see me reference only one statistic that is unique to goaltenders, and that is Goals Saved Above Expected (GSAx), a measure of the goals a goaltender saved minus the expected value of the goals he should have saved based on shot probability and frequency.

So there you have it: A guide to all things advanced stats! Hopefully, this helped to clarify any confusion about the metrics I reference in my articles. I look forward to expanding my understanding of advanced statistics in hockey, and no matter what I find, every statistic will be thoroughly defined in this guide.

All statistics are from Evolving-Hockey, Corey Sznajder’s transition data, or MoneyPuck.com.

Aidan Resnick

Aidan is a freshman at the University of Chicago, studying data science and business economics, and an aspiring sports analyst. In 2019, he attended the Wharton Moneyball Academy, the Carnegie Mellon Sports Analytics Conference, and the MIT Sloan Sports Analytics Conference, experiences that inspired him to pursue sports analytics. Aidan’s passion for sports analytics is best represented in his newest sports analytics book, “The Stats Game,” where he illuminates statistical tools and debunks myths in sports analytics, as well as in his victorious Diamond Dollars Case Competition project and in Resnick Player Profiles. A lifelong New York Islanders fan, Aidan always approaches his work with Drive4Five with an analytical mindset, focusing on the newest advancements in hockey analytics to maximize the precision of his content. Aside from sports analytics, Aidan is a dedicated violinist and chess player.