With each passing season more and more teams, analysts and fans are enriching their understanding of hockey with statistics that go beyond the boxcar stats of goals, assists, plus/minus, goals-against average and any other stat you could find on the back of a hockey card. There are a lot of names for this growing field — which started in baseball and spread like wildfire to other sports — but for the purposes of this article, we’re going to call them analytics.
To many who are new to analytics, the labyrinth of formulas, concepts, definitions and viz charts represent deep water to someone who never got past using arm floaties (or water wings, depending on where you’re from) in the pool. So, Sporting News is bringing you to the shallow end by introducing a few of the key concepts in hockey analytics.
The Shot Share and Possession
Wayne Gretzky’s quote about missing 100 percent of the shots that you don’t take — which was later immortalized by Michael Scott — is helpful to keep in mind for those who are new to the concepts of Corsi and Fenwick and to those who want to understand why they are important.
As we all know, the objective of hockey is to score more goals than your opponent. The best way to score goals (and to prevent them) is to have possession of the puck and to fire it towards the opposition’s net. That is why puck possession and shot attempts are foundational ideas in the analytics community. Corsi is also considered to be a predictive measure, and one of the reasons it outperforms stats like goals is that it deals with larger sample sizes.
The two main players in this corner of analytics are:
Corsi — All shots on goal (including goals), all shots that miss the net and all the shots that are blocked.
Fenwick — All shots on goal and all shots that miss the net, excluding shots that are blocked.
These, along with the other concepts we are looking at today, deal almost exclusively with 5-on-5 hockey. The reasoning behind this is that more hockey is played at 5-on-5. It’s important to note that context is everything for these stats; for example, it should come as no surprise that a player or team has poor possession numbers when they’re killing a penalty or inflated numbers when they’re on the power play.
Identifying the situations in which stats are used is essential to appreciating what insights they can give us to the play of an individual, a group of players or a team. While it’s true that stats like GAR (goals above replacement) and Game Score exist, there is no true, all-encompassing, one-stat-tells-all in hockey. Context is everything.
Individual, On-Ice and Team Stats
Corsi and Fenwick can be applied to an individual in two ways. The concepts can be used to measure a player’s iCF or iFF — their individual Corsi For and individual Fenwick For — in addition to playing a role as an ‘on-ice’ measure. By doing this, you can gain a better understanding of what happens when a player is on the ice compared to their peers and competitors. We’ll get into this more when we look at relative stats.
If a player piles up an “8 iCF” at 5-on-5 in a game, that means they attempted eight shots during 5-on-5 play that either went on net, were blocked, missed the net or resulted in a goal. When we look at what happens when a player is on the ice, we look at the ratio between two events, i.e. Goals For versus Goals Against (GF%), Corsi For versus Corsi Against (CF%) or rate stats (we’ll get there too).
We can also use possession numbers and other stats to measure the performance of a group of players, i.e. a forward line or a defensive pair. The same is true of teams. When we look at large enough sample sizes, we can discern which teams have more success in the shot share much like we can compare and contrast power-play success rates.
Expected goal models assign a value to shots based on their location and other factors such as whether the shot was a rebound, one-timer, etc. The concept of expected goals in hockey is based on the concept that some shots are more valuable than others based on how likely they are to result in goals.
Anyone who has played the game, even in a casual setting, or has watched it intently can attest to the fact that not all shots are created equal. A shot taken from the slot or right on the doorstep is far more likely to go in than a shot taken from a bad angle at the half-boards or a shot taken outside the offensive zone.
Just like Corsi and Fenwick (and other stats), we can measure the expected goals (xG) for a player, line, pairing or team, whether it’s for (xGF) or against (xGA). This makes xGF% a valuable tool in discerning the quality of shots that occur when a player is on the ice and helps us gain a better understanding of how sound their play is with and without the puck.
Concepts and stats related to expected goals for and against include: Scoring Chances For/Against (SCF and SCA) and High-Danger Corsi For/Against (HDCF and HDCA).
NHL rumors 2019: San Jose Sharks, Winnipeg Jets contenders for early-season trades
A scoring chance is any shot determined to have a significant probability of going in, depending on the model being used. Similarly, a High-Danger Corsi For or Against is a shot attempt that, in terms of its assigned value, has an even higher likelihood of going in than a scoring chance. As you might expect, you want to generate scoring chances and High-Danger Corsi For attempts while also limiting them for your opponent.
The graphic above represents probability areas broken into danger zones, as first defined and presented by War-on-Ice. As you can see, shots taken from the slot and close to the net are more valuable because they are more likely to go in.
Relative and Rate Statistics
Relative statistics compare a player’s performance to their teammates; in other words, it provides context to stats like CF% and xGF%. This is a particularly valuable tool because it enables us to discern which players are having a positive impact on the shot share on bad teams and to identify players who have underlying numbers that are heavily influenced by the strength of their team.
For example, the Edmonton Oilers had a 50.11 SCF% when Connor McDavid was on the ice last season. Of course, a 50.0 SCF% would mean a player is breaking even in scoring chances when they are on the ice — but that doesn’t mean McDavid is average at generating scoring chances. When we look at the relative stats, McDavid led the Oilers with a +3.55 Rel SCF% — the best on the team by a significant margin.
Rate statistics help provide greater context for raw statistics by scaling them per 60 minutes of play. This helps level the playing field when we compare players who have played a different number of games or have different average ice times. With rate statistics, we can compare/contrast stats like Corsi or expected goals, both for and against, between players, lines, pairs, and teams. Like so many of the other stats we’ve put a spotlight on, rate statistics can help provide context to the numbers we’re working with.