Hockey’s Advanced Analytics: What Are They & How Do They Work?

Hockey analytics is the analysis of quantitative data and statistics regarding players and teams to understand and track their performance at a deeper level. It goes deeper than the basic stats that we see everyday like goals, assists, shots, and plus/minus. Some of the most commonly used analytic tools include Corsi, Fenwick, and PDO, which take into account shots, possession, and luck of the game, but we will go into that more later in the article.

Background Into Hockey and Sports Analytics

When it comes to sports, people are always trying to get the edge and win. Winning brings in more money and is better for the team and city. Any advantage a team can get, managers and owners have tried to put measures in place to succeed.

It comes down to the right analytics, though. Sometimes the eye test just won’t cut it. Sure, you can see a player’s work ethic and other factors they can bring to the game by watching, but it’s the more in-depth numbers that really tell us the impact mathematically each player has on the team while on the ice.

Hockey’s Advanced Analytics: What Are They & How Do They Work?
Hockey’s Advanced Analytics: What Are They & How Do They Work? (The Hockey Writers)

The game of hockey has shifted from caring solely about goals and assists to what a player does overall while on the ice. Analytics has found that you can predict a win probability better when tracking shots from a player, offensive zone possession, and opportunities taken away from the opposing team. Hockey today focuses on the well-rounded ability of players because you gain no advantage if a player allows as many chances and goals as they produce offensively.

Why Analytics Are Important and How They Work

Analytics are important because they can be used to see exactly what players and teams are contributing to the success or failure of their on-ice product. It can help us see players that stand out and are doing more good than bad for their club on a bad team, and the opposite, a bad player on a good team.

Analytics has become a huge factor in evaluating a player’s value when it comes to contract negotiations, and a reason why we’re seeing young players signed to larger deals at such a young age. Some analytics predict future success based on previously tracked data. They can tell you the contributions that go deeper than points when a player is on the ice

You May Also Like:

Not all predictive measures are 100 percent accurate because of the luck incorporated into the game. This analytics measurement is defined as PDO and tracks the “luck” involved in the sport of hockey.

What is the Future of Analytics and How Do They Impact the On-Ice Product?

All stats have value, and the future of analytics will have us looking at metrics as they relate to each other, not separate. “Goals can only come from shots, shots can only come from the possession of the puck, and possession is an inevitable result of winning faceoffs, getting power plays, forcing turnovers, and so on.”

Hockey is a fluid game, not a game like baseball that is a stop-and-start game. This makes it more difficult to track and figure out exactly how many wins an individual player contributed to his team each season. More than one single player contributes to the offensive success of the team each play, unlike baseball where the first batter of an inning or game could hit a home run. In hockey, it starts with the faceoff where even the positioning of all the players matters.

With tracking chips in player jerseys and pucks, it will make it much easier to analyze every detail of any player’s game and answer better questions that could contribute to success. Stats that are too difficult to track through an entire game by eye will be more accessible to pull data from and see results.

Some stats like takeaways and giveaways we just attribute to the basic stat. But what if we go deeper and track each time it directly leads to a goal. Instead of just successful zone entries and exits, does a player directly contribute to either happening? From the defenceman retrieving the puck in the corner, passing it to the other defenceman and on to the winger, and the winger chipping it into the middle for the centreman to pick up and exit the zone.

Gaining knowledge in the future about even more analytics could further change the thinking of coaches and general managers. We already see defensive specialists who are better at draws or on the penalty kill start more faceoffs in the defensive zone. But more than that, at 5-on-5, we could see more wingers who make fewer mistakes defensively and contribute to breakouts be out there in situations that would normally see them sitting on the bench.

Depth on teams, offensively and defensively, could be shifted even more away from “tough guys” or energy lines to be filled with players that the coach can use in whatever situation is needed in a tight game. Sort of like bench players in baseball who are either designated pinch hitters, pinch runners, or players put in the game to help hold a lead defensively.

Any team that can figure something like this out could be ahead of the curve and will be able to acquire superstars to do that work while filling their depth for cheap with players that can help the team in every other situation to win games. It wouldn’t be a flashy job that gets these players a lot of money, but some teams look for specialists if they feel they have all other angles covered. There should always be a place for them in lineups.

Different Metrics

Corsi (Shot Attempts: SAT)

Corsi is one of the basic metrics that tracks the sum of shots, missed shots, and blocked shots. It is used to find the attempted shot differential between teams and players while they are on the ice. If a player or team has a Corsi of over 50 percent, that means they were controlling the puck more often than the opponent. Darcy Reiger was originally attributed with the stat, but named it after Jim Corsi because the name had a better ring to it. Jim was a former goaltender and goaltending coach in the NHL.

Fenwick (Unblocked Shot Attempts: USAT)

Fenwick is close to Corsi in that it looks at possession by tracking shots and missed shots, but doesn’t take into account blocked shots. A Fenwick percentage over 50 also means your team controls the puck more when the player is on the ice. The higher a player or team has, the better they are at controlling the game. Fenwick got its name from a blogger in Alberta who made the argument that Corsi could be improved to exclude blocked shots because it shouldn’t be considered a scoring chance.

PDO or SPSV%

This is hockey’s version of “luck” that was created for baseball to gain a better understanding of each player’s individual contribution to the score and games. It is calculated by on-ice shooting percentage plus on-ice save percentage. It can look at a team’s numbers or a single player’s. SPSV% stands for shooting percentage plus save percentage, and the NHL tends to define this stat as such instead of PDO.

Calculated as a team metric, the number of every team combined should always equal 100 percent. Individually, whether a player has a huge impact on the outcome, players could be well above the mean or below. A player’s PDO is calculated mostly when they are on the ice at 5-on-5 as to not have the data jumbled by a player that may play a ton of shorthanded minutes or someone who plays on the power play.

Players’ PDO can be affected by their teammates’ performances while they are on the ice as well, such as shots that are going in on your goalie more often than normal or a player making a bad turnover resulting in a goal against when the other players on the ice could do nothing about that. The bounces may just not be going in the net on the offence and can be tracked by an above-average Corsi or Fenwick.

Zone Starts (ZS%)

As you may know, there are players who specialize more in their offensive ability, while others are relied on much more heavily in their own end. This can really affect a player’s other advanced analytics and has to be taken into account when looking at metrics like Corsi and Fenwick.

This only takes into account the offensive and defensive zone faceoffs, not the neutral zone. Though half of shifts come on the fly and don’t start with faceoffs, this isn’t as heavily used as it once was, though still can give you a good, deeper understanding as to why players’ stats are good or bad.

Regularized Adjusted Plus-Minus (RAPM)

The use of RAPM is to evaluate an individual player’s contribution to the team by isolating him from all other outside factors that could have an effect on the raw results. Some metrics that are used to evaluate offence are goals for per 60, expected goals (xG) for per 60, and Corsi for per 60. Defence can be evaluated using xG against per 60 and Corsi against per 60. The rate per 60 (minutes) means how much data each player accrues for every 60 minutes they are on the ice. This helps compare a first-line player to a fourth-line player who would play a different amount of time each game and allows us to analyze and compare their performances.

This is just a basic rundown of what RAPM actually is. If you want to dive deep into a study, head here.

Expected Goals (xG)

All shots shouldn’t be considered equal. Some shots could be on an empty net, while others could be from the point with no traffic in front of the goalie. This is where xG comes into play. It corrects the problems that Corsi and Fenwick have and assigns a value to each shot individually instead of valuing every shot as an equal opportunity to score. It tracks who is taking higher quality shots and predicts them as more likely to score goals.

Goals Saved Above Average (GSAA)

Derived from baseball’s stat that tracks wins above replacement (WAR), hockey uses GSAA to compare an individual goalie to the league average. It calculates the goals a goalie has saved given his save percentage and shots faced compared to the league average save percentage on the same amount of shots faced. A positive GSAA means that’s how many goals the goalie has saved compared to the league average.

This stat is also useful to look at how much the goaltender is relied upon to win games. Teams that have a goaltender with a high GSAA would be in trouble defensively if they were to get injured or start to play badly. This metric doesn’t take penalty-killing into account, so a goalie can be plagued by a team that takes a lot of penalties. Only analyzing this at even strength would allow goalies to be compared better in most situations.

High-Danger/Low-Danger Shots

Shots can be given a value depending on where in the offensive zone the shot was taken from and the probability of it going in. Each shot is assigned either the No. 1, 2, or 3. A low-danger shot has a value of one and is usually from the point or near the boards. A scoring chance is defined as any shot with a value of two, while a high-danger chance is valued at three. Rebounds can be defined as any attempt made within three seconds of shot, block, or save. Rebounds add one point to the proceeding shot.

Advanced analytics are very useful and are moving the game of hockey forward, especially in the NHL with the increase in technology within the game. Teams will continue to utilize metrics and try to come up with the best ways to improve the results of their team and players, sign and draft the most promising targets, and craft team-friendly contracts.


Morning Skate newsletter Click To Subscribe