In the past, advanced analytics that measure athletes’ performance have typically been associated with the sabremetrics movement in baseball. Recently, hockey analytics have gained traction through blogs and articles. And now, it’s not uncommon for NHL teams to employ an analytics professional or even an entire department dedicated to deciphering big data from games.
However, widely used analytics still have flaws. This article is not meant to condemn analytics for these flaws, but to make readers aware of what the statistics stand for. Then, readers can make an educated decision on whether or not to qualify these quantitative metrics.
The first advanced analytics statistic we’ll dive into is Corsi.
Corsi: What Is It?
Created by a blogger and named after former Buffalo Sabres and current St. Louis Blues goalie coach Jim Corsi, Corsi is a proxy statistic to determine the amount of possession during a game. It counts the total shots for and against while a player is on the ice. “Shots” includes goals, saves, missed shots, and blocked shots. If a player is on the ice for more shot attempts for than against, it can be inferred that his team has possession more often than not.
There are many ways Corsi can be quantified on a player-by-player basis, including:
- Total Corsi for/against
- Corsi percentage (Corsi for divided by total of Corsi for and against)
- Average Corsi for/against per 60 minutes of ice time
- Average Corsi for/against per 60 minutes while on the ice with a certain player
- Using average Corsi for/against to determine quality of on-ice opponents
These five metrics are the most widely used Corsi representations among hockey journalists/bloggers to quantify team possession while certain players are on the ice.
Corsi: The Flaws
While there is some degree of accuracy with Corsi, there are still flaws with the metric that should be taken into account.
1. Equal Weighting
When counting shot attempts, goals, saves, blocked shots, and missed shots are all weighted equally whether for or against. If you talk to any NHL player, they’ll goals are not equal to other shot attempts, even when attempting to quantify possession. Consider this example of two defensemen who are on the ice for an equal amount of time and shots for:
- Player A: 12 shots against (four goals, five saves, one blocked shot, and two missed shots)
- Player B: 12 shots against (one goal, five saves, four blocked shots, and two missed shots)
If all else is equal, then the Corsi score (whichever measure being quantified) will be equal as well, despite the goal differential. Should these two players be considered equal given this information?
2. In-Game Situations
Suppose a goal is scored against your team. The five players on the ice would be given a negative Corsi score for the shot attempt(s). But what if the goal can be pinned on one player? Or the goalie? Or multiple players’ lack of backchecking?
These in-game breakdowns are not accounted for in the cumulative Corsi stats that track the average play of certain players. It is designed to cover a large enough timeframe that outliers are excluded, but game-to-game Corsi scores are skewed nonetheless.
Alternatively, if a player hops onto the ice during an on-the-fly change and a shot attempt is registered in either direction, should that player receive the Corsi for/against score? Calculating a player’s plus/minus rating encounters the same issue.
3. Team-To-Team Comparison
When comparing players on different teams, Corsi metrics are often used to determine which player is better. It is also often used when evaluating a newly signed free agent. One variable the different Corsi statistics do not account for is the differentiation in team-to-team strategy.
Teams like the Detroit Red Wings, Chicago Blackhawks, Pittsburgh Penguins, and Dallas Stars each have their own unique identity. Basing a player’s value on their Corsi scores does not take into account these differing strategies. For example, well-traveled defensive defenseman Rob Scuderi played for the Penguins, Blackhawks, and Kings this season—all of which with different playing styles. Consider his Corsi percentages with each team:
- Rob Scuderi (Pittsburgh): 25 games – 46.72% shot attempts for
- Rob Scuderi (Chicago): 17 games – 44.98%
- Rob Scuderi (Los Angeles): 21 games – 50.18%
A player’s performance and possession rating on one team does not always translate evenly to the next. Teams’ systems work best for certain players and that’s not something that can be directly inferred from Corsi scores.
4. Favoring the Offensive Side
Since Corsi ratings are designed to measure possession, they do not always lend useful information about a player’s defensive performance. Apart from a shot attempt (goal) equaling a shot attempt (missed shot from a bad angle), offensive defensemen are favored over defensive defensemen. For example, Dustin Byfuglien, Jakub Kindl, and Mike Green are not the most reliable defensemen in the defensive zone, but can drive shot attempts when on offense. Alternatively, players like Marc Staal, Kris Russell, and Mark Methot—who can be counted on to prevent scoring in the defensive zone—usually come away with poor Corsi ratings.
Hockey is a two-way game—both teams possess the puck for a significant amount of time during a game. Players with defensive roles cannot be fairly compared with those who get more ice time in the defensive zone. For example, Marc-Edouard Vlasic had a 52.77% Corsi shot attempts for rating as San Jose’s primary shutdown defenseman last season. Should defensemen with higher Corsi ratings like Brendan Smith, John Klingberg, and Hampus Lindholm be considered better than Vlasic since they drive possession better?
While Corsi has its benefits, there are still flaws with the popular advanced analytics metric. Fans should consider these when reading up on players. This is not to say that analytics should be not considered when determining team management decisions. Analytics definitely has its place in hockey. Perhaps today’s current measurements, like Corsi scores, have not been perfected yet and Moneypuck will dominate roster decision-making processes when more technology is introduced to the game.
Do you think the different Corsi ratings are reliable metrics for evaluating players? Comment below with your take.
Corsi statistics were provided by NHL.com.