Statistics are metrics used for many different things in hockey but mostly they are used to generate a player's value. Often times these values are then compared to other players to see who is more valuable. However statistics in hockey are flawed. They don't give us the whole picture and it makes it hard to quantify their abilities on paper. Click the jump and learn why.
Hockey is a game that is largely comprised of randomness and chaos. In a sport like football, every single step a player takes is mapped out and every player has specific things they are supposed to do on every snap, whether it is to run a certain route, draw defender's to themselves to isolate a different player, certain blocks to make or a hundred other options, hockey is much more chaotic. Each player plays by their team's scheme and they play within a system, but their moves and choices aren't mapped out, they are liquid and fluctuate depending on how the play develops.
In baseball stats are extremely helpful because everything that happens is clear and quantifiable. Baseball may be random but since every pitch is planned and there are stoppages in between each play everything that happens can be traced back to one event, whether it is a stolen base, a double, or a strikeout. This kind of rigid structure affords the opportunity to quantify almost everything that happens in a baseball game effectively. Baseball is simple and breaking it down is clear cut.
Basketball's stats are quite useful due to the amount of quantifiable events that happen. An average hockey game will have anywhere from 3-9 goals in it and most metrics rely on those 3-9 events. A basketball game however will have maybe 100 made baskets which can all be quantified and translated into metrics. In basketball a player's value is directly related to how many points they can create in comparison to how many they prevent. Since there are so many points scored this plus minus stat of sorts is a rather reliable way of judging a player's value to their team.
The plus/minus stat in hockey does an extremely poor job at projecting how good or bad a player is and gives us little usable information. The lack of scoring means that goals are rather random. Plus/minus doesn't distinguish between the quality of the goal or who is responsible for the goal. Imagine this scenario, a defenseman makes a great stretch outlet pass to spring his winger on a breakaway. It's the end of his shift so he goes off for a change before the player on the breakaway shoots. The defenseman gets off the ice and a new defenseman gets on the ice. The winger scores on his breakaway, and the way the plus/minus system works the original defenseman who made the stretch pass will get an assist but not a plus. The new defenseman however, having nothing to do with the play, will get a plus. The randomness and overall rarity of scoring in the NHL means that not every player in contracted to score goals. These are the grinder types, and defensive defenseman types. They are successful if they stop goals, any goal production is a complete bonus. It is not uncommon for the grinder types to go up against the other teams top line. They just don't have the skill or ability to generate offense against them. These players are going to have lower plus minus ratings due to their roles on the team of stopping goals instead of scoring them.
Now is the part of the conversation where we slide into why stats help us. Is a player who scores 30 goals a year more useful than 2 players who score 15. Putting it another way, is Bobby Ryan more valuable than Craig Smith and Jannik Hansen put together in the 2011-2012 season. Using stats, this question is pretty hard. Bobby Ryan is more dynamic than Hansen and Smith. Ryan's goals are more predictable because he can create goals with his strong shot and his finishing ability. Hansen and Smith are both 15 goal scorers but their goals are easily replaceable. A player's value stems from how hard it would be for that team to replace them.
Making this Wing centric again, it would be much easier for the Wings to replace Drew Miller and Joakim Andersson(with players like Patrick Eaves or Riley Sheahan) than it would be for them to replace a player like Franzen. Franzen, when hot, is a player who creates his own chances and commands attention and skilled defenders as Miller and Andersson's goals are much more random and can be attributed to chaotic garbage goals, which are less reliable when extrapolating stats on a larger timeline.
Some of the more advanced hockey statistics revolve around possession. This makes a lot of sense of course, if you have the puck you can score and your opponent cannot. Stats like Fenwick and Corsi are much better at predicting performance. These stats try to quantify things that lead to goal like events instead of traditional stats which generally measure goals. Goal like events are much more projectable over a long period of time and are therefore more reliable than goals when trying to establish the value of a player.
However, these stats are rarely used and are generally not understood by most hockey fans. These are mostly niche stats. Most fans go off of traditional metrics because they are more accepted and are concrete as opposed to advanced metrics which are more flawed and pretty complicated. Traditional hockey statistics are useful and can help you paint a picture and evaluate player performance, but stats are only as good as what they are used for. Stats can tell any story and the story they tell is entirely dependent upon the person using them.
In baseball stats are extremely helpful because everything that happens is clear and quantifiable. Baseball may be random but since every pitch is planned and there are stoppages in between each play everything that happens can be traced back to one event, whether it is a stolen base, a double, or a strikeout. This kind of rigid structure affords the opportunity to quantify almost everything that happens in a baseball game effectively. Baseball is simple and breaking it down is clear cut.
Basketball's stats are quite useful due to the amount of quantifiable events that happen. An average hockey game will have anywhere from 3-9 goals in it and most metrics rely on those 3-9 events. A basketball game however will have maybe 100 made baskets which can all be quantified and translated into metrics. In basketball a player's value is directly related to how many points they can create in comparison to how many they prevent. Since there are so many points scored this plus minus stat of sorts is a rather reliable way of judging a player's value to their team.
The plus/minus stat in hockey does an extremely poor job at projecting how good or bad a player is and gives us little usable information. The lack of scoring means that goals are rather random. Plus/minus doesn't distinguish between the quality of the goal or who is responsible for the goal. Imagine this scenario, a defenseman makes a great stretch outlet pass to spring his winger on a breakaway. It's the end of his shift so he goes off for a change before the player on the breakaway shoots. The defenseman gets off the ice and a new defenseman gets on the ice. The winger scores on his breakaway, and the way the plus/minus system works the original defenseman who made the stretch pass will get an assist but not a plus. The new defenseman however, having nothing to do with the play, will get a plus. The randomness and overall rarity of scoring in the NHL means that not every player in contracted to score goals. These are the grinder types, and defensive defenseman types. They are successful if they stop goals, any goal production is a complete bonus. It is not uncommon for the grinder types to go up against the other teams top line. They just don't have the skill or ability to generate offense against them. These players are going to have lower plus minus ratings due to their roles on the team of stopping goals instead of scoring them.
Now is the part of the conversation where we slide into why stats help us. Is a player who scores 30 goals a year more useful than 2 players who score 15. Putting it another way, is Bobby Ryan more valuable than Craig Smith and Jannik Hansen put together in the 2011-2012 season. Using stats, this question is pretty hard. Bobby Ryan is more dynamic than Hansen and Smith. Ryan's goals are more predictable because he can create goals with his strong shot and his finishing ability. Hansen and Smith are both 15 goal scorers but their goals are easily replaceable. A player's value stems from how hard it would be for that team to replace them.
Making this Wing centric again, it would be much easier for the Wings to replace Drew Miller and Joakim Andersson(with players like Patrick Eaves or Riley Sheahan) than it would be for them to replace a player like Franzen. Franzen, when hot, is a player who creates his own chances and commands attention and skilled defenders as Miller and Andersson's goals are much more random and can be attributed to chaotic garbage goals, which are less reliable when extrapolating stats on a larger timeline.
Some of the more advanced hockey statistics revolve around possession. This makes a lot of sense of course, if you have the puck you can score and your opponent cannot. Stats like Fenwick and Corsi are much better at predicting performance. These stats try to quantify things that lead to goal like events instead of traditional stats which generally measure goals. Goal like events are much more projectable over a long period of time and are therefore more reliable than goals when trying to establish the value of a player.
However, these stats are rarely used and are generally not understood by most hockey fans. These are mostly niche stats. Most fans go off of traditional metrics because they are more accepted and are concrete as opposed to advanced metrics which are more flawed and pretty complicated. Traditional hockey statistics are useful and can help you paint a picture and evaluate player performance, but stats are only as good as what they are used for. Stats can tell any story and the story they tell is entirely dependent upon the person using them.