clock menu more-arrow no yes mobile

Filed under:

Narratives, Stats, and Bias

Narratives tell stories, analytics tell the same stories differently. Combining them helps weed out our biases and become better consumers.

If you buy something from an SB Nation link, Vox Media may earn a commission. See our ethics statement.

Ed Szczepanski-USA TODAY Sports

Narratives are a part of sport. They will always be because we are dealing with humans and humans are stories. Luck is a part of sport. It has always been and it will always be. Stats have been a part of sports for a long time. Jeff Marek told a story about how the NHL's first game sheet is from the 1920s barely had any assists. The NHL did not even track +/- until 1968. Statistics evolve and yet the narratives stay the same: "good in the room", "plays the right way" so on and so fourth. The narratives can be coloured though and that may be a bigger issue than we think.

We are all consumers when it comes to hockey. We consume the game and we consume the media, both mainstream and blog, that surround the game. We are at the mercy of other people to relay us the quotes from the players with the proper context and without bias. That is the ideal, but we are all human and we all harbour biases; both recognized and unrecognized. Arpon Basu, the managing editor of, touched on how bias can affect media members when Josh Gorges was traded. He tweeted about how accessible Gorges was for the media and how that led to them being easier to him. They were easier on him if he had a bad night; they were quicker to excuse poor play. They simply were biased because Gorges made their jobs easier. That is human nature though, being kind to someone that makes your life easier, but we must be cognizant of the bias that is formed when we think like pay them back with a lack of criticism.

Separating ourselves from bias is hard. I really enjoy watching Tobias Enstrom play. I have a hard time criticizing him because of this. I am aware of this, so I simply remind myself that he can and will make mistakes like everyone else. Besides being fully aware of our personal biases is one thing, finding a way to counteract those biases is another and that is where analytics can help us.

Our eyes are harder to check for biases because checking ourselves for bias takes being fully aware and alert of the bias when watching. You may find yourself looking for Enstrom losing battles or Dustin Byfuglien giveaways. You are looking for them and you will see them. It doesn't matter that they are both very good to great at their position because you see their flaws and those flaws become major issues, even if they are not. There is a way to check these biases and test your eyes: analytics.

Some people don't like analytics and that's okay (you may want to stop reading now because I am going to talk about their merits). The thing with analytics is they are best used when they can show the why or the how of something. Tyler Dellow, now of the Edmonton Oilers, was great at this and he started running a series of articles on the Oilers corsi issues and how their tactics were killing them. He was able to take a set of numbers and dig into the video to see why the numbers were so bad. That is elite corsi analysis. But just because someone doesn't have elite analysis levels doesn't mean that corsi cannot tell us a lot.

With video, the lies our eyes tell us can be detected. Bias is eliminated because an unbiased source (computer code), has been able to quantify what we watched and simplify it to a few numbers. This is not perfect, nor is it a perfect description, but the idea that large swaths of the game (all even strength minutes) can tell us a story simply by combining all shot attempts including blocked shots (fenwick has no blocked shots) to see who had more shot attempts and therefore had the puck more makes sense.

There is the argument that analytics do not account for bad games that are a one off because analytics people are constantly citing analytics is wrong. Most analytics people want a bigger sample versus a smaller one. That is the reason why corsi is used by analytics writers more than fenwick; it encompasses more shots, therefore it has a larger sample size.

+/- is a stat that is pure luck. If a player cannot buy a goal and their goalie cannot save a shot, their +/- is going to suffer. Never mind that player could be possessing the puck 60% of the time, but the goalie is a sieve when the player was on the ice and that player suffered for something they did not actually have anything to do with. It happens a lot.

Narrative shaping opinion is something that happens a lot. Writers can use different forms of evidence to help shape both their opinion, as well as the opinion of the reader. If the writer has an opinion formed with biases, they will allow the reader to share the same biases. When someone says "Toby Enstrom cannot play in the West because he is too small," the answer very well may be "prove it." Enstrom can play in the West as a small defender because he does a lot of little things well and him doing those things well are reflected in his analytics numbers.

Writers can even weave a narrative using analytics versus traditional stats. Bruce Arthur and James Mirtle are two of the best at this, as they see the logic in understanding how numbers are not the whole story but a key part of the story. As writers, they are supposed to tell the story of what happened. If there is a better way, a more detailed way, to tell the story that is where their responsibility lies. Improving the product for the consumer is at the core of most business and that is what Arthur and Mirtle are doing. They are creating a more in-depth product for the consumer.

Pure narrative pieces still have a place in hockey. Narrative is there for when a story away from the game needs to be told. Narrative is there when we need to talk about cancer and all the evil it brings. Narrative is there when we need to celebrate the good, remember the lost. Narrative is there when we get to see beyond the game, when we get to meet the people who play the game. Analytics are there to help tell the story of the game.

Analytics allow someone to say "Dave Bolland has never been good at possessing the puck, even when placed in positions to succeed. Therefore, Dave Bolland is not the best player to build your team around." They can also allow someone to say "Bryan Little starts over 50% of his shifts in the defensive zone and is still a plus possession player. Maybe Little is a better player than I thought." Analytics simply allows you to have more information to make statements about players and come to conclusions about why some teams are failing.

*No numbers in the examples are real. I just used them to show the concepts. There are no actual quotes in this piece. This is a bit of a narrative.