Who’s up for some discussion on the relative merits of Corsi and other analytics?

………

Nuts to you all then because that’s what I’m writing about, like it or not.

Some of you may be familiar with a new analytical tool pioneered by local bloggers and Oiler fans Woodguy (Woodblog) and GMoney (OilersNerdAlert) called Woodmoney. I’m not going to go into a long description of it here, but only know that once it has collected enough information and been developed it could replace using numbers like Corsi and Fenwick and such for individual players. Why? Because it takes the old “shots are a proxy for puck possession” idea and adds to it the value that “not all shots are equal” and a measure of “not every minute on the ice is against the same level of competition”. Basically, it’s a smarter, more nuanced version of the core possession metric that Corsi and Fenwick are today.

Why bring it up here? Because it represents another step in the ongoing process of finding new ways of evaluating the game. Is it perfect? Well I didn’t go to GMoney or Woodguy for a quote for this article so I’m still technically allowed to say “no”. But it is an improvement.

See, as much as some of them may like to pretend, analytics guys don’t actually know everything. The good ones know what they can prove, to the extent that the proofs stand up to scrutiny, and yes, the eye-test counts as scrutiny (if an analytical tool is telling you that Luke Gazdic is a potential 2nd line winger, then you need to go back to the drawing board because entire seasons of play in the NHL clearly suggest otherwise).

Analytics advocates use the available tools to examine the game from a different angle, like changing a camera view to get the right call on whether a goal was scored or not. It is fair to be skeptical. In fact, I generally recommend it. But make certain your skepticism is grounded in an understanding of the process.

If you base your criticism of analytics on the argument that “numbers don’t tell the whole story”, well you’re responding to a statement that very few analytics people will actually make: that numbers would ever be used to tell the whole story. At best, this misunderstands the topic. At worst, it deliberately misrepresents the other side of the argument in order to enhance one’s own position.

Dissecting the Air

What is fascinating about human beings is that we are, in many ways, little changed from those ancient ancestors who divined meaning from the flights of birds or the way sticks fell in the dust, yet at the same time we hold the capacity for thorough and rigorous scientific thought.

In the 17th century scientists and philosophers (at the time there really wasn’t much, if any, difference – case in point, after developing his theory of gravity Newton began working on alchemy and studying the bible for hidden meaning) were trying to work out the physical world and the nature of matter. They were confounded by air. They knew it exists because a lack of it causes death, yet it could not be captured, measured, or seen. It can be hot, cold, wet, dry, carry disease, scent, and otherwise demonstrates all manner of qualities associated with physical matter.

Yet for the life of them they couldn’t classify it.

Yeah, “classify”. They were those kinds of guys. Rarely popular at parties, and often times poor conversationalists, but they were puzzle solvers and pioneers of many of the fundamental building blocks of the modern technological world.

Eventually these thinkers settled on referring to our atmosphere as the aether. It was a word chosen to represent the intangible, enveloping environment and allowed them to continue on exploring the physical world until they could better nail down what this stuff was.

Then around 1772 a stubborn, pain-in-the-ass, know-it-all rebellious English clergyman isolated the principal component of air in it’s gaseous state. He discovered oxygen. Anyone who has watched a movie where someone lights up a cigarette around pure oxygen tanks knows that it is not the sole component of air, but in discovering it Priestley had begun the process of dissecting the atmosphere around us and eventually we mapped out the rest of the chemical elements that make up the air we breathe.

*It should be noted, several others claimed to accomplished the same thing as Priestley, but they were a German-Swede and a French noble, so cultural bias being what it is, the history books tend to give credit to the Englishman.

Why the History Lesson?

Because before Priestley, air was difficult to categorize. It was critically important, this was obvious, but arriving at an understanding of it proved exceptionally difficult. Whereas some were content to simply call it air and describe it as either good or bad (that which was healthy for you versus that which carried disease) others were curious and wanted to explore it and figure it out.

Think of the game of hockey as taking place in a bubble located on the ice. The players and puck all racing around and colliding like molecules, and within the chaos a massive storm of individual events. Hits, shots, blocks, goals, saves, passes, penalties, and all manner of theatrics. Reporters have traditionally taken these events and woven them skillfully into a narrative that boils down the game to a storyline, complete with heroes and villains, victories and defeats. From this we also pulled out various numbers like shots on goal, passes, ice time and so on. These helped us colour in the spaces between the narrative.

For many, they were content to leave the game in this state. But eventually some came along who wanted to dig a little deeper, at first it was those motivated by their own livelihood, such as coaches whose employment rests on being able to identify and encourage successful conditions on the ice. After some time, and as technology improved, the interest spread to others in the amateur community.

What analytics guys are doing today is looking into the aether of the game and pulling out more information.

Just as philosophers began the discussion on the nature of our atmosphere which later scientists then built upon by discovering ever greater intricacies in the air we breathe, so too are the people who work in the analytics field of hockey. Some of the concepts are good and worth keeping, others are either discarded or need refinement.

This Brings Me To Another Point

Understand that there are limitations to any and all of the current hockey analytic tools. There is no single measure that tells us how good a player is (or isn’t). This is important because there are those who would seek to commodify analytics by marketing to those who demonstrate a gap in their knowledge of it. Put it another way, companies and individuals will be more than happy to sell you whatever flavour of snake oil you desire, exploiting your willingness to believe that it offers a remedy to your ailments.

But the point of the process isn’t to sell people on the idea that you have The Answer. Or it shouldn’t be, and I’d be wary of those who do.

No, the point of the process is to develop an ever more refined and nuanced way of measuring and evaluating players and teams. That is the scientific process. It isn’t terribly romantic and it doesn’t often resonate emotionally with fans, but it is how you go from guesswork and vague concepts to a more developed description of the chaos of the game.

I read critics of some analytical metrics and wonder if they grasp this concept. They will say that stat X isn’t useful or distorts the picture too much, often suggesting another metric they prefer. Yet in some cases what they fail to recognize, or sometimes omit from their criticism, is the fact that each metric is just one measure amongst many. Like someone railing against the Metric system because fluid ounces is completely unsuited for measuring a person’s mass, they’ve failed to grasp the very basic concept that not every measure is necessarily intended to be applied to every measurable facet of the game.

Now, nobody asks you to know everything that goes into the air you breathe. So long as it’s there when you need it, most of us are perfectly content to accept that it is there and that’s all we need know.

Nobody is asking you to know every arcane statistic that goes into the game, just as you don’t need to know about how to understand a quarterback rating to appreciate football.

The numbers guys are going to continue to work and refine their ideas, not all of them will be met with enthusiasm (in fact I’d bet that it becomes a badge of honour in some circles to publicly denounce them) but that is the point. This is the scientific process at work: have an idea, test it, find out something, test that something, prove or disprove, then repeat the process again to improve on the original theory or move on to related areas.

Tomorrow we will hopefully know just a little bit more than we did today. This doesn’t scare me, and I hope it doesn’t scare you because you must remember, the ocean air didn’t smell any less sweet the day after Joseph Priestley’s discovery than it had before.