This picture has nothing to do with the article below, but I really love it.

Life can only be understood backwards; but it must be lived forwards.
~ Søren Kierkegaard

With the release of Moneyball in theatres, advanced statistical analysis in sports has gone mainstream. Cam Charron did an excellent job a week ago outlining how the principles of Moneyball, the identification of market inefficiencies through the use of advanced statistical analysis, can apply to hockey and particularly in the search for the next underdog. We are experiencing a paradigm shift in hockey as more and more teams begin adopting this type of analysis and using it to drive their scouting, player acquisition, and tactics.

One of the dangers inherent in a paradigm shift, however, is swinging too far in the opposite direction. While I have used advanced statistics many times in the past, I’m always wary of losing sight of where statistics come from: the performance of individuals. As Søren Kierkegaard said, “Life can only be understood backwards; but it must be lived forwards.”

Kierkegaard is regarded as the father of existentialism, and this famous quote was in response to the dominant philosophy of his time, Hegelianism, which attempted to explain all of reality through reason. And what are advanced statistics but the application of reason to sports? For Kierkegaard, however, life can never be completely understood as it is constantly being lived. There is no static position of complete rest from which you can look backwards at your life to understand it. At least, not until you have died, at which time the point is moot.

The difference in hockey is that while life is constantly being lived, the game is not constantly being played. A game has a beginning and an ending, and once the game is over we feel justified in looking backwards and reasonably assessing what occurred. We do this through the most basic of statistics – winning and losing – or we delve into scoring chances, Corsi, Fenwick, and whatever other types of analysis we wish to use. Once we have done so, we feel like we have an understanding of what happened. If we take many of these games over a long period of time, we begin to paint a picture (or draw a line on a graph) and claim that we have enough knowledge to make judgements.

Here is where the danger and the limitation comes in. Though the game has an endpoint, a static place from which we can look back and analyse it, life does not. And life, for the players, goes on. Statistical analysis constantly treats sports as deterministic, that with enough knowledge, the results could be predicted, with any anomalies attributed to luck or natural variation. But there is one more element that they rarely take into account.


Hockey players make choices both on and off the ice that dramatically influence the results of a hockey game and, thus, the statistics. Players make choices in the off-season that influence their performance in the following season. They make choices in the middle of a game that influence whether or not they score a goal. They make choices between games that help or hinder how well they do in the following game.

So far, this has all been pretty abstract. Let’s get concrete in a hurry.

Also, Martin St. Louis is pretty dang good. That helps Stamkos a little bit.

Steven Stamkos is one of the best goalscorers in the game today. We know this because of the statistics: he scores a lot of goals. Simple. But how did he get to that point? He made a series of choices throughout his life: as a kid, he shot 300-400 pucks per day. In the off-season, he made the choice to work out with Gary Roberts and completely change his eating habits.

Tanner Glass is an excellent human being and a hard-working energy guy for the Winnipeg Jets. To state the obvious, he’s not as good at hockey as Stamkos. Why is that? Glass is intelligent and has obviously put a lot of time and energy into playing hockey as a career, but he doesn’t spend his off-season shooting hundreds of pucks every day or working out with Gary Roberts. Instead, a couple years ago he went to South Africa with his fiancée (now wife), saw four World Cup games, swam with sharks, toured Robben Island where Nelson Mandela was imprisoned, and visited Zanzibar. Only after this trip did he get back to training.

Some people make the choice to devote their entire lives to one thing and one thing only. Sidney Crosby doesn’t seem to do anything but eat, sleep, and play hockey. The result, which can be measured statistically, is that he is a superb hockey player. Most other hockey players make different choices with their life. They end up good at hockey, because they work and train hard, but they make other choices with the rest of their time.

Or, look at the example of Lee Sweatt, who signed with the Ottawa Senators this past off-season only to retire prior to training camp. He made the choice to retire at the age of 26 to pursue other interests. There’s also the example of Victor Oreskovich, currently playing for the Chicago Wolves in the AHL, who quit hockey for a year-and-a-half before returning to the game with the Florida Panthers. These players made a choice to stop playing hockey altogether, but it can be less extreme: if a player simply grows tired of the grind of training and chooses to work less hard during the off-season, his previous statistics will not be a reliable indicator of his future results.

I’ve written before about why intangibles matter: to sum it up succinctly, intangibles matter in hockey because they have a tangible effect. If a player is gritty or has heart or character, those traits should have results that are measurable by statistics. If not, then they shouldn’t affect how a team signs free agents or drafts prospects. The difficulty is in knowing how or when these intangibles will translate into tangibles. There’s a reason why the teams that have embraced statistical analysis still employ a wealth of scouts, even in baseball, where this type of analysis has had more time to blossom.

If a scout describes a player as gritty, he may mean that the player consistently makes the choice to risk injury or pain by going to the front of the net, where he constantly gets hacked, cross-checked, and elbowed. A player described as having heart may be one who chooses to battle through injuries. A player with good character and discipline makes the choice to follow a strict sleeping and diet plan so he’s always in the best shape physically and mentally on game day.

All of these choices should translate into tangible effects that can be measured statistically, but it is the tendency to make those kinds of choices that scouts should be looking for.

Does that high-end prospect coast through Junior hockey on the strength of his natural ability? When he faces stiffer competition, will he make choices that improve him as a hockey player or will he wither under the pressure? Is a spike in shooting percentage the result of luck or did the player make the choice to spend the bulk of his off-season improving his wristshot? Is that forward’s drop in production in the playoffs simply a by-product of small sample size or does he make the choice to stay on the perimeter and not get his nose dirty in front of the net?

Statistical analysis can explain that one player was better than another and make a reasonably accurate prediction of future results, but it has its limitations. Statistical analysis, no matter how advanced, can’t analyse choice.

Life must be lived forwards. Hockey must be played forwards.

Comments (14)

  1. Holy shit did you just talk about Georg Wilhelm Friedrich Hegel, Søren Kierkegaard, and hockey? Did we just become best friends?

    I’m still in awe. Great read man

  2. Such a great post. Well reasoned, man.

  3. Tanner Glass could go back in time to tell Mr. Glass Sr. to force his son to shoot more pucks than there are stars in the sky and he would not ever be close to what Steven Stamkos is as a hockey player. Not the best analogy.

    • How do you know? The existentialists say that we are defined by our choices. If the situations were reversed – if Tanner Glass made the same choices throughout his life that Steven Stamkos did and vice versa – would Glass have been the first overall pick with 119 goals in his first three seasons. Would Tampa Bay have “Seen Glass?” ads for his rookie year?

      I think my point still holds up. Sure, if Glass now spent the off-season shooting 500 pucks per day he might not become a sniper like Stamkos, mainly because we learn skills so much more quickly at a young age, but his shot would improve and that choice would change his performance on the ice. But, since he got married this past summer, I doubt he was going to spend his time shooting pucks.

      • I still don’t buy it. The gaping chasm between Glass’s and Stamkos’s skill sets aren’t just due to stick-to-it-iveness (to borrow a phrase from Deadwood). There has to be some accounting for natural ability and circumstance. I think Stamkos’s work ethic is admirable, but I bet his off-ice regimen has more to do with developing & honing skills that were already ingrained in him.

  4. “That was awe-some” ::clap, clap, clap clap clap::

  5. Mmh.
    I think that, to an extent, you can also model most people choices. Of course, the models wont be 100% accurate, but more often that not people are predictable. Worse, the more you tell them they are free to choose, the more accurate the predictions.
    On the other hand, I think that injuries are much more difficult to predict and have a way bigger randomness that’s going to lead to way bigger disparities in statistics from one year to another – or even from the first half of the season to the next, look no further than Crosby: 66 points in the 41st games, zero in the next 41- how ist that for statistical non predictiveness?
    But that’s the whole point of the philosophical debate of fate vs choice, and from my point of view neither extreem is valid: nor 100% choice, nor 100% luck, after that let’s meet somewhere in the middle. Where? I think purely NHL hockey speaking, luck has a small edge over choice. But it’s more a gut feeling than an assertion.
    But I have arguments to defend my thesis: just the small amount of people (only 690 NHLers in the world) and the little amount of time you have (very very best case is 20 yrs) renders luck necessary even if not enough.
    In life in general, I think choice is lore important than luck.

    • I don’t necessarily disagree. Luck is a major factor, as odd bounces can drastically change the outcome of certain events on the ice. The difference between a goal and a post can be mere inches. But choice often doesn’t even enter into the discussion. In discussions of advanced stats, players are frequently treated as machines that will consistently perform the same way time after time.

      I just want choice to enter the discussion.

  6. I think you’re beating up a straw man here – anyone who treats hockey stats analysis with any form of seriousness understands its limitations. Don’t let better be the enemy of perfect, the saying goes.

    The interesting question comes when you have a choice to make between a talented player who may not boast the “intangibles” lauded here, and a grittier but less-skilled guy.

  7. Rec’d for Soren-luv.

    Unrec’d for Nietzsche-lack.

  8. Great post, Danielson. Note in the paragraph, “The difference in hockey…” you’ve written “look” when it should be “looking.” Your argument does to an extent ignore the fact that some of us are naturally more athletic than others. For example I don’t have the eyesight to hit a baseball thrown from sixty feet away at 100 miles an hour. It wouldn’t matter how long I practiced, I would never be a Baptista. Thus I tend to agree with the comment that Tanner Glass could never be a Steven Stamkos, though the possibility exists. I just suspect Stamkos has more natural ability than Glass. Cheers

  9. Great article! Now can we discuss the fallacy in using counterfactual conditionality for establishing causality.

Leave a Reply

Your email address will not be published. Required fields are marked *