|
Summer is here so time to give some THANKS! |
|
|
|
|
|
It’s time to give thanks.
The gong show that is June, especially this year with the expansion draft, and then July with UFA period has come and passed. Hopefully most GMs are fishing, golfing, or doing whatever else they like now. The same for the players, even the ones who don’t see hockey in May need time to recoup.
It’s also time for an amateur hockey writer to give some thanks. Hey even we get some time off from thinking about new ways to anger some, frustrate others and mortify the rest of our readership. This ‘down time’ is also the same time to pause and reflect and the last couple of days on hockey Twitter seem like a good catalyst.
First the gratitude, thank you. On behalf of everyone who uses the free work that’s available to all hockey fans of all levels I’ll say ‘thank you’ even though no one anointed me to do so.
The work, which is really a representation of your time, oh so much time, is great. It has created a whole new dynamic to enjoying hockey and the sport. The discussions are vibrant, informative, hostile but important. I’m grateful.
I’m not a stats guy, nor am I some progressive thinker about predictive data or analytical thinking, I sure as hell appreciate those that are though. Again, thank you.
Thanks for the database work, (all of you), the websites, scraping, interfaces, tools, models and everything else I’ve missed. In years past I could name almost everyone fairly easily, now that list would be too long to remember let alone research for a piece like this.
I know that none of the stats-inclined folks read my work. I wouldn’t if I were them but if they did, I hope I didn’t make a mess of what they are trying to say/do with what they give away for free. I’m just grateful you do.
I’d offer you some advice though. I think I can at least return a bit of love and something to the cause seeing as I’m basically a taker.
While data use in hockey is relatively new compared to other sports, and the use in other sports is really new compared to other industries. I work in one of those industries and I think it offers some insight to what I see in the hockey data/analytics community.
The latest debate has everything to do with models. How some are able to take data and use it based on long standing and tested statistical formulas is really cool and fascinating but also controversial as the last few days have played out.
Where I come from ‘modelling’ is a process to analyze risk, events, places, things, and people. The modelling ultimately helps predict frequency to which they might fall susceptible to something or cause something to happen. At that point the models are tested and tested and compared and tested again before leading to action.
What am I talking about? The worlds second oldest profession: Insurance.
In this world instead of finding out what the magic scenario for success is actuaries look at what the worst case scenarios are, they look at risks and perils. What can cause loss. What I see in the controversial hockey models is the idea of an all-encompassing metric to identify what players have the best value. I was unaware that the problem of determining the ingredients which make up the best players was solved. I could be wrong :).
As insurance has progressed one of the trends that has emerged is using data to try to underwrite risk completely- reduce it to the bare minimum and assign an appropriate cost to those that cause the most loss. What has ended up happening is that you lose the perspective of what insurance does, it’s core purpose. The premiums of the many pay for the losses of the few but when the premiums become isolated to the identified loss causers, the system falls out of sync.
Taking this thought a bit further is that the insurance business has tried to isolate where losses come from so well that they consistently, over time, miss the fact that you cannot predict everything. I see this trend happening within the group that does such great work with hockey analytics. It’s the search for the ‘golden egg’ but what ends up happening is when the focus becomes too tight we miss the forest from the trees.
Hockey is a really different sport compared to other team sports and probably the only one that mimics it for variables is lacrosse.
In baseball the play is static and resets in the same form each time, as does football. While those games have great strategy and tactics they tend to lack free form spontaneous creativity. Basketball and soccer have more of it, as does rugby but within those games the setting of play and rules constrict them when compared to hockey. The only thing that hockey does not have that other sports deal with is weather. Unless it’s a rare outdoor game the weather never affects a game of hockey.
In a hockey game the glass, boards, feet, arms, shoulders, sticks, posts, netting, officials all influence the play of the game and with far greater frequency than any of the other sports. There’s nothing like it and those variables affect the game so much with such frequency that it is harder to measure the impact. Now add that hockey like soccer, lacrosse and basketball has players going back in forth from offence to defence and more variations come into play.
The analytics folks do a great job of compensating for chance and variables, what I’m saying isn’t new to them or for the reader either. However it’s worth remembering. If you look at insurance the same thing happens: variables/intangibles. Strange unexpected things that can’t be predicted because they are simply unpredictable. They become harder to quantify and prepare for because they are accidents, things that were unintended.
Hockey is a lot like that in the flow of the game. Many things that do happen maybe should not happen and were unintended. The models as far as I can tell do a decent job of dealing with these issues but again people are challenging them, and from what I have seen, because the focus is too tight.
There’s a great quote from a youtube video about a welder’s hangover cure. “You try to make something idiot proof but the world keeps making a better idiot”.
That’s where I see the problem with hockey data and analytics. It’s not that people are being idiots it’s that the harder we try to isolate and manage data to be encompassing there are things that happen which limit the outcome. It happens over and over in all sorts of settings.
My hope is that if any of these smart folks see this they pause for a second and realize that whittling down too far only creates other problems, thus what has been done is pretty damn good. However, there’s probably lots to learn at this level in a more nuanced way before change inevitably happens. Basically what I’m trying to say is don’t make the same mistakes others have made by trying to develop an all encompassing measurement. It probably doesn’t exist to the degree that it’s hoped for because there are too many things that change.
Don’t stop doing things though, that would be terrible but also recognize that there are limits and it may take longer than you wish to move past them. Again, this is simply an observation based on my experience and seeing a similar trend happen. Instead of trying to isolate and find something great that no one has seen perhaps simply identifying the bad is enough. A series of metrics to help everyone know and accept what won’t help teams until the next evolution of analytical value.
An ounce of prevention is worth a pound of cure and perhaps that’s a better focus until the necessary data points arrive to take the modelling and debates to the next level.
Winnipeg Jets Blogger •Winnipeg Jets Writer • 
Join the Discussion:
»