Vancouver to leverage statistical analysis?
There's a very interesting article this morning over at the Vancouver Sun, discussing the analytical work of former NHL GM Mike Smith, and how Mike Gillis, who just took over as Canucks GM, might make use of advanced hockey statistics in his new role. Here's a quote from Smith that pretty much says it all:
"Everybody will be doing some version of this in four, five or six years. Most owners are doing business analytics in their own businesses. It's here to stay."
In a salary cap-driven NHL, payroll dollars are a constrained resource, so it only makes sense to apply whatever tools you can to ensure the maximum return on that investment. Particularly for a team like Nashville, which doesn't have the luxury of living at the high end of the salary cap range, this is even more true.
One of the great frustrations I run into doing some of this work is that I have to spend too much time reassembling information into data; extracting play-by-play files, merging that with other tables, etc. If it was only possible to get my hands on the real, raw data that the NHL records, the possibilities could be incredible; envision performance metrics for a given player, with on-the-fly comparisons for how those change depending on linemate or opponent. I've got a lengthy To Do list of topics that could be very useful to a Hockey Operations staff, but building the data with which to start is a huge obstacle.
So here's an open message to David Poile and the Predators Hockey Operations staff; I'm in the middle of a job search right now (but for how long?), and am available if you're interested in a short-term consulting gig to help kick the tires on a few things. I can already suggest a strategy to improve your odds in the shootout (which might have helped this year), but there are many, many areas that are ripe for harvesting. It's only a matter of time before this becomes common NHL practice; the only question is, who's going to get there first?
Labels: statistics
3 Comments:
oooh - I just found your blog via Canucks and Beyond.... and I am lapping it up. I am a COMPLETE stats nerd!
It's what I do for a living (business stats not hockey stats) and I am convinced that the pretty numbers are a lot of what gravitated me towards North American sports in general and hockey in particular. We Brits just don't do sports stats with anywhere near as much enthusiasm.
Putting your blog in my reader to digest avidly on an ongoing basis.
Really, this stuff is like p0rn to me :)
And as for Van stepping up their numeracy - glad to hear it, it can only help. And I'd be happy to relocate to BC if they fancy getting me a work permit set up.... :D
It's interesting to see how all the tools are available for doing some serious work relative to hockey, all we need is the raw data to get things rolling.
For example, Information Builders put up a demo of one of their "Active Reports" covering very basic data from the 2006-7 NHL season online. If, from a business intelligence perspective, you were able to do something similar built from the most granular level possible (individual play data), stat-geeks could mine the possibilities for a looooong time.
Like I mentioned in the piece, it's inevitable that most teams will get there at some point, the only question is who's going to take advantage of it first...
that 'active report' is quite cute, but too constrained for me. I'm one of the really sad cases who wants the nitty gritty, just like you, I suspect.
I have a clunky method that I use to parse data from the play by plays into Excel, and occasionally SPSS, so that I can fool around with the stats to my heart's content.
I keep meaning to make a macro to do the whole extraction - but I have got used to my inelegant methods.
I do wish there was a way to get at the data in a more analysis-friendly format.
Oh dear lord I'm a geek
Post a Comment
Links to this post:
Create a Link
<< Home