|
Like a lot of kids in Canada, I was pretty much born with a hockey stick in my hand. I grew up playing pond hockey on backyard rinks and shooting pucks into the drywall in my basement, much to the frustration of my father. I played AA all the way through high school (this is pretty much the highest level you can go without getting drafted into junior hockey right away), but when I realized that I wasn't good enough to go pro (I was 15 when I found out that I didn't get picked for the junior draft), I turned to many other sports, one of which was Starcraft Brood War. Basically, I started picking up everything and just learning things. Learning is the best thing in the world man! It was the moment that I discovered I would never become a pro hockey player that I began to realize that my favourite thing is the process in which you go from complete and utter junk at something to a level where you can actually have a conversation with other competent amateurs and professionals.
Anyway, back to hockey. I still remain a HUGE fan, and I follow the NHL and its associated minor leagues (the o, the w, the q etc) almost as fanatically as I follow SC:BW and to a lesser extent SC2. I ended up discovering a passion for mathematics and statistics somewhere in my high school years, and ended up pursuing an engineering degree in university. I've always wanted to try to apply my knowledge to analyze and predict sports like hockey, and the recent book/movie Moneyball has stoked my interest in this subject once more. Note: If you have only seen the movie, and were curious about the actual analysis behind the scenes instead of just the drama surrounding the A's and their somewhat cryptic managerial staff, READ THE BOOK! Michael Lewis goes into such detail with regards to the statistics and methods surrounding sabermetrics* that I was almost immediately inspired with ideas for formulas and ratios that might apply to hockey players. From an adjusted time-on-ice metric to generating 3-D probability surfaces to model scoring chances in the offensive zone, I've come up with a million and one ideas of things that I want to explore and possibly publish.
The greatest issue in applying statistics to hockey to the extent that they have been applied in baseball is that hockey has an infinite number of possible "states". From the number of possible combinations of players on the ice, to shorthanded/powerplay to the simple positioning of the puck on the ice makes it INCREDIBLY hard to create simple statistics like "scoring average" or "goals generated per minute on ice". Baseball, on the other hand, has a finite number of "states" to the game. There are 3! possible combinations of positions of the players on base, and there are 12 different counts that are possible, and there are 18 different states with regards to the innings (top/bottom of each 9). No matter what happens, the rules of baseball dictate that the number of states be finite, so the "events" which dictate a game can be easily split up into independent events: Pitches. Strike outs. Singles into right field. Home runs over the left fence with two men on base. You can't delineate hockey so easily, although you can still identify states that have varying amounts of "potential energy" like powerplays, penalty kills, or 4 on 4 overtime.
More relevantly, hockey is a full contact sport. A young batter who has shown that he can hit 90+ mph pitches has shown a specific "skill" that can almost be transplanted into the major leagues. But a young hockey player who can skate just as fast a pro can't immediately be placed on the ice in a playoff game without worrying about his physicality compared to his older, more experienced counterparts.
But even despite all these problems, I've thought of a really really cool idea based on something that I read in Moneyball. I saw them breaking down the strike zone into quadrants and mapping how the on base percentage and slugging percentage of their players dynamically evolve based on how the count changed. Essentially, I thought of breaking down the offensive zone into a matrix and mapping the shooting percentages of different teams and different players in each entry of the matrix (area of the ice). To me, a shot on goal is as close to an "independent event" in hockey as you can get. Even thought you can have traffic in front of the crease or deflections which affect the shot itself, I was thinking about making different probability surfaces for even strength and the powerplay, since the on-ice strategy changes quite a lot depending on which of those two situations you are in. Unfortunately I don't have the raw data that the NHL teams do, so I've got a billion .txt documents sitting on my desktop with really really interesting ideas and formulas that I have NO GODDAMN DATA TO WORK WITH. QQ.
Thus ends my basic discussion of analytics. If you liked this at all, feel free to PM me or bounce ideas in the comments.
Check this out too: I googled NHL shot statistics and I found this analysis which is very similar to what I was thinking about doing, just at a much smaller scale.
http://hockeyanalytics.com/Research_files/SQ-RS0910-Krzywicki.pdf
I'm hoping to make this a weekly thing, so next week I'll be taking a look at a paper I found online, a regression approach to adjusted player plus/minus.
Later all, thanks for dropping by!!!  <3
|
Sounds interesting! I'll definitely follow your blog, but you should do the statistics even if its just for yourself!
In the U.S, I started playing hockey around 3 years old. I could skate pretty well by age 4, and I was "drafted" I guess you could say, to a 4 year old "travel league".
Unfortunately, getting up at 5am when I was 4 was about as exciting as getting up at 5am now. I remember my dad pulling my pads onto me at 5am and screaming and crying because I didn't want to get up.
So my parents eventually stopped forcing me and I quit, until I was about 9.
I got back into ice hockey and played all the way until my freshman year in High School, where I was the only freshman to make the Varsity squad. I was a left-handed shot and played RW. We made it to states that year and I put up two goals and an assist in a game we lost 4-3 to the team that later won states.
But I unfortunately realized something after that season: I hated my teammates and I really disliked my coach. In fact, it made me not want to play. I had played with the same travel team for 6 years before that, and unfortunately I think my uncomfortable-ness led to my decision to quit the team and pick up tennis more competitively.
TL;DR, I'm in the US, but I love hockey. Although I'm in Detroit, so I might as well be canadian anyways, eh?
Keep it up! I'm intrigued!
|
Very cool stuff. As a huge hockey fan and a bit of a stats junkie I'm quite interested in this. I always cringe when I look at the Devils' shot charts against the Flyers because they are always taking low percentage perimeter shots, now I can actually put numbers to how bad their shooting decisions are.
|
I fully endorse any hockey thread, and I look forward to follow ups.
Who is your team??
|
Considering he said he's in Detroit I would assume he likes the Wings.
|
But I am in New Jersey, yet I have a sophisticated taste in hockey teams
|
On March 22 2012 13:46 dGHaiL wrote:
Sounds interesting! I'll definitely follow your blog, but you should do the statistics even if its just for yourself!
In the U.S, I started playing hockey around 3 years old. I could skate pretty well by age 4, and I was "drafted" I guess you could say, to a 4 year old "travel league".
Unfortunately, getting up at 5am when I was 4 was about as exciting as getting up at 5am now. I remember my dad pulling my pads onto me at 5am and screaming and crying because I didn't want to get up.
So my parents eventually stopped forcing me and I quit, until I was about 9.
I got back into ice hockey and played all the way until my freshman year in High School, where I was the only freshman to make the Varsity squad. I was a left-handed shot and played RW. We made it to states that year and I put up two goals and an assist in a game we lost 4-3 to the team that later won states.
But I unfortunately realized something after that season: I hated my teammates and I really disliked my coach. In fact, it made me not want to play. I had played with the same travel team for 6 years before that, and unfortunately I think my uncomfortable-ness led to my decision to quit the team and pick up tennis more competitively.
TL;DR, I'm in the US, but I love hockey. Although I'm in Detroit, so I might as well be canadian anyways, eh?
Keep it up! I'm intrigued!
Yeah, if it's not fun then there's no point doing it. That's mainly the reason why I quit playing brood war and started playing SC2. I got to B- consistently and at that point, the amount of effort I'd have to put in to move up to B+ or A was so huge that it wasn't worth doing, it wasn't fun. Don't get me wrong, I still like brood war as a game so much better, and i still regularly stay up to watch spl, but playing sc2 is so much more casual and fun at this point.
On March 22 2012 15:14 yakitate304 wrote:
Very cool stuff. As a huge hockey fan and a bit of a stats junkie I'm quite interested in this. I always cringe when I look at the Devils' shot charts against the Flyers because they are always taking low percentage perimeter shots, now I can actually put numbers to how bad their shooting decisions are.
Yeah, I think a lot of teams endorse the method of "just get an asston of shots on net" way too much to be honest. If you've watched the leafs the past two seasons, they seem to always outshoot their opponents in games they've lost, and that's because half those shots are on even strength rushes from like the blueline or top of the circle lol. I'm still writing exams at this point, so I doubt I'll be doing any of my own statistics in depth over the next few weeks, but as I progress into the workplace (yay last year of school) I'll have a lot more time to prepare and publish original stuff.
Next week is the adjusted plus/minus paper, so check back for that if you're interested. I thought it was awesome on a cursory reading!
On March 23 2012 00:44 Hawk wrote:
I fully endorse any hockey thread, and I look forward to follow ups.
Who is your team??
Since I've spent equal portions of my life living in Calgary and Toronto, I have equal allegiances to both the Flames and the Leafs, but since I was born in Calgary I'd consider myself at heart a Flames fan. Cup run 2004 baby!
You say you have good taste in teams, who is yours?
|
Rangers all day urry day
anywho, look forward to your blog. There is a sizable hockey fan base here
|
On March 23 2012 04:44 Hawk wrote:
Rangers all day urry day
anywho, look forward to your blog. There is a sizable hockey fan base here
The best part about the rangers is Torts. That guy's a beauty man. no other coach has that kind of attitude and knowledge of the game and it makes him a real controversial figure because of it. i love him though.
|
kitaman27
United States9244 Posts
Looking forward to your blog! Hockey statistics can be pretty interesting, yet they are far less accessible than with sports, such as football and baseball. I'd image finding good data would be the biggest problem, like you mentioned, but you can probably find good sites online if you look hard enough.
|
|
|
|
|
|
EPT
