I've been working with a company for the past while now that deals with sports data. Perhaps you can verify this for me, but I've noticed a pattern: if a sport is American, and in a college, the stats and math behind it are disproportionately more convoluted than ANY OTHER sport, league, etc.
We've explored analyzing the impact that coaches bring to teams and athletes and have found some interesting results.
The big problem though, is that with the NFL at least, there are just wayyyyy too many coaches a player touches and that makes it hard to model to a degree.
I am interested in adding college and professional hockey stats to my site http://hockeybias.com. Please contact me if you are in a position to help me. Guy at hockeybias dot com
I'm not interested in sports analysis per se, but I am interested in interesting data analysis in general and have a soft spot for excellent visualizations.
I'd be happy to be proven wrong but I can't imagine there are many sports analyses that aren't written for sports fans who have substantial knowledge of the rules, sides, strategies, etc.
I generally appreciate sports analytics, but in the sport I watch the most (NBA) they have a long way to go.
While I admit this is a personal failure, I too often find the articles (especially by guys like Hollinger) obnoxious. They love to tout simple stat concepts (regression to the mean! sample size!) as keen statistical insight. Another example is the PER metric: it is a simple weighted average of multiple stat categories but the author regards it as the authority by which all players should be measured.
Additionally, nba stat analysts believe stats tell the entire story. The nba right now, more than any other sport, in my opinion, is the sport which can least be described by statistics. Reading one of their articles, however, would leave you wondering why they even bother playing the games.
Thanks! I am a fan of Daryl Morey and what he does in Houston. A tidbit you might find interesting: he among others hold a sports analytics conference in Boston at MIT every year. The big thing last year was evidence that the hot hand theory in basketball is false.
21 comments
[ 3.3 ms ] story [ 56.6 ms ] threadI'm working on an interesting project in this space and would love to talk to more data-oriented sports fans.
gmail me at burntreedown@...
Statsheet.com in my opinion is the best sports stats site out there.
Regarding the math aspect - I think prediction of performance or how certain teams/coaches affect different types of athletes is very interesting.
The big problem though, is that with the NFL at least, there are just wayyyyy too many coaches a player touches and that makes it hard to model to a degree.
If HN users are interested in it, they will vote up the stories. If not, they won't.
While I admit this is a personal failure, I too often find the articles (especially by guys like Hollinger) obnoxious. They love to tout simple stat concepts (regression to the mean! sample size!) as keen statistical insight. Another example is the PER metric: it is a simple weighted average of multiple stat categories but the author regards it as the authority by which all players should be measured.
Additionally, nba stat analysts believe stats tell the entire story. The nba right now, more than any other sport, in my opinion, is the sport which can least be described by statistics. Reading one of their articles, however, would leave you wondering why they even bother playing the games.
Edit: clarity.
http://news.ycombinator.com/item?id=481392
I found it fascinating even though my interest in the NBA is lukewarm at best.
So no, at least not analytics for sports that don't really exist outside the US. Data on Cycling or Motorsport would interest me.