Quote:
Originally Posted by Tinordi
A guide to advanced stats:
http://canucksarmy.com/2013/10/10/a-...d-hockey-stats
I almost posted this as a thread the other day but think it has a good place here. Pretty much discusses what advanced stats are and how they're used.
A good article in all and describes why they're so compelling. |
I would not call this article a reason to consider these advanced stats as compelling. In fact, these aren't even advanced stats, they're just non-traditional stats without any basis to their validity. Consider these issues that I believe makes them untrustworthy and impossible to validate.
Methodology. There is no clear or consistent measure for many of these stats. What is a shot on goal varies from arena to area. Boston is nefarious for pumping the stats up for the Bruins. The same can be said for other areas around the league. There is no clear definition of what a shot is, or when a shot should be counted, and applied across the league with any consistency. It is subjective and interpreted differently across the league. To me this invalidates anything to do with Fenwick measures.
Consistency in measures. I think this falls outside of method because of the way the variables float around. Take Corsica, which measures the quality of opposition. The constant is sorely lacking. The bad teams are all going to have better Corsica numbers because everyone they play is better than them. Conversely, the good teams are always going to be rated low because everyone is worse than them. This measure can wildly fluctuate during the season, as a bad team gets better or a good team heads south. I've always sat back and looked at this and thought, well of course this guy is going to appear to be playing against stronger competition, he plays for the Edmonton Oilers. Everyone IS stronger! It doesn't take a Phd in advanced statistics to figure that one out.
To me these stats are an answer searching for a question. The first time I ever saw Corsi as a stat was when the Oilers sucked hard and were trying to pump the tires on some of their players. This stat was dreamed up to show that a guy like Shawn Horcoff was comparable to Sydney Crosby, if not better in some ways. It was a bull#### stat then and it's even a bigger bull#### stat now.
The lack of consistency of data collection and application of method makes these things useless. These stats are used to make arguments that have no validity in the real world. There is no way that anyone who watches and understands the game would make some of the leaps in logic these stats drive people to, unless they had an agenda to promote. Again, an answer looking for a question.