For those who aren't familiar with Point Shares, it's a new single-number statistical metric that hockey-reference rolled out recently.
There's a thread on the HOH forum discussing it.
Short version of this post: Point Shares, and especially Defensive Point Shares, are statistical smoke.
As far as I can tell, defensive point shares for Housley and other defencemen of his era are calculated by 1. Dividing up team defensive performance among all players, and 2. Giving players positive or negative adjustments based on their plus-minus.
In my opinion this does not work at all. There are two obvious errors. First, plus-minus is at least as much an offensive stat as a defensive stat. Housley was a relatively high-scoring defenceman at even strength. Point Shares is double counting that value. Second, as far as I can tell Point Shares does not distinguish between power play, even strength, and shorthanded performance. Housley spent very little time on the penalty kill. Therefore the metric is giving him credit for team defensive performance to which he did not contribute.
Also, a large part of a defenceman's value is the quality of competition he faces. I doubt Housley was being matched up against the best players. But I'll give Point Shares a pass on that. It's very difficult to measure this statistically and we don't have the data for the vast majority of NHL history.
Defensive Point Shares are poorly designed. If the list of leaders looks like a list of good defensive players, it's because long-career defencemen on good defensive teams tend to be good defensively. Housley was an exception, IMO.
The person who runs hockey-reference.com and implemented this stat, Justin Kubatko, is not a hockey guy. He runs some of the sports-reference.com family of sites. He mentions that it is based on work done by Tom Awad, Iain Fyffe, and Alan Ryder, who are leading hockey stats guys. But I think his lack of subject matter knowledge has caused some design errors, and the Point Shares stat needs more work.
