This is a superb, superb answer to an "advanced stats" question from ESPN's Craig Custance's Friday Insider entry/mailbag feature:
As I try to be more of an advanced stats believer, I find myself with more questions than answers. First, how does a team get better at advanced stats? Is it putting more pucks on net? I'm pretty sure every hockey player ever has given that as an answer. But seriously, what would fix your team’s problem? Is it a certain style of play? If it is, then one could argue that once a team wins the Stanley Cup with a different style, we’d see that be the new "in" thing.
Good to hear from you, my friend. There’s a lot to tackle in this question. First, there are definitely things coaches can do to improve their team’s possession analytics, but whether or not it leads to wins is up for debate.
For instance, I remember talking to Dallas Eakins last year after he was fired from Edmonton and they made systematic changes that resulted in improved analytics.
“Very, very subtle changes,” he said at the time. “But they were paying off.”
And the numbers suggest that zone entries impact the possession numbers, so if teams make a concerted effort to carry the puck in rather than dump it in, that would theoretically improve the analytics.
To me, the analytics we see are more useful for roster construction. If I’m trying to improve the possession numbers, I’m doing it through roster improvements. I’m focusing on defensemen who can quickly close the gap on players in the neutral zone to force a dump in. I’m focusing on defensemen who are also great skaters who can quickly get the puck and get it out of the zone to start the rush the other way. All things being equal, I’d like a goalie who can play the puck to help those defensemen.
But to your last point, there is more than one way to win in the NHL. The Tampa Bay Lightning play a different brand of hockey and are built differently than, say, the Los Angeles Kings when they were at the top of their game. Possession numbers aren’t the end-all, be-all, they’re just part of the equation.