Something I want to make clear: several people have noted that the difference between the best faceoff teams and the worst faceoff teams isn't that large, perhaps at most three wins in a single season. The correlation between faceoff percentage and winning percentage isn't particularly high either, and faceoffs "explain" at most 10% of winning. So does this mean that faceoffs aren't significant?
No way. Other than goaltender career save percentage, tell me another basic tracked statistic that predicts 10% of winning and represents a true talent that doesn't regress very heavily to the mean. Hits? Nope, zero correlation with winning. Giveaways, takeaways - not so much. PP% and PK%? Huge regression to the mean. Shots for and shots against? They're not basic statistics - they're driven by other more basic events on the ice that we don't track.
As we try to figure out what players are worth, we're going to need to collect a lot more real-time data. There are a great many significant things that don't make a 3-win difference in hockey; we shouldn't ignore them.
It's quite clear that there are some organizations in the NHL that care about faceoffs, and some that don't. Detroit obviously does - they have the highest winning percentage since the league started keeping stats, and they're the only team to win more than 50% of their faceoffs every season for the last 12 years. Some of that is good luck: the star players they've drafted also happen to be great faceoff takers, unlike Pittsburgh, for example, who have two terrible faceoff men in Jordan Staal and Evgeni Malkin. But some of that is clear strategy: after Todd McLellan took over the San Jose Sharks coaching job, San Jose's faceoff percentage jumped to be tops in the league.
Here are the team faceoff percentages since 1997-98:
I'm surprised that Minnesota's as bad as they are - they were run by defense-first coach Jacques Lemaire for their first eight seasons, and it's only this year that they've won more than half of their faceoffs. You'd think winning faceoffs would make playing defense easier.
The number 1 and 2 teams since the lockout are predictable, but after that, it's a bit of a free-for-all of good and bad teams: