The Puck Stops Here

A couple days ago James Mirtle wrote about which teams were doing the best on faceoffs. In the comments Jes Golbez responded that faceoffs are an overrated stat and there is little correlation between faceoff success and winning hockey games. This statement is true, but contradicts conventional wisdom about hockey.

Conventional wisdom argues that hockey is a game of puck possession. Faceoffs are one way that teams gain possession of the puck and are thus important. We have all seen an important faceoff late in the game (which rarely leads to a goal) and that reinforces the idea.

If we look at the statistics James Mirtle quotes 29 of 30 teams are well within 4% of the 50% success rate which is average. Four percent is one in 25 faceoffs. The difference between a good faceoff team and an average one is winning one in 25 extra faceoffs. The difference between a bad faceoff team and an average one is losing less than one in 25 extra faceoffs. That is not very much at all. The lone exception to that is the Edmonton Oilers who have a 56.2% faceoff success rate. The difference between the best team on faceoffs (and the one outlying point) is winning slightly over one in sixteen extra faceoffs. This is why there is little correlation between faceoff wins and winning hockey games. There is little difference between the faceoff success rate of the best and worst teams in the league.

Hockey is a game of puck possession. Each team has possession of the puck a couple hundred times in a game. If they are a good faceoff team, they might get around three extra puck possessions. These extra puck possessions are far more easily made up or lost by players turning over the puck or taking it away from the opposition. Far more difference in puck possessions come from other facets of the game then from faceoff wins and losses. So faceoff success is more or less lost in the rest of the game. Since most facets of the rest of the game are hard to quantify in any meaningful sabermetric method, there is little reason to overanalyze small differences in faceoff success.

If we look further at the faceoff stats, we find that the teams that are good on faceoffs are a further distance from average then the teams that are bad on faceoffs. The worst teams on faceoffs are the Pittsburgh Penguins and the New York Islanders. They still win 47.2% of their faceoffs (this is only 2.8% less than half). The best three teams are all quite a bit better than 2.8% above average. Edmonton has a 56.2% faceoff success rate. Boston has a 53.7% success rate and Nashville has a 53.4% success rate. I think this is because there are only a handful of players who are very good on faceoffs (by very good I mean they win 60% or more of their faceoffs - which means they would still lose 2 out of 5 faceoffs). Most players are average or slightly below average on faceoffs. Most teams faceoff success rate is a combination of several players who are roughly average faceoff men. Thus most teams are roughly average. The few teams with a very good faceoff player or two are the furthest from average.

When I say there are only a handful of players who are very good on faceoffs (using my definition of 60% success rate or better) this corresponds to only three players with any significant number of faceoffs taken. They are Jarret Stoll of Edmonton, Yanic Perreault of Nashville and Rod Brind'Amour of Carolina. These three players play on the four best faceoff teams (Carolina is fourth). Boston's success is due largely to two players Patrice Bergeron and Travis Green. They are one of only three teams (Edmonton - Stoll and Mike Peca and Nashville - Perreault and Greg Johnson are the other two) with two players in the top twenty faceoff men. In fact, Boston nearly has a third in Dave Scatchard who they traded to Phoenix in mid-November.

While it is a good thing to have a player or two who is very good on faceoffs, as it will earn your team a few more puck possessions, I would not be too concerned about it. You could just as easily find players who do not giveaway the puck or frequently take the puck from the opposition to more than make up this difference.

Faceoffs are not a particularly important statistic in the NHL because nearly every team is almost average. Even the best and worst teams are not that far from average. If we were to ever have a team that wins 80 or 90% of their faceoffs or loses all but 10 or 20% of their faceoffs, there would likely be a strong correlation between this success or failure and the team's success or failure, but it will be a very weak correlation at best when the difference between exceptional and average is roughly one faceoff in 25.