Faceoffs, and the inability of the Oilers’ two offensively gifted centres to win them consistently, have been a bit of a thing in Edmonton over the years. You don’t have to watch very many hockey broadcasts to see a team lose a faceoff and, as they do that dispiriting skate down the ice to retrieve a puck, hear some ex-hockey player explain that you have to win faceoffs on special teams because otherwise it costs you twenty seconds.
There’s a lot of stuff out there about faceoffs generally, with evidence suggesting that the impact of winning or losing one isn’t particularly large. I was curious about this specifically in the context of the Oilers and, even more in the context of special teams, so I gathered the data for all of the 5v4 situations in Oilers games between 2011-13 that started with a faceoff in the offensive zone for the team on the PP. A warning for those who may consider such foolishness: this is considerably more difficult than you’d think. Like, hours and hours of work more difficult. The NHL does not give up her secrets easily.
Really, I’m thinking about two questions here. The first question is this: is there a significant advantage that accrues to a team that wins an offensive zone faceoff at 5v4 or a defensive zone faceoff at 4v5? The second question, which is often forgotten, is this: assuming a significant advantage does exist, are there teams that win such a high percentage of their 5v4 or 4v5 faceoffs that they can turn this theoretical advantage into something tangible?
Just to explain that last question a bit: a penalty shot is one of the best chances in hockey. You’ve got a 30% shot at scoring a goal or so. That said, you’d be foolish to build an offence around generating penalty shots because there aren’t enough of them to live off of. That second question is kind of asking the same thing about 5v4 and 4v5 faceoffs – is there enough of a margin available in winning them to make it a big focal point?
Turning to the first question, I sorted the power plays in Oiler games in the 2011-13 seasons into four groups: Oiler PPs where the first faceoff was won, Oiler PPs where the first faceoff was lost, opposition PPs when the first faceoff was won and opposition faceoffs when the first faceoff was lost. What I’m trying to do here is isolate a single variable – winning that first faceoff – to see how it impacts on the numbers that a PP puts up.
There’s an awful lot of information packed into that little table but there are two points that I want to focus on. First, you can see that both the Oilers and their opposition did better when they won that first faceoff than they did when they lost it. Most of the time, the data should be confirming what we already know – we suspect that it’s better to win that first faceoff on the PP than it is to lose it and, lo and behold, that’s the case.
It’s a reasonably sizeable difference too; for the Oilers, winning that first faceoff resulted in a bump of 8.1 S/60 over 5v4 situations on which they lost it. For the opposition, winning that first faceoff added 9.9 S/60 to the volume of shots that a team will generate over the course 60 minutes of 5v4 time. It’s substantial and you can understand why teams care.
Except…how much can you reasonably expect to win that first faceoff at 5v4? How much marginal gain can you generate by being the best team in the NHL at it as opposed to, say, the worst? I assembled the 2011-12 faceoff data for the NHL at 5v4 and 4v5 to look into this. The NHL being the NHL, the data isn’t bang on – there appear to be some draws missing – but it gaves us a range to work with.
You can see that the best team at winning faceoffs in the NHL at 5v4 in 2011-12 was Colorado, at 57.5%. The worst was New Jersey, at 47.1%. A ten point spread. At 4v5, there’s a bit of a bigger spread – Calgary’s at 39.9% and Toronto’s at 55.9%. We’re talking about 5v4 here though, so we’ll use that spread.
We can calculate a 5v4 SF/60 number for the Oilers and their opposition under two different scenarios: if they win 57.5% of the faceoffs that start a 5v4 and if they win 47.1% of the faceoffs that start a 5v4. I did this and the results are as you see in the table at left.
The median team in the NHL in 2011-12 played about 6.9 hours at 5v4. If we assume 1 shot in 8 is going in, we aren’t quite at a goal’s worth of effect going from worst in the league to best in the league at winning that first faceoff on a PP. Overall, I’d estimate the difference between the best and worst teams in the NHL at winning faceoffs at 5v4 as being something below two goals. Two goals. If you go from one extreme to the other.
In other words, the difference is big between what happens when you win that first faceoff versus when you lose it but everyone is pretty similar in terms of how often they win it. Logically, rationally, you shouldn’t get too caught up in this. I’ve never seen anyone who’s been able to come up with a compelling competing argument. The Oilers have won 54.3% of their faceoffs at 5v4 the past few years; I doubt this is really an issue with them.
The other thing that caught my eye is more interesting and more subtle. Note that the impact of a faceoff win to start a 5v4 is larger for the opposition when they face the Oilers than it is for the Oilers against their opposition. It’s 1.9 SF/60 bigger for the opposition. This is interesting to me because it happens to click in nicely with some unpublished work that I’ve done on the Oiler and Shark PPs, the yin and yang of NHL 5v4 shot generation over the past six years. I’ve got a pretty strong suspicion that the Oilers are notably weak at this and that other teams are running set plays to generate shots off of faceoff wins.
If I was right, this would show up in shots generated immediately after an OZ faceoff win. Guess what? It does. Over the last two years, in the dataset I’m using, the Oilers have won 241 OZ faceoffs at 5v4. In the seven seconds following the faceoff, they’ve generated a total of 15 shots (amazingly, 7 of these have gone in). The opposition have won 273 OZ faceoffs at 5v4. In the seven seconds following the faceoff, they’ve generated a total of 38 shots. Seven of these have gone in. Expressed as a rate, Edmonton generated 32.3 SF/60 after a OZ faceoff win on the power play; their opposition generated 72.5. This is a problem.
What if the Oilers generated shots at that rate? The table at left summarizes what the Oilers and their opposition actually did in terms of shot generation at 5v4 in the data I’ve included in my dataset and what Edmonton’s numbers would look like if the Oilers generated 72.5 SF/60 in the seven seconds after an OZ faceoff win at 5v4.
You can see that the difference is pretty large – they go up by 2 SF/60. It’s twice the jump that a team would expect to get if it went from being the worst 5v4 faceoff team in the NHL to the best and it’s just for going from Oilers-level to what the opposition has done against the Oilers level, which is probably pretty league average. Over the past two seasons, you can attribute 40% of the gap between the Oilers and their opposition in terms of shot generation at 5v4 to the Oilers being lousy at generating shots in the seven seconds following OZ faceoff wins. It’s an astonishingly large amount to be attributable to such a small part of the game.
I’ve never done this but I thought it might be interesting to look at what a team that’s good at generating shots from faceoffs looks like. I’ve looked at this a couple of different ways and San Jose stands out as being exceptional at it. I found it striking when I watched their faceoff wins back to back to back etc. last summer so I thought I’d loop together what they’ve done on the faceoffs in my dataset for 2011-13. There were 15 faceoffs in the set – I cut a few that occurred in garbage time in that blowout loss in the Rexall opener this year and the odd one where they won a draw that hit a ref and was turned over or something. 10 of them are still here.
Here’s what you watch for. Watch how a forward, generally the forward on the wall, drops back as the faceoff is won to give the Sharks a third man high along the blue line. The defenceman on the wall skates into the middle and the defenceman who starts out in the middle shoots across to the wall. The Sharks end up with three guys spread out along the blue line. As you’ll see, it leads to shots awfully quickly.
Not all shots are created equal, of course, and the Oilers have been a high shooting percentage team over the past two years. They’re as subject to randomness as anyone, although I kind of suspect that they do create higher percentage shots. You would want to reason through whether you reasonably expect this to increase your goal output – the Oilers fiddle around and fiddle around and fiddle around to generate that easy chance off a cross-ice pass down low. They’re notably weak at generating shots off a faceoff win – it’s worth them exploring whether they’d benefit by implementing something like this into their 5v4 setup.
Anyway, conclusions: 1) We should all relax about winning and losing faceoffs because, although it makes a difference in the expected outcome, the difference between good and bad teams is tiny. 2) A more significant issue for the Oilers seems to be an inability to generate shots immediately following an offensive zone faceoff win at 5v4. You can see from the video how one of the elite teams in the league does it. Edmonton’s got a much slower, more deliberate setup. It’s questionable whether that makes sense although I don’t think that this evidence completely closes the door.
(A technical note on the dataset I worked with here. I didn’t include all 5v4 powerplays. Basically, it had to become 5v4 when the puck was dropped and the faceoff had to be in the offensive zone. There are narrow circumstances when that’s not the case – if you take a penalty after a goal is scored or if one team is short already and takes a penalty. I excluded those power plays from this analysis, on the sensible grounds that I was trying to isolate a variable that they did not contain – who wins the faceoff in the 5v4 team’s offensive zone to start a 5v4 situation?)Email Tyler Dellow at email@example.com