One of the strongest findings that hockey analytics has produced is that teams that have dominant power plays for an extended period of time are teams that generate a pile of shots. In the short run, a power play can live on shooting percentage but in the long run, teams that succeed on the power play (which I’m using to mean 5v4 here) are teams that generate a massive pile of shots.
Take a look at this table of the 18 teams that managed to produce at least 7.0 GF/60 over a three year span during the BTN era.
In particular, note the average SF/60 rate at 5v4 and the average shooting percentage. There are 120 three year team spans in this period and the average SF/60 of teams that scored at least 7.0 GF/60 would be the tenth best. The average shooting percentage of 13.5% is good – it would rank 17th – but not quite as good.
The worst SF/60 in there is Philadelphia’s 49.5, which is 50th out of 120. The worst shooting percentage, San Jose’s 11.3% over a three year span, is 93rd. If you want to be a team that scores at an elite level on the PP over an extended period of time, we can see that teams can do it with a lower shooting percentage. Nobody does it without being amongst the best at generating shot volume.
Which brings me to this season. There’s something interesting happening in the early going. Here’s a table with the 5v4 data league wide in the BTN era.
A couple of things jump out at me. First of all, the gains in 5v4 shots that came with the 2008-09 rule change that put the first faceoff of a PP in the defending team’s zone has kind of been ebbing for a few years. The leaguewide 5v4 shot rate was down to just 47.39 SF/60 at 5v4 last year. This year, it’s up about 10.6% to 52.4 SF/60. It’s early for individual teams but it’s not that early for the league – we’re almost 15% of the way through the NHL schedule. Goal scoring is up too, despite the shooting percentage falling.
The other interesting thing to look at is which teams are enjoying big jumps in their shot rates. I’ve put that in the table to the left. Shot rate has been reasonably consistent year over year – a correlation of 0.54 from 2007-08 to 2012-13. What really catches my eye in that table is the fact that there are five teams that have seen increases over 10+ SF/60 and three that have generated SF/60 increases of 15+ SF/60. The biggest increase that an NHL team has ever recorded in a season was 14.3 SF/60 for the 2008-09 Ducks, the year that the rule changed.
Only six teams out of 150 have managed a 10+ SF/60 increase since 2007-08. It’s early yet but to have five over ten shots better (and three over 15) does seem awfully unusual. I’ve no easy way of checking this but I’d be surprised if the Oilers have had some runs like this in the recent past. It’s hard to generate this many shots without it being a deliberate strategy, I think. There’s been an increased emphasis on generating shots in Edmonton, something that Dallas Eakins has talked about.
To bring this back round to what I opened with, about this being a tenet of hockey analytics folk, I’ll be very interested to see if it continues. I’ve no doubt that many NHL teams, more than we might guess, are poring over the data that’s been generated and, if they’re good and we’re good, arriving at some of the same conclusions.
They probably won’t talk about it publicly though. I certainly wouldn’t. The way we’ll find out that they’re changing tactics is by seeing it show up in the data. The rise in stolen base success rates in baseball is an example of what I’m talking about. People outside of the game produced compelling arguments that teams were making too many outs on the basepaths and these findings were either adopted or arrived at independently by people within the game. Stolen base attempts are lower than ever and the success rate is near an all-time high. Even if nobody comes out and says why this happened, you can see the impact of the logic in the data.
5v4 is kind of analogous to this because it’s about finding the sweet spot between the quality of shot that you generate and the volume of shots that you take. If a team only shot when a goal was certain, they might take three shots an hour – only shooting when they had an empty net, passing the puck into it. If a team shot every time they crossed the blue line, they might take 90 SF/60 but almost never score. It’s about finding that spot between volume and quality that maximizes the goals.
One final note. 28 teams have generated at least 55 SF/60 at 5v4 over the course of a season since 2007-08. 17 of those scored at least 7.0 GF/60 at 5v4, which is kind of the mark of an elite 5v4 team. Only two teams were truly bad at 5v4 (below 6.0 GF/60) when they accomplished this: Washington (a season with 5.6 GF/60) and Columbus (4.83 GF/60). The world isn’t a deterministic place but being a team that generates a ton of shots at 5v4 usually means you’ll end up with a good PP. It’s not certain but it’s probable. If you can’t handle a lack of certainty, you probably aren’t cut out for sports in particular or functioning in the world in general.Email Tyler Dellow at firstname.lastname@example.org