Shall we talk about the Capitals’ power play?
Yes, let’s do that.
For the four-year stretch beginning with the lockout-shortened 2012-13 season through the League’s last full campaign, the Caps’ power play converted at a far-and-away League-best 24.1 percent (three full percentage points better than the Penguins), lighting the lamp 227 times (nearly ten percent more than the second-place Flyers, with 84 of those tallies coming of Alex Ovechkin’s stick, mostly from that familiar spot in the general vicinity of the face-off dot to opposing netminders’ right) and generating shot attempts at the third-highest clip on the circuit. Those seasons occupy three of the top six and four of the top 13 slots in efficiency over the span, including the top two, both clocking in at better than one-quarter proficient. That’s likely historically dominant.
Thanks, Adam Oates. (No, seriously.)
But so far this season, the Caps have sputtered a bit with the extra man. They’re still generating shots at a (very) high rate, but have only converted five times (once at four-on-three) in 38 opportunities, a paltry 13.2 percent, 24th-best in the League.
So the answer to “what’s wrong with the Caps’ power play?” would seem pretty clear: bad luck. Further to that point, Corsica Hockey has the Caps’ expected Goals-For (xGF) at five-on-four at 8.42 goals (whereas they’ve actually scored just four times in that most common manpower advantage). Add in the four tallies we’d reasonably expect the Caps to have scored that they haven’t, and suddenly they’re at 9-for-38 on the power-play, a 23.7 percent clip that is very much in line with the productivity to which we’ve grown accustomed.
Small sample results explained away by random variance that will almost certainly regress as the data set increases. Tidy, no?
Well, maybe not.
The Caps’ current 12-game/13.2 percent slide doesn’t look a whole lot better if you quadruple the sample to 48 regular-season games (which ignores the eight power-play goals they dropped on the Flyers in the first three games of the playoffs and the subsequent 5-for-33 the rest of the way), a span over which they’ve scored on just 14.8 percent of their power-play chances (and one of those 22-for-149 was into an empty Nashville net).
That 48-game sample dates back to a somewhat arbitrary-but-convenient point in the 2015-16 schedule – the All-Star break and the two postponements due to snow that preceded it and left the Caps with a week of mostly hockey-less days in between games. Prior to that break, the power-play had gone an eye-popping 38-for-140 (27.1%), nearly twice the success rate they’ve had since.
Random variance and regression in action? Sure. To an extent. But let’s take a closer look at the top power-play unit of Ovechkin, Nicklas Backstrom, John Carlson, T.J. Oshie and Marcus Johansson. (As a proxy for this PP1, we looked at the five-on-four power play with Ovechkin and Backstrom on the ice together, which is imperfect in that it includes brief periods in which Johansson was replaced by Evgeny Kuznetsov or Jason Chimera (?), but it’s nonetheless informative and certainly close enough for our purposes.)
In the first segment (Segment 1: 46 games, through January 19, 2016), the Caps’ five-on-four power play generated 125.9 shot attempts per hour of ice time with Ovechkin and Backstrom on the ice; in the 48 games since (Segment 2), that number has dropped to 111.2. That’s an 11.7 percent fall-off, but 111.2 shots per sixty is still a very effective shot-producing power play. It’s hard to pinpoint a culprit here, but one area of inquiry would certainly be zone entries – if the Caps are having a bit more trouble getting the puck into the offensive zone under control, it would stand to reason that their shot rates would be impacted.
Another reason for a lower shot rate could conceivably be increased selectivity in shots, either by design or necessity. To that end, let’s take a look at who’s shooting in each of our segments:
What you’re seeing there are individual shot attempts (per sixty) with Ovechkin and Backstrom on the ice at five-on-four and the corresponding percentages. Substantively, what you’re seeing is a 16 percent reduction in Alex Ovechkin’s shot share (probably not ideal), big reductions (in already relatively small contributions) for Nick Backstrom and Marcus Johansson, and increases of 14 percent in John Carlson’s share of those shot attempts and just over 18% in T.J. Oshie’s share (which is a good thing, all else equal… which it isn’t). Essentially, the Caps’ shot rate has dropped off a bit and the distribution has also shifted. Are teams doing a better job of taking away Ovechkin’s shot? Sure looks like it. And while the uptick in Oshie’s chances is good, the Caps would probably prefer that “Ovechkin slack” to be picked up by Johansson (a 20.7-percent shooter at five-on-four) and Backstrom (13.2 percent) to Carlson (6.1 percent). The Caps’ power play has indeed evolved from relying on Ovechkin to relying on the threat of Ovechkin, but increasing the volume of the lowest-percentage shots isn’t a great recipe for “survival of the fittest.”
Now, we’re obviously talking about subsets of relatively small samples here – a handful of bounces one way or the other and these numbers tell a different story. But the take away here is that the Caps’ power play is slumping (as opposed to broken), but there may be contributing factors here beyond what’s random or otherwise inexplicable. The Caps have too much talent and are generating enough shots that they’ll break out of this slump, but improving zone entries (perhaps) and focusing on who’s taking shots (likely) will help them make those breaks, as it were.
One last point – during this 48-game slide, the Caps are just 5-for-50 (10.0%) on the power-play against Metropolitan Division teams (17.2% against others) and 7-for-75 (9.3%) against Eastern Conference teams (20.3% against the West), so don’t discount the impact that familiarity may be having on the Caps’ power-play success. It’s been one hell of a run… hopefully it’s not coming to an end.
(Thanks to Muneeb for pulling the data that isn’t otherwise attributed via links or text.)