Let's talk about the penalty kill.
A season ago, the Caps finished 14th in the League in penalty-kill efficiency at 81.2 percent, which was up two spots from the season before, despite actually killing off penalties at a slightly lower rate than 2013-14's 82.0 percent. But where there was real improvement was in the shot rate the Caps faced, which pretty much went from historically awful to mediocre (the overall shot attempts-against didn't show as large an improvement, but with the team's focus on shot-blocking, perhaps we can chalk some of that up to a strategic preference).
The man-down unit had its ups and downs, part of which we can see by looking at the goals-against rate over the course of the campaign, with rolling eight-game samples:
The kill was pretty much hemorrhaging goals until around mid-December, then got better, with some hiccups the rest of the way. As points of reference, here's what the best (Minnesota) and 15th-ranked (Detroit) teams looked in this same metric over the course of the season:
Not bad, and the Caps actually finished in the top half of the League in goals-against rate at four-on-five. But the shot rate-against was a different story...
That chart shows unblocked shots towards the Caps' net and it got noticeably worse over time, especially at the end (though it wasn't without its good stretches). Again, for points of reference, the best (New York) and 15th (Edmonton) teams in the League: So the goals-against rate looks pretty good while the Fenwick-against pace doesn't, and you know why that is...
Braden Holtby. Holtby's four-on-five save percentage was terrific. Layer it on top of that goals-against chart and it looks like this:
Generally, when Holtby's save percentage was up, the goals-against rate was down and vice versa (duh). Unfortunately, shorthanded save percentage is notoriously variable, so rather than rely on great goaltending to save the penalty kill, teams really need to work on reducing shot volume against, which brings is back to, well, shot volume against.
The big question here is why did the Fenwick-against rate move the way it did (i.e. steadily upward)? One factor to consider would be distribution of ice time. Here's how the percentage of shorthanded ice time was divvied up among the defensemen who got meaningful time on the kill:
After spending the first five months of the season giving significantly more minutes to the Brooks Orpik-John Carlson pair, Barry Trotz started leaning on Karl Alzner and Matt Niskanen more in March and by the end of the season, that duo was getting more ice time (note that Orpik did miss four games in mid-March). Put that chart on top of the Fenwick-against rate and...
The big spike at the end of the season moves with the increase in the second pair's ice time and the decreased reliance on the top pair. Interesting... but is it correlation or causation?
Looking at earlier trends, the shot-rate increase in November correlates to an increase in ice time for Orpik. The increase in December looks like Niskanen. The increase in February looks like Carlson/Orpik. As for decreases, the November and December decreases look like the move with increased ice time for Alzner and Niskanen, and the January decrease Orpik/Carlson. And the February decrease coincided with a rise in ice time for Niskanen and Alzner... which continued and led to the spike at the end. So there's no real "there" there, especially when you consider all the factors we haven't looked at here - quality of opponents, the forwards, frequency of penalty kills, etc.
What about the forwards? Here are the guys that averaged more than a minute per game on the kill:
Put it on top of FA rate and...
That's a pretty crowded chart, but the spikes look like they go along with Brooks Laich and Troy Brouwer getting a lot of ice time. Jay Beagle, too. Things look better when Joel Ward and Eric Fehr were getting decent time (which, of course, isn't going to be happening any more). But, as was the case with the defense, there's no "a-ha!" moment in terms of deployment (and if there is, it's probably "don't play Alzner, Niskanen and Laich too much... which is unfortunate, given that those are probably three of the team's five most likely penalty kill ice time leaders).
There's another factor to look at here, and that's face-off percentage. Conceptually, if a team's getting killed in the dot on the penalty kill, they're likely missing out on some relatively easy clear opportunities and will pay the price for that before long (and the numbers show that it takes upwards of 30 seconds off a lost face off for a power play to reach the same shot rate as if it had won that draw). Here's what the Caps' penalty-kill face-off percentage looked like over time:
So what's the conclusion? Well, there really isn't one. There's no clear culprit for the Caps' penalty kill's concerning rate of shots allowed in either deployment or face-off percentage, which leaves the usual suspects - the system and the personnel. As between those two... good luck apportioning blame.