The Capitals’ latest stretch of poor play has seen them lose five out of seven games, with three of those five losses coming against contending division rivals, and the two wins needing improbable comebacks. They’ve yielded the game’s first goal in six straight games, given up 29 goals against overall in those seven games (and 20 of those goals at five-on-five).
It’s a stretch that has fans and, we can only assume, Caps’ brass scratching their heads.
So just how bad has the Caps’ team defense been? Let’s use Natural Stat Trick’s expected goals model to analyze this over time.
The thing that jumps out here is that Reirden’s entire tenure has been more porous than any of those that preceded him. Some of this can be chalked up to a more open style of play, which we get into a little bit in this post from last week — and we’ll step aside for now from the question of whether “more open” is a desirable, or even purposeful change. But it bears noting that the current run of play for the Caps is one of the worst defensive stretches they’ve had since it’s been possible to measure this sort of thing in this manner, with most of the other contending stretches also occurring during Reirden’s tenure.
Well, with one very notable exception…
On February 17th 2018, the Chicago Blackhawks steamrolled the Capitals 7-1, and their fifteen-game rolling xGA/60 hit a highwater mark of 2.92. Ten days later, one author from this website, not going to name any names, advocated for major change. The point being, we don’t have to reach too far into the past to extract a lesson about not overreacting to performance lulls at this point of the season. Trotz’s Caps went on to hoist the Stanley Cup fifteen weeks later.
The difference here, of course, is that the sustained defensive lapse was an anomaly during the Trotz era; it frighteningly resembles the new normal under Todd Reirden.
Let’s dive in a little deeper. After all, a model is only a model, and we’ve written plenty about how these expected goal models never seem to be a particularly good fit for evaluating the Caps. So here’s a look at how the model and reality have looked over the same time period.
It is admittedly difficult to make heads or tails from this, other than to note that the two datasets do share a trend, suggesting that perhaps the xGA side of the xGF% formula is a better fit than the xGF side, which jives with what we know about these models failing to capture the Caps’ exceptional shooting talent. The salient piece from this plot, however, are where the two trends diverge from each other, and in which direction they do. Here’s a plot to drive the point home.
Notice that during Barry Trotz’s tenure, visualized in orange, there are several long stretches where the values are negative. This indicates that the team made a habit of allowing fewer goals than expected, with fits with Trotz’s profile of being a defensively minded head coach. Todd Reirden, visualized in blue, has not had such good fortune. Not only are his troops expected to allow goals at a rate that exceeds almost any from the last thirteen years, but they’re also in a constant state of underperforming that expectation.
There are a lot of factors that go into a big change behind the bench, but history shows us that poor defensive performance is likely high on the list of considerations. It’s difficult to say what proportion of culpability should be doled out to Reid Cashman, and what rolls up by default to Todd Reirden. That, and any attendant personnel actions, are for Brian MacLellan to decide. But if the defensive performance doesn’t round into shape soon, in the words of Jonny Lang, there’s got to be a change. Things just can’t stay the same.