Photo: Patrick McDermott

The Washington Capitals are a dangerous team when they’re trailing. We saw this twice last weekend as the Caps erased multi-goal leads by both the Predators and the Stars. And though they lost both those games, we learned a lot about how the Caps can perform when facing adversity.

When down a goal, the Capitals are the fourth most aggressive team in the league, possessing 59.7 percent of shot attempts. That’s a big jump from tie games, when the Caps hold a still-respectable 10th place shot-attempt percentage with 52.4 percent.

But when the Caps manage to get the lead, which they do most of the time, their possession drops to 46.9 percent, a stark drop-off, and a middling 13th in the league. That partially explains why the Caps rarely win by more than two goals.

All teams do this to some extent, but the Capitals’ meekness with the lead has been one of their noted weaknesses— even during that torrid winning streak last month.

Coaches often direct their teams to become less adventurous once they gain the lead, which is a primary factor in relatively lower possession. Barry Trotz does that to some extent (you can see it in the forecheck), but his personnel choices based on score are playing a role as well.

Below is a graph of how Caps forwards’ ice time gets distributed depending on the score. The blue bar is when the Caps are leading, the gray bar is when the teams are tied, and the red bar is trailing. So if the red bar is bigger than the blue bar, that player is getting used more when the Caps are trying to mount a comeback. My observations follow.

• Barry Trotz uses Jason Chimera, Jay Beagle, and Troy Brouwer relatively more when the Caps are protecting the lead (2.3 percent, 1.9 percent, and 1.4 percent more, respectively). Is that because they’re more defensively stout or because he’s resting his weapons?
• Because, yes, Alex Ovechkin and Nick Backstrom get much less ice time once they’ve secured the lead for the team (which they usually do on special teams, not at 5v5). Backstrom’s ice time drops 1.3 percent from tied and Ovechkin drops 1.6 percent.
• When mounting a comeback, it’s the young guys and part-time second liners who see the biggest change compared to tied games. Andre Burakovsky‘s ice time jumps up 4.0 when the Caps are down, and Evgeny Kuznetsov jumps 2.8 percent. Next is Alex Ovechkin, who scored three goals while trailing over last weekend alone. He’s not bad, that guy.
• For all his fondness for them, Barry Trotz shortens the bench for some of his meat-and-potatoes guys when the Caps are losing. Jason Chimera and Jay Beagle, along with Laich and Ward, see drop-offs when the Caps are down in the score (just 0.9 percent for Chimera, but 3.2 percent for Ward).
• But no one’s ice time drops more than Eric Fehr when the Caps are losing. He sees a 4-percent decline in ice time compared to leading.
• Tom Wilson looks like an anomaly in this and all the other data in this report, perhaps due to his smaller number of games played, his jumps up and down the lineup, and the preponderance of time he spends in the box.

Now, if you’ll indulge me, I’d like to isolate Trotz’s usage of those players from how those players perform in the various circumstances. The question I asked myself went something like this: Are there any players who do markedly better or worse based on the score? And do those improvements or declines happen on offense or defense?

Offense by Score

The table and chart below show the number of shot attempts the Caps get per 60 minutes of 5v5 based on the score of the game.

And for our visual learners:

And here’s that same data expressed as percentage change ([leading – tied] / tied and [trailing – tied] / tied).

• For starters, I’ve got not idea what’s going on with Evgeny Kuznetsov. When he’s on the ice with a lead, which isn’t much, the offense goes up a whopping 13.9 percent. Maybe the circumstances of Kuzya’s scratches play into it? It doesn’t appear to be sample size alone, but it’s such an outlier than I’m skeptical.
• Andre Burakovsky could be another anomaly. I don’t think we should draw too many conclusions off him except that Trotz doesn’t use him– or the other young guys– much.
• Onto the less wild part of the spectrum. When the Caps are losing, Alex Ovechkin‘s and Nick Backstrom‘s offense drops off the least after Kuznetsov and Troy Brouwer. Whatever his faults, Brouwer might have a good skill set for comebacks.
• I’m hesitant to single people out (or double people out in this case), but, when the Caps have the lead, Jay Beagle and Jason Chimera generate offense only slightly better than the Buffalo Sabres (39.6 shot attempts per 60). A lead will seldom grow when they’re on the ice, so their primary value should be in preventing shots against.

On that note, onto the defense.

Defense by Score

This is the same calculation and visualization, just with the opponent’s shot attempts per 60 instead of the Caps. Think of it as shot suppression; lower numbers and shorter bars are now preferable.

And the chart version:

And the percentage change compared to tie games:

• When the Caps have the lead and Jay Beagle and Jason Chimera are on the ice, opponents shoot at a pace (61.9 shots attempts per 60) that is higher than the most offensive team in the league, the Chicago Blackhawks.
• That’s the lede, but it’s not the whole story. Chimera seems dependable as hell in a comeback effort– reducing opponent shot attempts from 51.8 to 37.9 when the Caps are down compared to a tie score. That could make for a sturdy checking line when you’re resting your main attackers for another assault.
• And even Chimera’s numbers aren’t as sterling as Joel Ward, who fully earns his reputation as a solid two-way player. He cuts the opponent’s attack down by 42 (42!) percent when the Caps are down and only allows a 2.7 percent increase when the Caps are protecting a lead. That’s stellar.
• But the stingiest Caps player when the Caps are up, and this will not surprise you, is Nick Backstrom.
• I’m on the record championing the Ovechkin – Backstrom – Fehr line, but those three players allow the most opponent shots when trailing in the score. This might inform Trotz’s reluctance to use that line since October. (Then again, those three players also generate the most offense when trailing.)
• The guy you gotta worry about when the Caps are in the lead is Jay Beagle, who allows a scary 37.3-percent increase in opponent shot attempts when leading and only a 16.7-percent cut when the Caps are trailing. I wonder how much that colors Alex Ovechkin’s numbers given the 77 minutes of 5v5 they’ve shared. It seems to me that Trotz might prefer Beagle to play unadventurous and sober hockey in all situations– despite some proof that it’s not helpful.

This was an open-ended study without a hard agenda, but I think I might’ve accidentally come up with some instructive conclusions.

First and foremost, winning lots of games by one goal is not the foundation of a championship season. The Caps need to prove they can blow teams out, and that requires keeping one’s metaphorical foot on the metaphorical gas once you’ve got the lead.

But if you’re trying to come back, you want your European forwards to get a lot of opportunities to attack. That means big offensive-zone shifts for Ovechkin, Backstrom, Kuznetsov, and Burakovsky. A top line of Ovechkin, Backstrom, Fehr would likely generate the most Caps shot attempts, though they’d also allow the most by opponents as well. That’s probably worth the trade-off.

When trailing, Jason Chimera and Joel Ward will do a splendid job not letting the other team run up the score, though they’re not being used that way now. Once you get that lead back, it might be wise to shorten the bench lest Chimera or Jay Beagle allow a shooting gallery.

There might be a lot more wisdom to mine from this, though it’s also possible that unseen biases are corrupting these data. Let me know below if you see anything I missed.

Notes: All stats are current as of January 14th. For changes in ice time percentage compared to tie game, I used absolute percentage change, not relative. For changes in shot attempts / 60 based on score, I used relative percentage change. I’ve kept all the graphs sorted alphabetically for better comparison among them.