clock menu more-arrow no yes

Filed under:

Leading Indicators: Personnel, Tactics and the Washington Capitals

New, comments

Taking a closer look at what's going on when the Capitals lead by the slimmest of margins

Winslow Townson-USA TODAY Sports

Perhaps you've noticed - the Washington Capitals have had some trouble holding leads lately. They led in every game in their recent five-game losing streak only to, you know, not win. In fact, they've led in every game they've lost this season, with the exception of the shootout loss to San Jose. Part of the problem has been goaltending, but that doesn't tell the full story.

The Capitals are still pretty much an average possession team when they are leading by a goal, but that's quite a drop from their dominance when the game is tied or they're trailing. By looking at the underlying numbers we can identify some of what the Capitals are doing differently when they have the lead versus when the game is tied, and why those differences may be contributing to the team's struggles.

Here's a look at the rates of which the Capitals are both allowing and taking shot attempts per 60 minutes during five-on-five play when the game is tied and when up by a single marker:

Score State Fenwick For / 60 Fenwick Against / 60 Corsi For / 60 Corsi Against / 60
Tied 43.7 35.7 58.8 51.3
Leading by 1 33.0 36.4 47.0 51.3

Note: A Fenwick attempt is any unblocked shot attempt while a Corsi Attempt is any and all shot attempts.

The Caps are still doing a good job of suppressing shot attempts when they are leading (ninth in the League, a few ticks below their fourth-overall ranking when tied), but their offensive production drops off the table, from eighth to 25th. That gets us closer to identifying the issue(s), but what's driving this dramatic drop-off (both raw and relative) and in offensive production?

One of the best ways to visualize multiple 'advanced' hockey stats at once is through the use of player usage charts (first created by Rob Vollman). Thanks to the good folks at war-on-ice, we can look at these charts in various game states, including when the game is tied and when the Capitals are leading by one goal. Let's start by looking at how Barry Trotz and his staff are using their players when the game is tied (note: both of the charts and all analysis are based on five-on-five play through November 8):

11-11-14 Tied game

The Capitals, as a team, have an offensive zone start percentage of 51.1% when the game is tied. Offensive zone start percentage (or, more accurately, the percentage of non-neutral zone face-offs that a player starts in the offensive zone), is displayed on the X-axis. When the game is tied, no Caps forward has a larger percentage of his non-neutral zone draws in the offensive zone than Andre Burakovsky. A high percentage of offensive zone starts may indicate that a coach doesn't have faith in the defensive or face-off abilities of the player in question and/or the coach wants to put the player in a position where his ability to create offense are most likely to be leveraged (for Burakovsky, it's probably a bit of both). The Y-axis measures the quality of competition that an opponent is facing (as measured by opponents' ice time) - the higher on the Y-axis the tougher the competition. So by using the chart you can see that Brooks Orpik faces the toughest competition of any Capitals defensemen, while opponents are putting out their best players to try to stop the Caps' big guns.

Before going into too much depth about how the team is deploying it's players in tie games, let's take a look at the usage chart for games that the Capitals' are leading by one and then compare the two:

Leading 11.11.14

Before making claims based on the two graphs together, it's important to note how dramatically different the scale of the X-axis is here and for good reason - when leading by a single goal, 59.0% of the Caps' non-neutral zone face-offs take place in the offensive zone, the highest percentage in the League through Tuesday's games. Yet despite all of these offensive zone starts, the Caps are struggling to generate shot attempts (and were flat out awful in terms of goals-for when up by one before dropping three up-one goals on Columbus the other night).

So which players are being deployed in a (relatively) different manner than they were when the game was tied? Time on ice is a good place to start, and here's a look at the distribution for defensemen:

Defense Tied and +1 TOI

The percentage only includes games in which the player has played.

Mike Green and Nate Schmidt have been the Capitals' best defensive pairing in terms of puck possession and they both see a larger share of ice time when the team is leading than when the game is tied. (Bet you didn't see that coming.) Whether this change in time on ice is due to the coaching staff wanting to produce another goal, keeping Orpik and John Carlson fresh or some other rationale (or no rationale whatsoever), giving Mike Green more ice time shouldn't hurt the team, generally. Based on the above player usage charts, Schmidt and Green start a large portion of their shifts in the offensive zone, so there's no reason to expect a significant drop-off in-shot attempt generation. And yet...

[Okay, a brief aside regarding that motion chart. Here's how it works - the slider goes from tied ("1900"; Google requires dates for the sliders) to Caps Up-1 ("1901"). So click on a bubble or select a box in the lower right, hit play in the lower left, and watch how your selected player's (or players') performance changes between the two game states in terms of Corsi-For rate (x-axis) and Corsi-Against rate (y-axis). Color represents ice time, bubble size Corsi-For percentage. Oh, and the star is thrown in there to make the axis scale look right. Carry on.]

When the game is tied the Capitals generate around 23 shot attempts per 20 minutes of five-on-five play when Mike Green is on the ice; when they are up by a goal that rate drops all the way to 15. While not to the same extent, every Capitals' defenseman sees a decline in their on-ice shot attempt production. Generally speaking, teams will play more conservatively with a lead and will trade a reduction in shot generation for an increase in shot suppression - they don't need another goal to win, so the less likely it is that another goal will be scored, the better; traps happen, low-event hockey then rules the day, and so on. But for the Caps, as we saw above, they may be giving up too much offense for what they're getting in terms of shot suppression.

But again, what's driving the reduction in shot generation? Perhaps the Caps are relying more on their third and fourth lines when up a goal, which logically would result in decreased production. Let's take a look at time on ice changes:

Forward TOI by score differential

Well it turns out that's not the case. Just like with the defense, all of the shifts in player usage make sense. The Capitals' best players are seeing more ice time when the team is defending a lead. Alex Ovechkin and Nicklas Backstrom have had a solid start to the year, and despite being a healthy scratch for a few games this season, Eric Fehr and Brooks Laich appears to be go-to guys for Trotz when they're in the lineup. On the flip side, Evgeny Kuznetsov and Tom Wilson haven't yet proven themselves trustworthy, so they see less ice (Marcus Johansson and, surprisingly, Joel Ward a bit as well).

And how have the forwards performed in the different states?

Laich, again like (almost) every other forward, sees a drop off in his on-ice shot attempt generation... what makes Laich of note is that he experiences an even more significant drop off in the rate of attempts against than attempts for. Admittedly, that would be a little more impressive if his CA/20 was not so high when the game was tied, and the sample on Laich is even smaller than the already small sample for other players, due to injury.

But if we focus in on a few elements from this graph and both of the player usage charts, two players really stand out. Burakovsky and Johansson rank first and second on the team in terms of offensive zone start % when the game is tied. When the team is up by one things are very different, and they appear to be having an effect on the duo's possession numbers. Burakovsky sees a drop off in CF/20 that is as significant as the one experienced by Green, while Johansson sees a significant drop off in both his CF/20 and an increase in his CA/20. The struggles of Burakovsky and Johansson are apparent (and somewhat explicable), but they don't provide a satisfying answer as to why the team is struggling to generate shot attempts.

Given the above, the Caps' decrease in offensive production while leading by one seems to pretty clearly be an issue of tactics and not personnel deployment. They'd likely see an increase in attempts if they deployed Burakovsky and Johansson in the offensive zone more often, but perhaps the lack of shot attempts is partially by design. It's possible that the Capitals are playing with a lead similarly to how some teams play when they are shorthanded. While based primarily on visual evidence rather than data, it appears that, with a few exceptions, the Capitals at times prefer to simply retain possession of the puck, rather than generate shots when they enter the offensive zone. That strategy makes sense at the theoretical level - if you have the puck the other team can't score (it's almost like puck possession matters). So, has this strategy (if it is indeed the strategy) limited opponent shot attempts?

Yes, yes it has. If we go all the way back up to that first table (all the way back at the top), we can see that the Capitals are allowing the same rate of shot attempts in each of the different score states... which is impressive, even if somewhat counter-intuitive - ideally, you'd hope to suppress more shots when playing with the lead. But that's not the reality of score effects, protecting leads and such. Here's a look at how this year's Capitals' shot attempt rates change when they lead by a goal versus when the game is tied compared to the average changes seen across the League last season.

Entity FF/60 Change FA/60 Change CF/60 Change CA/60 Change
14-15 Capitals -10.70 0.70 -11.80 0.00
13-14 League Average -2.51 2.40 -4.52 4.12
13-14 Top 4 FF% (Tied) -2.05 4.93 -4.93 4.88

Note: Top 4 FF% Teams: Los Angeles, Chicago, San Jose, And New Jersey

So, while the Capitals are doing a very impressive job limiting shot attempts, it's coming at a high cost. The drop-off in attempt generation is staggering, especially when you recall that Ovechkin, Backstrom, and Green all see a larger percentage of ice time when the team is leading than when the game is tied and the Caps are getting all of those offensive zone starts.

Thanks to a piece that Matt Pffefer wrote for Hockey Prospectus we know that teams shoot at a higher percentage when they are leading, and the Caps are putting their top offensive players on the ice... but they need to take shots in order to score (a little luck wouldn't hurt either - the Caps are in the bottom-third in up-one shooting percentage). Up to this point in the season the Capitals have fallen into an extremely defensive posture when leading (and have had plenty of practice doing it), and doing so has cost them. It's now up to the coaching staff to identify that a change in tactics is necessary in order to help this team secure better outcomes.