No team had a better January than the Washington Capitals, going 9-1-4 to lead the Eastern Conference. Their Saturday night overtime loss to the league-leading Winnipeg Jets was fun but telling: despite a thrilling comeback effort, they didn’t size up particularly well.
Since our last snapshot, the Caps have a 0.767 points percentage (second, behind the Islanders), but they’re 25th in shot-attempt percentage and 21st in expected-goals percentage. They’ve got a lot of leverage out of 10 power-play goals (6th place) and allowing only 7 goals while shorthanded (8th). That’s one way to win hockey games, but it’s not the most dependable if you want a deep playoff run. Logan Thompson is amazing, but you don’t want to rely on him to gift-wrap a couple round wins on his own.
A few weeks ago I wrote that the Capitals were in first place …and a bit of trouble. But they won four of their next five after that article, so what do I know? I obviously don’t know what will happen next, but I do know how they’ve been so successful lately, at least superficially.
Washington is getting elite goaltending (96.4 percent during five-on-five in the last ten games, absurd, LT for Vezina) and generally finishing their chances (sixth best shooting percentage during five-on-five play). Some of that is repeatable but most won’t be. The Caps won’t keep outscoring opponents 21 to 8 forever if they’re only outshooting them 21.5 expected goals to 21.3.
So thank you, LT, but it’s time for the skaters to improve. This is the snapshot. The tables below show the team’s statistics while each player is on the ice during five-on-five play. Interesting points are highlighted for discussion. A glossary follows.
Forwards
| Player | TOI | SA% | xGF% | GF% | PDO |
|---|---|---|---|---|---|
| Frank | 106 | 54.5 | 61.6 | 77.9 | 1.07 |
| Dubois | 722 | 53.2 | 57.5 | 62.9 | 1.04 |
| Eller | 366 | 53.1 | 49.4 | 50.3 | 1.00 |
| Protas | 715 | 52.9 | 50.2 | 63.9 | 1.06 |
| Mangiapane | 581 | 52.2 | 53.4 | 60.1 | 1.02 |
| Raddysh | 617 | 51.6 | 55.2 | 51.9 | 1.01 |
| Wilson | 688 | 51.2 | 55.0 | 57.8 | 1.02 |
| McMichael | 699 | 50.3 | 54.7 | 61.0 | 1.03 |
| Strome | 693 | 49.8 | 49.6 | 62.6 | 1.05 |
| Dowd | 645 | 49.2 | 51.7 | 58.8 | 1.03 |
| Ovechkin | 440 | 48.6 | 45.1 | 68.9 | 1.08 |
| Duhaime | 577 | 48.3 | 49.4 | 55.3 | 1.02 |
| Vrana | 229 | 45.3 | 39.9 | 65.6 | 1.10 |
Defenders
| Player | TOI | SA% | xGF% | GF% | PDO |
|---|---|---|---|---|---|
| Carlson | 843 | 55.2 | 56.1 | 57.3 | 1.01 |
| Fehervary | 802 | 51.3 | 50.9 | 50.7 | 1.00 |
| Sandin | 879 | 51.2 | 56.5 | 60.0 | 1.03 |
| Chychrun | 798 | 50.9 | 47.3 | 63.3 | 1.05 |
| van Riemsdyk | 808 | 49.1 | 48.7 | 63.3 | 1.06 |
| Roy | 713 | 47.8 | 49.6 | 60.0 | 1.04 |
Glossary
- TOI – Time on ice in minutes. Only five-on-five play is included here.
- SA% – Shot-attempt percentage. The share of total shots attempted by Washington while the skater is on the ice during five-on-five play. 50 percent means even.
- xGF% – Expected goals percentage. The share of expected goals generated by Washington while the skate is on the ice during five-on-five play. Expected goals weighs how likely to become a goal each attempted shot is. 50 percent means even.
- GF% – Goals-for percentage. The share of total goals scored by Washington when the skater is on the ice during five-on-five play. 50 percent means even.
- PDO – The sum of Washington’s shooting percentage and saving percentage when the skater is on the ice during five-on-five play. One means league average. The acronym doesn’t stand for anything, and yes, I hate it.
Notes
- There’s a lot going on, so I’ve kept a few players off the snapshot: Miroshnichenko (no games since early January), Lapierre (who will have to fight to get up in the bigs again), and McIlrath (no games since before Christmas). And I’ve added Ethen Frank, who is very interesting. Frank hasn’t exactly been battle-tested yet (9:45 per game), but the early returns are positive – outscoring opponents 5 to 2.
- Almost exactly upon the new year, Dylan Strome crashed out. After great underlying play to start the season, he’s been around 45 percent in on-ice shot attempts in January. Fifty percent would be even – i.e. equal possession for the Caps and opponents – but we’re used to seeing him well above that.
- Strome was playing between Ovechkin and McMichael before and after the sharp part of that drop, and he had been centering a few variations (Mangiapane and Wilson, McMichael and Raddysh) in the weeks leading up to it. If I had found that one player was obviously sinking him, I would have written about it by now. Maybe it’s Strome himself, though the full-season with-or-without-you (WOWY) data doesn’t suggest that. Here are on-ice shot-attempt percentages for when players are on the ice with or without Strome; again, above 50 percent is good.
| Partner | Together | Without Strome | Diff |
|---|---|---|---|
| Ovechkin | 49.0 | 46.7 | +2.2 |
| Protas | 53.3 | 52.7 | +0.6 |
| Chychrun | 47.0 | 52.7 | -5.7 |
| Sandin | 51.1 | 51.2 | -0.1 |
| Carlson | 54.5 | 55.5 | -1.0 |
| McMichael | 47.9 | 51.5 | -3.6 |
| van Riemsdyk | 45.5 | 50.5 | -5.1 |
| Wilson | 52.6 | 50.8 | +1.8 |
| Roy | 47.3 | 48.0 | -0.7 |
| Fehervary | 53.8 | 50.6 | +3.3 |
| Raddysh | 56.0 | 50.8 | +5.2 |
| Mangiapane | 47.4 | 53.1 | -5.7 |
- The poor performance with Chychrun and TVR is apparent. Mangiapane didn’t fill in well with Strome, but Raddysh did. McMichael probably is impactful elsewhere. But overall, I won’t pretend to make perfect sense of this. Tactics or injury feel more likely as explanations than personnel. Someone needs a camel ride.
- One thing I wish I could do but can’t: get skating speed data over stretches of time. I’d love to know if, all of a sudden, a skater dropped by half a mile per hour. I’d love to know if someone like Strome (a hard skater but without a high top speed) has further lost his top gear over the past few months. NHL Edge has that information, but it’s not readily available and it’s not broken down by game. That used to be the same situation with shot-attempt data – where you could only get season-to-date stats if you knew the arcane querystring parameters of behindthenet.ca. Not having reliable game-by-game data was why I started the snapshot in the first place – so I could follow progress.
- I’m so grateful to have that attempt-level information now, so we can see how the team is changing without me having some byzantine spreadsheet. Here are the on-ice expected-goal percentages for forwards over the last three months. Higher is better.
| Forward | Nov | Dec | Jan |
|---|---|---|---|
| Dubois | 53.0 | 59.9 | 58.2 |
| Eller | – | 47.3 | 55.6 |
| Mangiapane | 50.9 | 53.2 | 54.3 |
| McMichael | 53.2 | 52.3 | 52.6 |
| Wilson | 50.4 | 54.7 | 51.6 |
| Protas | 37.1 | 59.4 | 50.5 |
| Raddysh | 56.4 | 53.8 | 50.3 |
| Strome | 43.0 | 51.3 | 43.4 |
| Ovechkin | 40.9 | 41.2 | |
| Dowd | 54.5 | 54.8 | 40.9 |
| Duhaime | 50.2 | 50.7 | 40.6 |
- One thing I didn’t see coming: Lars Eller has settled in. He had posted a 47.3 percentage of on-ice expected goals in December, but hit his groove in January with 55.6. (Have I mentioned yet that Jakub Vrana hasn’t got a sweater since January 6?) I get that there was a lot of consternation about Eller, and he’s still not producing – two goals and two assists in 14 January games – but I don’t think there’s a strong case that he’s dragging the team down.
- I’m not doing the Pierre-Luc Dubois for Selke bit in this month’s snapshot, but here’s a list of every forward who had a better on-ice five-on-five actual-goal differential in January: Anders Lee and Mathew Barzal. They (common linemates for the red-hot Islanders) were plus-15 and plus-13, respectively. PLD was plus-11. He’s Washington’s second-best acronym-able player and overall best skater this season.
- I got this sassy comment earlier in the month, lightly formatted by me:
Peter, you post numbers without context, as per usual, try to say so-and-so players are performing poorly without understanding why or trying to explain why. Fehervary being under 45 percent is ok because his zone starts and offensive zone faceoffs are a lower percentage than his xGF%. Chychrun is really the one that is the glaring red flag as his o-zone starts and o-zone faceoffs are both well above 50 percent in January.
Strome is also a huge warning sign because he’s getting over 70 percent of his o-zone starts and o-zone faceoffs in the offensive zone and his xGF% should not be that bad.
- I have to insist that this decade-plus long column in an exercise in exploring the context behind the numbers, but more importantly, let’s talk about the substance of the point, which is good. We already went over Strome, and he’s still Washington’s most offensively deployed non-🐐 forward, and it’s still a major concern. The plot below shows each Caps skater and how they’re deployed (up and down, where up means more offensively) and how play trends during that time (left and right, where right means the Caps are controlling play).
- Jakob Chychrun, frankly, doesn’t worry me much. He’s got a big gap in his on-ice possession stats when you jump from expected goals to raw shot attempts: minus-3.6 percentage points. There are two things going on there, and you can see them both in HockeyViz’s heatmaps, below. At left is where opponents shoot against the Caps, at right is where the Caps shoot against opponents. Red means more attempts are coming from those spots – at left that’s bad and at right that’s good.
- Left is a real problem. Chychrun is getting burned on high-danger chances – the dark red blob in front of the net at bottom. This is a combination of having trouble getting the puck back in the defensive zone (Washington’s biggest system-level problem in my mind, though more on Fehervary and Roy than Chychrun), plus allowing too many unimpeded rushes in general. But at right: I wouldn’t sweat it. A lot of Washington’s offense is coming from high in the zone (which is at the bottom of the chart, sorry). That’s Chychrun himself taking an outsized number of shots and being a freaking sniper with them. Only 11 defenders shoot more than him – Brent Burns being number one, obviously – and only four score more than he does.
- Being consistently good at shooting from outside is two things: 1) rare and 2) bad for analytics. The weighting system for expected goals doesn’t know that you – Chychrun in this case but also think of Ovechkin on the power play – are positively deadly from that spot on the ice – because usually players don’t score well from there. Your expected goals numbers won’t show it, but you’re still good. There’s a lot of problems in Chychrun’s game, but looking at the context makes him better in my eyes, not worse.
- I don’t normally do this, but if anyone knows a good way to get period-by-period microstats at scale, I’d love to see it. I’m generally suspicious of split stats (home/road, no thank you), but I’m even more suspicious about Washington seeming to play one terrible period every game, including that first period against the Jets on Saturday. How common is that? Does everyone do that? Are we just imagining it?
- Let’s game out Alex Ovechkin real quick. Snapshot statistics no longer apply to this man, whose reality distortion field has defeated all measurements but one: puck in net. In his career he has 877 goals. He needs 895 to pass Wayne Gretzky. This season, he has 24 goals in 36 games. There are 30 games remaining.
24/36*30=20,
20+877=897;
897>895.
I could say something about an unsustainable shooting percentage, 18.2, a career high, but I can’t – Ovi’s reality distortion field is too strong. I’ll just state my preference again: I wish him to finish the season with 894, tying the record, and then have his first goal of next season be the golden goal, anticipated and watched by the whole world, with Joe Beninati and Craig Laughlin making the call on TV, and Wes Johnson announcing it in the rink. That’d be a nice day.
This story would not be possible without Natural Stat Trick, All Three Zones, Evolving Hockey, and Hockey Viz. Please consider joining us in supporting them.
