One fatal flaw: April snapshot

The Washington Capitals have eight games remaining in a very irregular regular season. These last few games will be massively important for many reasons, but much has already been settled:

The Caps will make the playoffs. Despite losing Jakub Vrana, they have a lot of high-end scoring talent. Despite brutal deployments, they have strong forward depth. Despite some liabilities, their defense is above average.

Overall, and despite predictions to the contrary, the Capitals are still a damn good hockey team.

We can say all this because we have the confidence granted to us by the standings (first place in the East by a point) and by the deeper data this snapshot exercise explores. But that data also gives me the confidence to say I’m not at all confident in the Capitals’ goalies.

Forwards

Defenders

Glossary

• GP – Games played.
• TOI – Time on ice in minutes
• SA% – Shot-attempt percentage. The share of total shots attempted by Washington while the skater is on the ice. 50% means even.
• SA% Rel – Relative shot-attempt percentage. The difference in SA% when the player is on the bench versus on the ice. 0% means even.
• GF% – Goals-for percentage. The share of total goals scored by Washington when the skater is on the ice. 50% means even.
• PDO – The sum of Washington’s shooting percentage and saving percentage when the skater is on the ice. 1 means even. The acronym doesn’t stand for anything, and yes, that it is annoying.

Notes

• Before we get to the main dish, I want to say that I’ve removed Panik and Vrana from the snapshot. They’re both good NHL players, as long as your expectations are in check. For Panik, that means “I expect him to score only when it’s totally an accident.” For Vrana it means he now has to prove in Detroit that he is the superstar we know he can be. So far, so good.
• Earlier this month, the Caps traded defender Jonas Siegenthaler to the New Jersey Devils for picks. That’s an ambiguous return for a player I think is unambiguously NHL-ready, but you won’t see that in this season’s numbers (cf. weird year). Siegenthaler’s 45.7 on-ice shot-attempt percentage (SA%) is the lowest among eligible defenders, but I think his 76-minute sample was extraordinarily weird. He had some normalish deployments in January, then hostilely spare usage in February and March before shipping up to Jersey. I think all the last seven Caps games of Siegenthaler’s career did more to obscure his value than reveal it. Now he has to deal with whatever got him on the COVID protocol, but I’m excited to see him get a bigger role with the Devils soon.
• Out with the old, in with the new: Anthony Mantha has dominated play since joining the Caps six games ago. In addition to his individual scoring (four goals and two assists), he’s seen his team control two-thirds of shot attempts. And while he’s shooting hot individually, the Caps are relatively unlucky during his five-on-five shifts, with goalies saving under 90 percent and the Caps shooting 8 percent (resulting in his 97.0 PDO). Mantha isn’t going to score on every third shot forever, but I think he’ll remain a strong asset for the team.
• Other deadline pickup Michael Raffl has played under thirty minutes so far, and they’ve been thirty weird minutes against the Islanders. I don’t put much stock in his snapshot numbers yet, but I can see why he’s such a well-liked player among coaches. For now, I’ll punt on real analysis.
• The players in the tables above are ordered by their on-ice shot-attempt percentage, so I’m kind of comforted that the fourth liners are at the bottom of the forwards. Like we discussed last time, the deployment decisions around the grinders are dramatic and I think should be dialed back a bit, but at least those are players for whom the least is loss with less offensive-zone time. It’s impressive nonetheless that Nic Dowd is shooting 15 percent during five-on-five.  Everyone above him in shooting percentage is also above him in bench order: Kuznetsov, Eller, Sprong, Backstrom, and Sheary.
• Even with Panik and Vrana gone, it seems like there’s a wealth of top-nine forward options right now. Conor Sheary (53.7 SA%) and Daniel Sprong (51.2 SA%) have earned their sweaters on a regular basis not just through their scoring, but by being reliable players even without the puck. I feel like we’re due for some lineup drama soon, but that’s only because there’s too many viable players. These are a buncha hot-shooting dudes.
• It’s time once again for a Nick Jensen appreciation bullet. Jensen does not top the list in on-ice possession stats (the Caps control 0.3 percentage points fewer of the shot attempts when he’s on the ice), and there’s surely some offense lost from his defensive reliability, but just look how reliable he is defensively. I suppose there are a lot of ways to clear the crease, but Jensen is definitely good at some of them. These heatmaps show where opponents are shooting. Blue blobs mean fewer shots from that location and red blobs means more. Left is when Jensen is on the ice; right is when he’s on the bench.

• According to HockeyViz, opponents are a quarter less effective (compared to league average) when they’re up against him. And, to the broader point, the Caps are still pretty good when he’s on the bench, which means it’s time for us to talk about something outside the scope of the snapshot but strongly implied by it. But first, let’s we gotta talk about patterns versus tangible details real quick.
• We as humans aren’t great at detecting patterns from hundreds of hours of stuff like hockey, but we’re real good at remembering certain dramatic incidents. So without the support of numbers, I might say that the Caps have big defensive problems and cite a bunch of opponent odd-man rushes that have wrecked the team lately as evidence. But those anecdotes stick in our head even if (or perhaps because) they’re not indicative of the underlying pattern. And that underlying pattern is really good! Only five teams in the NHL allow their opponents fewer high-danger chances than the Caps do (yes, the Isles are one of them). But that good “team defense” isn’t necessarily showing up in goal stats because of, well, you know: goalies.

• In the graph above, the Caps have an expected save percentage of 92.8, eighth best in the NHL — putting them on the good/right side. But the team’s actual save percentage is 91.6, ten spots lower — putting them on the lower/bad side of the chart. That’s a lot of good defensive play squandered by goalies Ilya Samsonov and Vitek Vanecek. And if we isolate just the situations when the Caps are leading in the score, hoo boy, get ready for this.

• The Caps don’t (on a pattern-level) play any worse defensively when protecting a lead– so they’re still around the same spot on the right-left axis. But then the actual saving percentage when leading is 90.2, third worst in the NHL. That is costing the team wins and points, and it could be disastrous if it continues in the playoffs. (By the way, the Philadelphia Flyers got cut out of both of those graphs because of how low their actual save percentages are. Poor Carter Hart.)
• But who knows what will happen in the playoffs? Those are smaller samples against less diverse opponents. And this is — if you haven’t heard this before —  a Weird Year. I have no strong preference about Samsonov or Vanecek (I guess gimme Vanecek by a hair), but that pair is the weak point of a very strong team. The Caps have high-end scoring, tons of forward depth, stalwart grinders, and stout defenders. They just need their goalies to get a little better —  or a little luckier.

This story would not be possible without Natural Stat Trick and Hockey Viz. Please consider joining us in supporting them. A very of this story and its underlying data appeared on RMNB’s own Patreon.

Screenshot courtesy of NBC Sports Washington