It’s easy to look at the Caps’ goal production in recent weeks or longer and assume (perhaps correctly) that they are not good enough offensively to be competitive right now. To wit, the Caps are 30th in the League in goals per game in March (and would be dead last if not for the seven goals they dropped on one of the two teams below them), 27th since the Olympic break, 25th since the start of February, 22nd in calendar year 2026 and so on.
It’s just as easy (and perhaps more dangerous) to look at their goals-against and assume that they are good enough defensively to be competitive right now. That may or may not be the case. Consider that, on the season, the Caps are sixth-worst in the NHL in expected goals against per 60 at five-on-five per MoneyPuck, clocking in at 2.65 (Vancouver is worst in the League at 2.74). But that off-putting ranking is being masked by a sparkling 2.17 actual goals against rate in those situation, good for fifth-best in the League (Natural Stat Trick’s data tells a very similar story, if you’re looking for corroboration). For you visual learners, that looks like this:
Of course, last we checked, games were won and lost on the basis of actual goals, not expected goals. But while actual goals can tell the descriptive story of what has happened in the past, expected goals are arguably more important going forward, because regression is a thing…
…unless it isn’t.
(It is, but stick with it for a second.)
If someone told you that a team had given up two goals in each of its last ten games, but had an expected goals-against in each of those games of four and asked you to bet on whether they’d give up two or four in the next game, which are you betting? Sure, you’d be right to question the model that’s expecting twice as many goals than reality is delivering, but assuming the model is good (and we have reason to believe these models are pretty solid), your next question – at least if you’re a Caps fan – might be, “Who’s in goal?” Now we’re getting somewhere.
Logan Thompson has been a human eraser this year, saving a massive 24.4 goals above expected at five-on-five in just 46 games (so saving around half a goal per game at fives). Among goalies with 25 or more appearances, LT is tops in that stat in both count and rate:
HockeyViz tells a similar story for Thompson, with just a tick under 20 goals saved above expectation:
In other words, Thompson’s play has masked a lot of the defensive shortcomings of the skaters and/or structure in front of him.
Thompson’s backup, Charlie Lindgren, is… well, more or less exactly as expected – in 20 games, Lindgren has allowed 65 goals in all situations, on an xGA of 63.96, and 40 goals on 39.72 xGA at fives. Remarkably close. (To fill in one blank, Thompson has allowed 17 four-on-five goals on 17.11 xGA.)
So, back to our hypothetical, if Lindgren is in net, you’re betting the expected goals against and if Thompson’s in there, you might be justified in thinking that he’s going to continue to outperform that. Which brings us to the conclusion here – if the Caps are going to continue to be near the bottom of the League in expected goals against, they’re not going to be particularly successful unless they a) minimize the number of games Charlie Lindgren plays (unless he gets better/luckier), and b) continue to get the same Vezina-caliber netminding from Logan Thompson without regression or injury (see point a).
Despite glowing goals-against numbers overall, the Caps need to tighten up defensively going forward (i.e. next season) if they’re truly going to be considered a top team. Getting great goaltending from your #1 goalie is great; getting great goaltending from your #1 goalie playing behind a rock-solid defense makes you a contender.