Read on Twitter
Hot take: If Erik Karlsson were born 5-10 years later, the analytics community would call "peak Erik Karlsson" a far worse defenseman than they currently do.
Instead of calling him generational, they'd compare him to to maligned offensive D like Burns, Rielly, and Carlson.
Instead of calling him generational, they'd compare him to to maligned offensive D like Burns, Rielly, and Carlson.
Back in the day, we had neither xG nor RAPM. We had Corsi in place of xG and on-ice/rel stats in place of RAPM.
With these stats, we still did a decent job of isolating a player's contribution to their goal differential and were wrong less frequently than those who ignored them.
With these stats, we still did a decent job of isolating a player's contribution to their goal differential and were wrong less frequently than those who ignored them.
In 2011-2012, Karlsson was the nearly unanimous Norris choice among the nerds: Scoring rates (and Corsi For) showed he was superior offensive player to Shea Weber, but Corsi against showed he was comparable defensively.
Throughout his peak, Karlsson's Corsi against was strong.
Throughout his peak, Karlsson's Corsi against was strong.
The Corsi narrative from the nerds holds up well against today's scrutiny. Using RAPM, we can see that Karlsson was a comparable Corsi suppressor to Weber in 2011-2012 while being a much better driver of Corsi For, and over his career, he's generally been a good Corsi suppressor.
We've switched to using expected goals against as the top proxy for defense for good reason; it's more accurate because it doesn't treating all shots as equal.
Using expected goals against (or real goals), it's clear that Karlsson was not as good defensively as Corsi suggested.
Using expected goals against (or real goals), it's clear that Karlsson was not as good defensively as Corsi suggested.
(Chart courtesy of @JFreshHockey.)
