Our final Gelo ratings for the 2022–2023 season are posted. We’ve got observations and thoughts.
First, the observations:
- Vegas finished as the best-rated team in the league (3.22), followed by Boston (3.16), Edmonton (3.11), Florida (3.07), Colorado (3.04), and Dallas (3.02).
- Edmonton finished with the best Ogelo rating (3.65), followed by Florida (3.49), Toronto (3.37), Buffalo (3.35), Colorado (3.22), and Vegas (3.21).
- The Islanders finished with the best Dgelo rating (2.60), followed by Carolina (2.65), Boston (2.66), Dallas (2.74), Los Angeles (2.75), and Vegas (2.83).
Now, the thoughts:
- Gelo is very “of the moment,” by which I mean it’s heavily impacted by recency, owing to our impression (from the data) that the quality of a hockey team changes quickly and is—more than any other sport—something other than a sum of parts. Going off of this, I would not say the Knights were better than the Bruins this season. I would say they were better than the Bruins as of June 13th, though.
- Vegas being top-six in Ogelo and Dgelo, and being the only team with that status (the Avalanche were the only other team that finished top-13 in both categories, which is nuts) is a real testament to what they did, which to my impression is most simply: They built a good, well-rounded team. None of hockey’s biggest superstars, nothing unconventional in terms of being super-high scoring or super-strong defensively, just solid hockey on both ends of the ice. This can get a little circular—our only inputs being goals scored makes Ogelo and Dgelo just as arguably measures of style/approach as they are measures of offensive and defensive capacity—but any way you slice it, Vegas was a well-rounded team who didn’t do a lot of things in extreme fashion. That worked out.
- The mean Ogelo/Dgelo rating was 3.05, sizably up from the 2.85 where the model started it. Embarrassingly, I don’t remember off the top of my head whether the model started this season with 2.85 or started last season or a prior one with 2.85. Either way, it’s a reflection of the increase in scoring, but we’d have to dig more to figure out the duration of increased scoring that’s affecting this. Is this a multi-year thing? I know it’s been going on longer than just in the playoffs, but it’s possible playoff scoring is also different? Should we have the time this offseason (Gelo’s a low priority among the models because we’re more a college sports shop than pro sports), we’ll have to figure out whether we should start next season at a new mean.
- We’d like to look into playoff differences overall, but at the same time…Gelo had a really, really good postseason, judging by its performance as a futures market guide. We got a 28% return using it, and we were hyper-cautious in ways that did not eventually gain us units. We did lose 25% last year, so in a two-year sample, it’s more or less even, but even breaking even is a pretty good result when you’re betting in markets with a huge natural vig.
- We’d also still like to look into the back-to-back games thing. We don’t have any current aspirations of introducing a goalie rating or things like that, which will probably keep Gelo from being the most useful in individual game predictions, but determining the impact of rest and/or travel is within our realm of capability.
- Building a model and using it to bet futures on a sport’s postseason is a good way to “get into” a sport. Much like last year, I’m exiting these playoffs loving hockey and excited for next year. Which is great as its own reward.
