There's a post over at SportsGrid right now complaining about a whole bunch of things with statistics in hockey. The general theme is that the statistics certainly and obviously matter, but (a) writers aren't using them properly; (b) they can't necessarily beat Las Vegas; and (c) a bunch of other stuff sprinkled in there, like neglecting on-ice shooting percentages when creating projections.

I don't disagree with everything in that post -- I do agree, for example, that there are mainstream guys starting to pick up the new-age stats who are just bludgeoning its usefulness. This was going to be an inevitable issue at some point: comprehension and application of material comes with time, and I think everyone, invariably, will make mistakes on this sort of stuff. It's kind of a reality if you're hoping for better analysis of the sport from the media who cover it. For now, we just have to sit and observe, point out the mistakes, and hope it's not an issue a few years from now.

The one thing that I found sort of interesting was the underlying point about beating Las Vegas. I'm not sure that's true, or close to true for that matter. Unfortunately, I don't have a full data set for series prices -- I've got about 75% of the prices dating back to 2007, but not the complete one I'm looking for (if you have these numbers, contact me). What I do have is Score-Adjusted Fenwick%, last twenty games, for every team that's made the playoffs since 2007. And I have the win/loss records of every home team -- home teams are darlings in sportsbooks; it takes a herculean mismatch to be a road favorite in one of these draws -- in the same timespan.

I mean, if this doesn't convince you that possession matters and has pretty significant predictive value, I don't know what will:



So, blindly picking the team with the better Score-Adjusted Fenwick% over the last twenty games gave you a 67-31 record, or a 68% success rate. Picking the home team -- home teams are, on average, 56% likely to win a single-game and a seven-game series -- in that same timespan? 53%. So, probably a touch-low, but we are still sixteen (!) games apart here.

This matters, of course, because I can promise you that in many instances, Score-Adjusted Fenwick% actually recommended taking the road team, which an awful lot of the time meant taking the Las Vegas underdog. This year alone, Minnesota (win) and Chicago (win) and Los Angeles (win) and Columbus (loss) would've given you a 3-1 record in RD1 on four road dogs.

Possession is not everything, but it matters quite a bit. Understanding the role of puck possession and the presence of variance in a single-game / seven-game series is key for everyone, and that extends to front offices who may do something bat-shit crazy in response to an early exit.

And yeah, you can still beat Vegas. But this window is starting to close, one would guess.

--

Thanks for reading!

Join the Discussion: » 4 Comments » Post New Comment

» Wrapping Things Up

» Enforcer

» Random Thoughts

» Shot Coordinate Fun

» Any Room?