Lessons hidden in sports betting markets

With sports betting now legal in several US states, I might as well give away my number one piece of advice for amateurs looking to gamble:

Don’t bet.

It’s an easy recommendation. Numbers implied by betting markets are too good, too close to the truth that, when accounting for the vig, it’s nearly impossible to make a long term profit.

But just because your local statistician tells you not to bet doesn’t mean you shouldn’t check out betting market odds. In fact, it’s just the opposite. There’s no single piece of public information that is more accurate and informative with respect to a professional sports game than the closing odds. Related: there is no better standard with which sports scientists can compare prediction models than to compare to betting odds.

In a recent paper, Ben, Greg, and I took a deep dive to learn about what betting odds tell us about the four major North American pro sports leagues, the NBA, NHL, NFL, and MLB (sorry, MLS, but three-way odds are a different story). It’s titled “How often does the best team win? A unified approach to understanding randomness in North American sport,” but our alternate title might as well be “Betting markets: if you can’t beat em, use them in a statistical model.” I’d welcome readers to check out a pre-print of the manuscipt, forthcoming in Annals of Applied Statistics, if they are interested in more details. As the paper’s focus more on the technical side, what’s missing is a cohesive summary of what it all means. In other words, how can we use our results – built using betting market data – to make better decisions in sports?

I’ll provide several ideas in the following set of posts.

Hopefully you’ll be able to use these posts to learn a bit of insight into how each sports leagues compare to one another, and how various factors can play a major role in team success.

Note: this series of post is based off of a forthcoming academic paper and does not reflect views of my current employer (the NFL).

Related