It's Friday afternoon, and like many Americans - like many people, probably - I've spent the past few days pulling in election analysis from an eclectic set of sources: network news, the FiveThirtyEight live blog, people calling and texting; each new update had an undeniable effect, though generally very small, on my mood at the time. But if I think about the source that had the strongest influence over me, there's only one possibility - the betting odds, which I tracked across a handful of European markets. It was a bit of a roller coaster - Biden started on Tuesday at around 65%, which doesn't strictly equate to a 65% chance of victory (but thinking of it in those terms is appropriate for the context of this piece). Later that night, Trump jumped up to almost 75% at 10 PM (EST) as Ohio and Florida cleared their throats, then Biden clawed back to almost even by around 1 AM for the lack of other surprises, which is when I finally called it a night. I learned the next morning that Trump had actually surged past 70% again at around 3 AM, but by the time I saw it news from Wisconsin and Michigan had pushed Biden above 70%, saving me the trouble of worrying about Trump's brief rally. As you know, Biden's looked more likely to win with each passing day; he's at 95% at time of writing.
I think my undergraduate background in statistics coupled with my lifetime interest in sports - and perhaps the occasional winner I've cashed at the Keno desk - predispose me toward accepting the betting market as the most reliable source of real-time election analysis. But when I share this preference, I'm often met with confusion - what makes someone putting up the odds or betting the house more reliable than a news network, or a pollster? It's been an interesting experience trying to answer this question, almost like being asked to explain my positions on certain issues, like ending racism; it's hard to explain what we take for granted. My standard responses have included attempts to describe a variety of related factors - the underlying science of probability, the nature of betting markets, or the inability of media coverage to correctly link cause and effect.
I realized last night - while listening to complaints about certain networks calling Arizona too early - that I should have offered a far simpler answer: the betting markets work because there are consequences for being wrong. A person who wagers $100 on an election outcome ends up with one of two results - more than $100, or $0. Which of those outcomes do you think is more likely to result in another bet? The far more straightforward consideration is that casinos go out of business very quickly if the oddsmakers set ill-advised betting lines. On the other hand, if Fox News turns out to be wrong about its early Arizona call, will they be banned from covering the 2024 election?
These considerations alone ensure nothing, but it's suggestive of the type of person who prefers to wager on an outcome rather than speculate endlessly on camera about all remaining possibilities. The confident bettor is also more likely to wager large amounts, and dollar totals are reflected in the odds (with wagers, unlike in an election, the tally is almost always irrelevant). The betting market of my imagination likely has no time for the faces that filled my screens this past week, these pundits who earnestly analyze the results in real-time and weave the newest event into an ever-evolving web of predictions. It didn't take long to realize the same people who were wrong four years ago were back to explain why they were wrong again; you'd be forgiven for thinking the media operated like the Supreme Court, and offered lifetime appointments.
I don't blame the media for the state of election analysis (I have other media complaints, which I'm saving for later). I certainly don't blame any individual within the media for the state of election analysis, even if some of them - like the pollsters - should probably know better, having been burned once by their pie charts in the sky. I believe the media is doing the best it can as it covers an endlessly complex story, which is essentially a series of fifty counting exercises that is confused by the countless ways each next vote can count, or when, and that's assuming they count at all, and then of course the question of whether we can count on those folks who would rather count ballots than count votes, and then this all leads to a final count being tracked simultaneously to the smaller counts, which is what really counts; I just don't think it's possible to cover this adequately, like a five-foot blanket on a six-foot man. But it's important to me that I understand the system of incentives and consequences that contextualize someone's opinion because I'll know what's at stake; the architect can tell me the bridge is sound, but I'd much rather he cross first.
