Drink, Vote, Gamble: The Power of Magical Thinking

Harry's New York Bar in Paris' 2nd Arrondissement has been serving cocktails to ex-pats and other fellow-travelers since 1911. Starting in 1924, way before Americans abroad were able to vote in US presidential elections, Harry's has conducted a straw vote among citizens who stop in for a drink. According to lore, the "winner" of the Straw Vote has lost the "real" election only three times in 96 years. 

Being traditionalists living abroad in a very traditional country, we decided to take the métro over to Place de l'Opéra and cast our straw ballots over Manhattans and Daiquiris (the normal, decent kind, no blender abominations here). The voting operation was very professional, with tight ballot security measures and and strict identification requirements--only American citizens who produce their US passport for the serveur are provided a paper ballot and a pen.

 

Chain-secured, padlocked ballot box, passport mandatory.

Oski, the official polling place monitor for Harry's Bar.

 

Ballot secrecy is optional, with early voting tallies updated weekly. Results shown as of 28 October 2024.

The day of our visit, and before we performed our civic duty, the votes stood at 53% for Harris, and 47% for Trump. This is very close to the Iowa Electronic Markets average contract prices for the presidential vote share for the month of October, $0.52 for Harris and $0.48 for Trump.

Since election markets tout themselves, and the entire enterprise of election gambling, as "more accurate" than the polls at predicting election outcomes, I've applied the Werewolves Adjustment Factor (WAF) to account for IEM traders' historic over-valuation of Democrats and under-valuation of Republicans. This method predicts 50% of the the two-party vote share for Harris, and 49% for Trump--virtually the same as the New York Times and Silver Bullet poll-trackers.

IEM VS: IEM vote share contracts; WAF VS: Werewolves Adjustment Factor applied to IEM VS

But as we know, in the US presidential election, final popular vote share is effectively a non-binding referendum. Just a suggestion, to be taken under advisement. What counts is winning the electoral college. In an earlier post, I mistakenly included the Lott-Stossel Election Betting Odds predictions (as of September 30) at 52% for Harris and 47% for Trump. Which at the time, was pretty close to the IEM VS and to the poll-trackers

In fact, what Lott-Stossel tracks is the price of a contract for a specific candidate becoming President of the the United States; that is, winning at least 270 electoral college votes. But I did not realize this until mid-October when I checked the latest numbers after returning from Vienna.

Whereas at the beginning of October, gamblers had placed $1.2 billion to give Harris a 5 point edge over Trump, by mid-October they had bet about $2.5 billion to give Trump a 24 point edge. This 15 point swing in favor of Trump, in just two weeks, had no parallel in any of the polling trackers. ABC News' 538, which simulates election probabilities based on polls, also showed a swing in favor of Trump winning it all, but the chance is only 2 points higher than Harris's chance (a 6 point swing from September).

As of this writing, Lott-Stossel reports that over $3 billion has been bet, with gamblers giving Trump a 62% chance compared to a 38% chance for Harris. It was this rapid and dramatic shift that me realize my error. Since September, $1.8 billion additional bets have been made; back of the envelope, 71% of the money is on a Trump presidency, compared with only 47% of pre-October money.

I was not the only one to notice the large amounts of late-breaking money placed on Trump at very high contract prices (remember: the ostensible goal of election gambling is to make money off of information gaps, not to predict elections). It was also noted by Forbes, Barron's, and New York magazine's Intelligencer. The Intelligencer posits a number of explanations for the high volume of October bets on Trump.

One explanation is that election gamblers may have really good information that Trump will win, and they were just waiting for the right time to go all in. But unless that good information was only received in October, they should have been betting on a Trump victory back when prices were much lower. As it stands, a $0.62 contract will return $0.38 should Trump win, which is roughly 0.6:1 odds. Which is not a really risky bet, but has huge opportunity costs compared to what these very large positions could have earned if placed earlier.

Another argument is that election gamblers are naturally predisposed to vote Trump. The reasoning is that the largest gambling site, Polymarket, is partly-owned by Peter Thiel, a Trump supporter, and requires gamblers to purchase and trade in crypto-currencies. In essence, the market may be full of crypto-bros who are betting with their hearts on their preferred team regardless of what other information such as polls are saying. And they are doing so at high prices.

The third explanation for me is the most intriguing. Basically, the Intelligencer describes wealthy Trump supporters--avowed ones like Elon Musk talking up Trump's chances and anonymous "whales" taking inexplicably large gambling positions--as trying to create a perception that Trump's win is foreordained. After all, markets reflect "the wisdom of crowds," and "studies find that ... markets tend to be better at predicting elections than polls." If the markets are confident it's going to happen, it's already a done deal.

To me, this perfectly describes an effort to perform a kind of "chaos magick" that doesn't merely predict a Trump victory, but wills it to happen by "speaking it into existence." This is best explained by Gary Lachman in his 2018 book Dark Star Rising: Magick and Power in the Age of Trump, which details the esoteric and occult influences in far-right movements in general, and within the Trump orbit in particular. For Lachman, speaking about the period from 2015 to 2018, the critical elements were the manipulation of grievances and negative emotions through effective storytelling, symbols and memes. Think of the imagery of the golden escalator and the invective-laden speech that followed, torch-processions in Charlottesville, or the co-option of "Pepe the Frog" meme. These were all elements in a process through which, in the words of one Trump supporter in 201, Trump was "dreamed ... into office." In 2024, taking a $14 million position on a Trump victory--as a Polymarket trader named Freddi9999 has done, at an average of $0.54 per share--could be viewed in this same vein (though, of course, we know nothing about Fredi9999's politics, only that he has been buying his shares on the expensive side).

Now, the idea that Trump was magically willed into office in 2016 is of course complete bunk. But the idea that belief in "speaking things into reality" would motivate some people to take large gambling positions as an act of demonstrating faith is not farfetched. Nor is it a niche belief that exists only in occult or far right political circles, despite Lachman's evocative label of "chaos magick."

One of the core tenets of Christian Science, as founded by Mary Baker Eddy in the late 19th century, was that the physical world is an illusion, brought into being by human society's thoughts and mental states as they collectively grasp at spiritual reality. Most famously, the Church of Christ, Scientist holds that illness and other catastrophes are the result of erroneous, negative thinking. Evil thoughts can cause real harm, and conversely, personal health and well-being, and a more just social order, can only be realized by prayer, contemplation of scripture, positive thinking, and good works. Shorn of their overt theological roots, these ideas form the underpinnings of perennially popular self-help titles such as Dale Carnegie's How to Win Friends and Influence People (1936), Napoleon Hill's Think and Grow Rich (1937), Norman Vincent Peale's The Power of Positive Thinking (1952). They also appear in more countercultural circles such as the Human Potential Movement.

And I think we all can agree that the illusion of control is very important in many aspects of life. It makes us feel like we don't live in a cold, random universe where we're lifted up or cast down without a lot of regard to effort, ability or character. It's why people vote in the first place. What is a vote except an act of faith that you'll get what you think you're asking for?

So how can we best explain why so much of October's election bets have shifted towards a Trump victory? We can begin by applying Occam's Razor, the principle that among equally explanatory hypotheses about a phenomenon, the simplest one with the fewest suppositions is preferred.

According to the "wisdom of crowds" explanation, we need to accept that October gamblers taking big positions are (1) rational, that they are (2) betting in their self-interest and therefore are (3) utilizing the best available information to make decisions about their bets, and that (4) this information was generally not available before October and (5) is still not widely available.

According to the "home team bias" thesis, these gamblers are (1) intendedly rational, that they (2) believe they are betting in their self interest, but (3) they are basing their bets on a combination of intuition, wishful thinking, and widely available information.

According to the "chaos magick" thesis, these gamblers (1) want Trump to be elected President, for any number of reasons, (2) believe they can effect the outcome of the election by exhibiting sufficient faith through their bets, and (3) they believe they will make a large amount of money--though will not maximize their returns, which they could have done by betting in the summer or in September--in the event that their magic works.

Maybe I have not characterized these explanations correctly. But what is clear is that we should not automatically defer to the "wisdom of crowds" explanation that market proponents rely on when explaining why betting on elections is "more accurate than polls" at predicting presidential outcomes. It's overly complicated compared to the alternatives, and takes no account of political gambling behavior as reflective of political preferences or acts of faith. And if it fails to explain seemingly erratic betting behavior in October, why should it be believed in any other months?

Worse still, this year at least, the markets don't agree with one another. Contract prices for IEM's winner take-all-market (WTA, a bet on who will win an absolute majority of votes, regardless of who wins the electoral college) are $0.81 for Harris; the same bet can be placed on Polymarket for $0.59. While both pay out in the event that Harris wins 50.1% of the popular vote, the amounts won by gamblers would be vastly different. They simply are not the same predictors.

So what do we know? Not much. 

We know that the popular vote will be close, but that doesn't matter much for who will be President of the United States in January 2025.

We know that whoever wins will be put in office according to the preferences of a small number of voters, in a handful of tightly-contested districts in a handful of evenly-divided states--not because of the power of prayer or any other ritual acts that "spoke it into existence."

We know (or at least I've convinced myself) that election markets may turn out to be right this year, but considering the divergence of their predictions and the huge blind spots in the theories on which these markets justify themselves, any relationship between bets and outcomes may be statistically spurious--or perhaps a meaningfully coincidental "synchronicity," to borrow Carl Jung's paranormal explanation.

And we know that like voting itself, predicting elections and obsessing over "what they are telling us" every four years are also rituals. They may not magically produce the outcomes we want or expect, but at least if you perform the ritual at Harry's in Paris, you get to drink.