Media “Polls” or Cold, Hard Cash?

“Prediction is very difficult, especially if it’s about the future.”

– Neils Bohr

---

November 27, 2024— With three days left in November, you can still bet on whether bitcoin will hit $100,000 before December.

And odds are 54-to-46 that Tulsi Gabbard will be avenging her “no-fly” status before you know it.

The UN-skeptical Elise Stafanik is practically a lock for its ambassador. Even RFK Jr. could be having a go at Anthony Fauci’s legacy soon.

Despite being raided by Federal authorities for nailing the outcome of the election weeks in advance, Polymarket, the betting site, is still up and running.

It’s probably a no-brainer, but Trump’s lead widened dramatically on Polymarket in July after Biden’s feeble debate performance. The gap narrowed as the DNC anointed Kamala Harris and her “vibe election” appealed to anyone with Trump Derangement Syndrome.

In the end, while the media polls grunted and strained to produce a “dead heat,” the odds on Polymarket for a Trump win roughly equaled the popular vote.

Okay, so, moot question: What’s the better forecasting tool?

Today, our friends at The Free Press take a peek behind the curtain. For specificity, they track one French trader who bet over $80 million – and doubled his money – on Polymarket. It’s a good holiday read. ~ Enjoy, Addison

How a French Whale Made $85 Million off Trump’s Win

Joe Nocera, The Free Press

On October 15, three weeks before the presidential election, Ashley Grosse heard that a Frenchman wanted to run a study with the firm where she works, YouGov. In an email, the prospective client told the firm he wanted to poll “the current U.S. elections focused on three swing states, specifically Pennsylvania, Michigan, and Wisconsin,” investigating the “shy voter effects” within the next seven days. “The faster, the better,” he wrote.

Soon Grosse, the general manager for YouGov’s Scientific Research unit, found herself in a Zoom meeting with the Frenchman, where she agreed to conduct the polling he wanted. What she went on to discover with her poll led him to place the single largest bet ever made on an election. His wager was that Donald Trump would win—and win big.

“I had conviction,” the high-rolling whale told me over the phone last week, “so I bet a lot of money.”

That a trader living in Europe was able to predict such a decisive victory when every major professional pollster was forecasting a dead heat is one of the most surprising tales of the U.S. election. Now, the Frenchman is revealing to The Free Press exclusive details about how he did it.

In our conversations by phone and email, I discovered he is a trader who wanted to go by the pseudonym “Théo,” because he said disclosing his identity “would disrupt my life, as my family and friends are not aware of the extent of my wealth. It will also seriously endanger my safety and that of my kids.”

Back in October, Grosse—who knows Théo’s true identity—told him it was extremely difficult to gauge the preferences of so-called “shy voters,” who either can’t be reached or refuse to talk to pollsters. But Théo said he had an idea. He had seen some academic literature about “social circle polling,” in which pollsters asked people not only who they were voting for but who they thought their acquaintances were voting for. Those polls had performed well. He wanted Grosse’s team to do something similar: Ask people in the three swing states who they thought their neighbors were planning to vote for. Grosse later told me she found his idea “intellectually interesting,” but she was skeptical the neighbor question would unearth anything useful.

But Théo was convinced it would smoke out a major preference for Trump over Kamala Harris that was hiding in plain sight. “It was clear to me that most of the polls were skewed to the advantage of Harris by. . . ignoring the shy Trump voter effect,” he wrote to me in an email.

What Grosse didn’t know—what nobody knew at the time—was the reason he was commissioning the poll: He was placing multimillion-dollar bets with Polymarket, the popular prediction market. Even before engaging YouGov, he had wagered $30 million on Trump. Once he saw the results of the neighbor question at the end of October, he went all in, using 11 different crypto accounts to raise his stake to $80 million. In fact, he was so confident Trump would win, he told me he sold off virtually all his liquid assets to raise the $80 million for his wager.

And thus it was that on November 6, with Trump’s victory in hand, Théo more than doubled his money, taking down a staggering $85 million. He bet not only that Trump would win the Electoral College but the popular vote as well. He also put money on Trump winning the three Rust Belt swing states. On the eve of the election, Nate Silver said the race “was a pure toss-up.” Théo knew better.

In our first conversation, I asked Théo for his backstory. He told me that two decades ago, he began his career as a financier on Wall Street, working for a “major firm” that he declines to name. (This was among the details he wouldn’t tell me for fear that someone might identify him.) Upon returning to France, he started his own business in Paris, which he later sold. When I asked him how he makes his living now, he said, “I manage my own money.” He told me he trades bonds, commodities, oil futures, cryptocurrency—pretty much anything that might yield a profitable trade. Prior to his Polymarket bet, he said his trades had added up to a $16 million profit in 2024.

With his thick French accent, Théo can be a little hard to understand at times. But he made it clear that he felt he deserved the money he had won because he had outsmarted the American polling industry. Speaking of his calculation that the “blue wall” swing states in the Rust Belt would all go for Trump, he asked me in one email, “Why am I the only guy who thought of calculating the odds as I did?”

He told me he got the idea to calculate the shy voter effect from the work of a research team at the Santa Fe Institute, MIT, and the University of Southern California Dornsife. Since 2016, the team has been conducting polls every two years, asking people, “What percentage of your social circle will vote for which candidate?” In 2016 and 2020, according to an article written by two members of the team, the “social circle” question was “better at predicting the outcome of the national popular vote than the traditional questions.” This was not only true in the U.S. but also in Europe, where the team had similar success using the social circle poll in four European elections.

Mirta Galesic, one of the members of the team, described the social circle question as an example of a famous concept called “the wisdom of crowds.” As James Surowiecki recounts in his book The Wisdom of Crowds, research has shown that when a large group of people is asked to guess something—like how many jelly beans are in a glass jar, or how much an item weighs—the group estimate is usually quite close even though individual guesses are all over the place. So it is with the neighbor question: The wisdom of crowds kicks in.

According to the report YouGov prepared for Théo, shared exclusively with The Free Press, the variation between how people said they would vote and how they thought their neighbors would vote was extraordinarily wide. In Michigan, for instance, the approximately 1,200 people polled chose Kamala Harris over Donald Trump by three percentage points, 49–46. But when asked who their neighbors would vote for, they chose Trump by 52–37. In Pennsylvania, the same number of people made the same choice: They were for Harris 49–47 but believed their neighbors were 52–37 for Trump. In Wisconsin, the traditional polling question had Trump squeaking past Harris 48–47 percent, but crushing her 52–36 percent in the neighbor poll. (YouGov added North Carolina to the three swing states, with similar results.)

For Théo, this confirmed his suspicions. The result of the neighbor question, he told me in an email, “aligned perfectly with the market pricing on Polymarket at the time, where the odds of Trump winning Michigan, Pennsylvania, and Wisconsin were all between 55 percent and 60 percent. This contrasted sharply with regular polls, which showed a 50–50 tie in these states.”

With the YouGov poll in hand, Théo used his finance background to do some math. If a comfortable majority of people thought their neighborhoods would go for Trump, there was a very low probability for Harris to win all three of the blue wall swing states. If Trump had a 60 percent chance of winning each of the three states, as both the Polymarket and his neighbor poll were showing, that meant Harris had, at best, a 10 percent chance of capturing the entire blue wall—which she had to snare in order to win the election.

Thus, Théo calculated that Trump had a 90 percent chance of winning. Not only that, he calculated that Trump had a 65 percent chance of winning the popular vote, too. With Grosse’s help, Théo had indeed uncovered the shy Trump voter.

(via YouGov)

So how did the pros get it so wrong? Harry Crane, a statistics professor and polling expert at Rutgers University, cited a concept called the herding effect, “where all the pollsters herd around each other.”

“Pollsters compete with other pollsters,” he told me. “If you turn out to be wrong, the only thing that really matters is that nobody else got it right. It’s not a coincidence that all the polls were within one or two points of each other.”

When Théo went to bed the night of the election, he told me he had no trouble falling asleep—“I was exhausted mentally and physically after two months of intensive research and enormous bets,” he said. He added that he was “both nervous and euphoric,” but as he closed his eyes, a famous Latin phrase popped into his head: alea jacta est. The die is cast. When he woke up the next morning, he was $85 million richer.

“I’m not a Trump supporter,” Théo told me in one of our conversations. “I didn’t have a political agenda. I did it to make money. When I looked around at possible trades, the election offered the greatest opportunity. The risk-reward was very good. And Polymarket was the only place where I could bet that amount of money.”

As they say on Wall Street, it was a no-brainer.

Two years ago, Polymarket signed a consent decree with the U.S. government that prevents it from taking bets from Americans. (The Commodity Futures Trading Commission takes the view that betting market contracts are essentially derivative contracts and must be regulated by the CFTC. Another prediction market, Kalshi, is contesting that designation in court, and was able to accept bets from Americans when an appeals court agreed with the company. The case is continuing.)

Even without American bettors, though, Polymarket had plenty of liquidity, which is what Théo needed to be sure he would be paid out if he won. Indeed, the day before the election, Polymarket’s trading volume passed $3 billion, with bets on Trump totaling around $1.3 billion. (The week after the election, the FBI raided the home of Polymarket’s CEO Shayne Coplan, taking his computer and phone, though he was not arrested. The FBI has not commented on the raid, but a Polymarket spokesman called it “obvious political retribution by the outgoing administration against Polymarket for providing a market that correctly called the 2024 presidential election.”)

In the weeks before the election, as Théo’s whale-sized bets moved the odds on Polymarket in Trump’s favor past the 60 percent mark, rumors began to circulate that someone was trying to manipulate the election by creating momentum for Trump. Théo was so annoyed he called The Wall Street Journal and denied that he was motivated by politics. (The Journal,which called him “the Trump whale,” also knew him only by his pseudonym.)

On the day after the election, he spoke to The Wall Street Journal again, revealing for the first time his reliance on the neighbor question without disclosing many other details. In his talks and emails with me, his big point was that the polling industry had become yet another elite institution that was corroding the faith of Americans, and it needed to improve. Although some pollsters have included the neighbor question, they have largely ignored its significance. Théo felt the pollsters should start making the neighbor question an important measure in gauging results.

“Having spent months analyzing U.S. media and polling in depth as a European, I would like to share my experience and perspective,” he told me. “From this external vantage point, I have seen how biased narratives can distort public understanding and create distrust among citizens.”

Having spoken to some experts, though, I’m not sure they will take his advice. Kristen Soltis Anderson of Echelon Insights downplayed the utility of the neighbor question, saying in an email it wasn’t a substitute for “a normal ballot question.” She pointed out that the popular vote was much closer than the neighbor question would suggest. But, of course, elections aren’t won on the popular vote in the U.S.—it’s the Electoral College that matters.

Ashley Grosse told me she thought the polling industry had actually done relatively well this year; after all, the popular vote was close, and most polls showed that. “It’s a net difference of two or three points,” she said. “It’s really not that bad.” But she understood that, given the size of Trump’s victory, most people didn’t see it that way.

When I asked Grosse how she explained the neighbor poll results, she grasped for an answer. Maybe, she said, it was because Democrats in Michigan often lived in heterogeneous neighborhoods, and they knew that many of their neighbors would be voting for Trump.

Then, she paused for a moment to ponder the big failure of her industry: “What you start seeing is a portrait of America that pollsters don’t see.” ~~ Joe Nocera, The Free Press

Regards,

Addison Wiggin,
Grey Swan

P.S. Markets will be closed tomorrow and we’ve got family hanging around for the Thanksgiving holiday. Thus, the Grey Swan will be back on Monday, December 2, 2024. Enjoy your own holiday weekend!

Until then, send your thoughts on Trump’s tariff wars, 2025 predictions, or what Polymarket bet is catching your eye here: addison@greyswanfraternity.com.