With polls neck and neck and voters’ preferences shifting daily, predicting a clear winner at this stage is like trying to pinpoint a storm’s trajectory. One thing’s for certain: the outcome will be shaped by the candidates’ campaign strategies, each party’s mobilization efforts, and the whims of swing state voters come election day.
While some may simplistically predict the outcome of political contests with certainty, as a seasoned journalist I’ve come to realize that such declarations are often nothing more than wishful thinking. In reality, predicting election results is an inherently uncertain business, and even the most advanced forecasting models based on polling data can only offer a coin flip’s chance of accuracy. As I calculate the probabilities, Donald Trump’s chances are assessed at approximately 53% by one source, 51% by another, and an even higher 53.7% by yet another authority. While it’s close, that’s still a bit short of 50/50.
In recent months, an unconventional method of evaluating percentages has gained popularity: prediction markets. Individuals can now wager on the outcome of various events in real-money markets, including the winner of the presidential election, as well as other elections, sporting events, and movie awards. While established platforms such as and the have a longstanding presence in the market, this cycle has seen the emergence of two significant new markets.
The IEM (Intrade) was the first primary non-academic prediction market officially sanctioned in the United States, with attempts made by some to prevent it from allowing betting on presidential elections. As currently stands, the market for the election outcome suggests a 61% likelihood of Trump’s triumph, given the staggering $50 million invested thus far.
Despite its small size, Kalshi is a significant player in the global crypto market, facilitating bets from users worldwide. The Presidential Market boasts an impressive investment base of over $2.3 billion, with a substantial following among its loyal supporters. While Polymarket prohibits users from placing bets, individuals can still circumvent this restriction by leveraging a digital private community. As of Wednesday, the trend has begun to take shape, with some notable momentum. Currently, this poll suggests Trump enjoys a commanding lead, boasting an impressive 64.3% support.
As a long-time enthusiast of prediction markets, like many others, I’ve always been fascinated by their potential to accurately forecast various events. By providing a useful accompaniment to polling, they consolidate widely known information about candidate probabilities and, concurrently, serve as a deterrent against misinformation. Numerous discussions about politics are indeed cheap, and it’s often assumed that if you’re willing to stake your reputation on such claims, then those concerns should be taken seriously. When making a wager, you’re essentially backing your own money with your prediction, which tends to encourage more informed decision-making as you’re less likely to make rash predictions.
While these markets have not faced a test like the one anticipated for 2024 previously, observing their behavior with billions of dollars invested offers valuable insight into how they will operate at scale and to what degree they may be influenced to produce a desired outcome. Can we really believe this information?
Prediction markets for newcomers
For centuries, people have been wagering on election outcomes. Henry David Thoreau is famously known for his dissenting views on the importance of individual civil disobedience and nonconformity. Economists Paul Rhode and Coleman Strumpf have investigated instances where betting exercises dominated transactions within the inventory exchanges on Wall Street, their findings reveal. In 1916, the frenzied heights of the Wall Street markets were reached, with a staggering 12-month total of approximately $290 million in today’s dollars.
These markets operate unlike traditional sporting events. The neighborhood bookmaker sets her own odds, considering market trends but ultimately making autonomous decisions about pricing, influenced by personal insights and a deep understanding of the industry. Whenever you place a wager alongside someone, you are essentially betting against them, not competing with other bettors; a good bookmaker sets the odds to ensure a steady profit.
Unlike placing traditional bets, these prediction market platforms – including Polymarket, Kalshi, and PredictIt – enable users to purchase “shares” in outcomes, which payout if a specific event materializes at some point. The initial sale of shares takes place in the market, after which the buying and selling are confined to merchant-to-merchant transactions, rather than involving the home directly? The market maker’s primary function is to facilitate dealer transactions. Odds can shift dramatically and dynamically, fluctuating in real-time as new information emerges. It’s extra like an investment than a sports activities wager.
This construction implies that it’s crucial to pay attention to these numbers. When scrutinizing websites like Polymarket, one’s initial instinct is to focus on the headline figures – in this case, the impressive 61.7% odds favoring Trump – and mistakenly infer that these numbers accurately reflect the collective opinion of bettors at the site regarding a potential Trump victory.
What lies beneath its surface is another story altogether. The pseudonymous finance blogger argues that the primary objective of a prediction market is to attain a state of equilibrium where the value of a share in a particular candidate aligns with the demand for those shares, ultimately reflecting the collective wisdom of the market’s participants. The equilibrium implies a precise match between the perceived probability of a candidate’s success and the assumed value of that individual in the market, thereby facilitating accurate forecasting when markets are involved. While these two may diverge at times, especially when the market’s constraints are diverse.
Economists Justin Wolfers and Eric Zitzewitz have found that, in reality, the costs of prediction markets are relatively low, allowing for a form of confidence interval to be established, within which one can reasonably assume individual beliefs reside with high probability. Zitzewitz highlights the significance of market constraints, specifically those imposed by platforms like PredictIt, where traders’ maximum investments are capped at a relatively low level. These guidelines may inadvertently lead to significant pricing errors, as they greatly increase the difficulty of placing bets on low-probability events. “In a prediction market where constraints are nonexistent, as Zitzewitz famously noted during a phone call, ‘we’re much more likely to arrive at a price that reflects the average of opinions.'”
While awaiting further developments, caution against conflating prediction market prices with polling results, lest you replicate a now-amended error in The New York Times.
Pundits seeking a reliable gauge of election outcomes crave more than just polls; they require a sophisticated model, akin to those developed by esteemed entities such as FiveThirtyEight’s, The Economist’s, or Nate Silver’s.
The fundamental principle governing betting markets remains consistent regardless of the scenario, as it is based on the probability of a particular outcome. If shares of, for instance, Nvidia are misvalued, savvy investors can profit by wagering that the market’s perception of its worth will shift. In a market where tens of millions of individuals leverage trillions of dollars, it’s unlikely that there will be numerous obvious mispricings. Markets where novel concepts have recently emerged tend to be swiftly capitalized on by savvy investors seeking to capitalize on their newfound potential?
That’s a sound principle, albeit one that’s often tested by companies whose costs seem entirely at odds with their inherent value. The consensus on prediction markets is that they have performed well in practice. Wolfers, Zitzewitz, and Erik Snowberg scrutinized the proof thoroughly, finding it to be remarkably persuasive. Compared to traditional methodologies, such as surveys of professional forecasters, macro derivatives demonstrate a comparable level of accuracy in predicting financial outcomes like financial progress and inflation.
According to a study by David Rothschild, an initial analysis revealed that prediction markets were significantly more accurate than polls early on in the election, with their accuracy gap narrowing towards the election date. According to a study by a researcher at the Iowa Digital Markets, the longest-running prediction market in the United States, it was found that the margin of error in polls (1.91 times) exceeded that of markets (1.58), even in the later stages of the electoral cycle.
Firms of diverse sizes and types have taken the initiative to introduce internal prediction markets as a tool to inform their decision-making processes. Zitzewitz and Eric Cowgill from Google, Ford, and another undisclosed company collaborated. The study examined market performances by tracking subjects such as demand, product quality, meeting deadlines, and external factors. The findings were subsequently compared to those of internal consultants, revealing that the average error of the markets was a remarkably low 25% less than initially predicted.
In an unconventional setting, economists Anna Dreber, Thomas Pfeiffer, Johan Almenberg, and Magnus Johannesson proposed a novel framework where psychologists could wager on the replicability of specific psychological findings, essentially creating a marketplace for testing the robustness of research results. Investigations revealed that financial markets had demonstrated an uncanny ability to anticipate which research studies were likely to yield replicable results, thereby surpassing simplistic surveys conducted among consultants.
Economists Lionel Wengel and Robert Clements provided conclusive evidence in their 2012 paper, as demonstrated by the data presented. Researchers found that markets with maturity dates of 12 months or more were significantly flawed; often, they lacked sufficient liquidity to establish a market price or suffered from systematic biases in pricing. As the event drew near, markets became increasingly well-calibrated.
While presidential elections may pose a uniquely challenging environment for evaluating prediction markets. If a market is well-calibrated, when all its instances run, it should accurately predict outcomes, where results with a 70% chance occur 70% of the time, those with 20% odds happen 20% of the time, and so on.
To accurately evaluate this form of analysis, you require diverse market scenarios and reliable predictions. While presidential elections are singular events, historically, there hasn’t been significant election-betting market activity beyond a few instances, excluding notable exceptions in Iowa. Although Polymarket has only recently entered its second presidential cycle, it’s still premature to gauge its effectiveness in accurately predicting presidential elections, as there isn’t sufficient data yet to determine whether it’s well-calibrated in this regard?
Are the markets being manipulated?
While prediction markets often provide accurate forecasts, their reliability is not always guaranteed. Some people harbour reservations about these measures because they fear that such drastic steps will only serve to alienate the very communities we’re trying to bring closer together.
In the United States, the most significant opposition to online betting platforms such as Kalshi originates from a small but influential group of Democratic senators, spearheaded by Senator Jeff Merkley of Oregon. With the court’s greenlight for Kalshi, billionaires and behemoth corporations can now risk tens of millions on which political party claims the House or Senate majority, subsequently pouring massive sums into crushing opponents to safeguard their investments.
The story of the so-called “French whale” that surfaced on Polymarket seemed to validate concerns about misinformation spreading like wildfire. Four online sportsbooks offer notable accounts with wagers placed on Trump’s presidency, allegedly driven by motivations such as. While that’s not an enormous amount within the broader context of a $2 billion market, it’s still significant enough to raise eyebrows. Despite widespread coverage by various outlets, including The , , and, the most insightful reporting on this topic comes from an anonymous author known only by their pseudonym (they utilize a Ryan Gosling image from as their avatar), a prominent Polymarket dealer driven to uncover the identity of their counterpart.
After conducting research, Domer found that the organization had been overseen by a specific entity. Polymarket later revealed that all four accounts had been operated by a French national with “in-depth trading expertise and a financial services background.” “My best guess is it’s a wildly risk-loving, ultra-wealthy Frenchman who’s quite confident Trump will win,” Domer concluded.
It’s also possible that a wealthy French individual is trying to manipulate the market to influence Trump’s chances. Polymarket concluded that its investigation into the dealer found no evidence of him attempting to manipulate the market; instead, he made gradual, small bets, whereas an attempt to control the market would involve making large purchases or sales in a short time frame to rapidly move the price. ( additionally made this remark). The prediction market, Polymarket, found that this individual’s betting activity primarily reflected their personal opinions about the election, rather than objective information.
While it’s impossible to completely eliminate the possibility that the dealer was attempting to manipulate the market, it’s also plausible that they were skillfully trading in a way that wouldn’t have raised suspicion about their intentions? The concern is that someone could manipulate a prediction market by flooding it with capital to artificially inflate a particular candidate’s chances, thereby triggering media coverage and public perception that mistakenly elevates their standing.
The interconnectedness of markets means that a single action can have far-reaching consequences, potentially impacting multiple markets simultaneously. If Polymarket quotes Trump at 65% and Kalshi offers him 55%, it’s possible to generate risk-free cash by buying Trump at Kalshi and Harris at Polymarket. To do this, you’d spend only 55 cents to get a dollar if Trump wins, and 35 cents to get a dollar if Harris wins. In either case, you could spend 90 cents to get a dollar if either Trump or Harris prevails – a certainty with over 100% odds that one of them will emerge victorious. Market arbitrage is known for its ability to eliminate discrepancies between market prices over time.
Supporters of prediction markets argue that their mechanism is inherently resistant to manipulation, as the collective wisdom of participants reflects a more accurate assessment than any individual perspective. When someone injects a massive amount of capital into a market to artificially inflate its value and create a false sense of security, which contradicts the true reality beneath the surface. Websites like Kalsi and Polymarket, which operate similarly to hedge funds by allowing users to create and trade predictive markets.
Given the disparity between my perceived risk tolerance and Polymarket’s aggregated estimate, I wouldn’t wager my life savings on Harris’ prospects with confidence. While the potential gain is enticing in theory, I still harbor a significant 40% concern about losing everything. However, hedge funds primarily operate by placing speculative bets, leveraging significantly larger resources and possessing greater risk appetites. To prevent market manipulation, these regulations could effectively curb the ability of manipulators to swing markets. As trading wraps up around 10 pm ET on Tuesday, a single large-scale buyer () makes a significant move by purchasing more than $2.1 million worth of “Harris wins” shares, hinting at the emergence of at least one counter-whale willing to take the opposing side in this wager.
Historically, attempts to control markets have consistently ended in disaster for those who sought to manipulate them. In 2012, a mysterious figure spent heavily to manipulate the outcome of prediction markets and boost Mitt Romney’s chances, ultimately achieving their objective. In 2008, a notable anomaly caught attention, prompting a swift crackdown from the now-defunct prediction market, Intrade. These individuals suffered a resounding defeat.
According to Rhode and Strumpf’s research on financial history, it is observed that in the examined cases, a speculative attack initially drove prices, but these changes were swiftly reversed, and prices subsequently reverted to their pre-attack levels. There is scant evidence to suggest that political event markets will be subject to systematic manipulation over extended periods.
The notion of predicting French merchant intentions is inherently uncertain, leaving open the possibility that they may be driven by either strategic manipulation or genuine conviction in Trump’s election prospects, with massive financial investments potentially stemming from either scenario. What do you think could be driving the interest in prediction markets beyond popular outlets like FiveThirtyEight and Nate Silver, which have already democratized access to data-driven insights? There were two plausible explanations, he suggested. Markets may absorb information faster than fashion trends. As trends evolve, they identify nascent shifts in fashion, leading to increased value, followed by adjustments to their displays a few days later. While there’s an additional nuance that the market merely grants a premium to the Trump brand, what’s more likely is that individual merchants may simply hold a slightly stronger pro-Trump bias than the market as a whole.
Determining which perspective is accurate in real-time proves extremely arduous. Sethi aggregates a team of digital merchants that buy and sell shares in prediction markets, leveraging insights from esteemed sources such as FiveThirtyEight, Nate Silver, and The Economist to inform their investment decisions. He discovered that none of these merchants had any cash, instead misplacing funds to varying degrees on Polymarket and to an even greater extent on PredictIt’s other models. Don’t fashion trends perform less well than market indices in terms of their overall performance? Tentatively, sure,” Sethi writes. “Yet, this scenario may undergo a significant transformation in the coming days, with considerable alterations to its trajectory and outcome.”
Can collective wisdom trump individual bias? The true promise of prediction markets lies in their ability to aggregate diverse perspectives, fostering a more informed and nuanced understanding of complex issues. By incentivizing participants to share their insights, these markets create an environment where the crowd’s collective knowledge can outperform even the most skilled experts. Will this democratic approach revolutionize decision-making?
Some methods consider presidential elections to be one of the most underwhelming applications for prediction markets. Elections are often the sole catalyst that sparks widespread interest and fosters a greater desire for information surrounding the event’s predictions. We’ve already developed polling-based forecasts, which provide significant insights into the contest. The website features an exhaustive compilation of numerous election forecasts, drawing from polls, models such as FiveThirtyEight’s and Nate Silver’s, and the expertise of political scientists.
Shouldn’t our focus be on refining market trends rather than relying on speculative predictions? Perhaps not.
While the notion of leveraging markets to forecast unpredictable events has merits beyond an electoral context. Predictive analytics faces significant challenges when integrating information from diverse sources, but prediction markets offer considerable potential in addressing this issue.
Take pure disasters. Occurrences such as hurricanes, typhoons, and earthquakes have far-reaching humanitarian and financial implications, despite some statistical models predicting their likelihood. However, these forecasts do not provide actionable guidance for businesses, residents, insurance companies, and other stakeholders invested in disaster-prone regions. Might facilitate companies preparing for supply-chain disruptions and serve as a clear warning to residents that it is crucial to invest in more resilient housing options.
Betting on the place is fraught with uncertainty, its consequences as unpredictable as a hurricane. The potential danger lurking beneath is unsettling, yet perhaps no more unsettling than the actuarial tables that insurance professionals meticulously construct every day. It’s often more of a chore than an exciting activity, much like placing bets on sporting events or following a political campaign.
I worry that high-value prediction markets may not only disappoint brokers but also forego cash, instead: insurance firms, such as Lloyd’s, might pay for a market where meteorologists help forecast hurricane probabilities or the National Institutes of Health subsidize a market where medical researchers wager on trial outcomes, thereby gaining insight into which medications appear most promising and worth investing in.
Until now, prediction markets have been driven by two primary influences: a pragmatic assessment of their potential as a valuable analytical tool for understanding the world, and an innate instinct among gamblers to speculate on uncertain outcomes. As someone with a modicum of experience in this regard, I thoroughly understand the pleasure involved. When prediction markets finally yield actual social value, we’ll crave even more of the deliberate scrutiny pushing for change effectively.
