SINGAPORE: On prediction market Polymarket, hundreds of thousands of dollars are at stake in bets on what the highest temperature will be in Singapore on any given day.
Since the trend began in March, the size of this daily market reached an all-time high of over US$260,000 (S$330,000) in total volume of bets on April 13. On that day, the highest temperature recorded by Weather Underground (which the market uses as its measure) was 34 deg C.
Similar markets exist for users to bet on what the highest daily temperature will be at other locations, from Hong Kong to New York.
Prediction markets are platforms where users can bet on events, as varied as the outcome of sports matches to whether the United States and Iran will negotiate a peace deal.
Proponents of prediction markets, including the platforms themselves, have long argued that individuals placing a financial stake on an outcome can more accurately predict a result than experts or surveys.
However, as users embrace betting on topics such as the weather, the inherent contradictions of prediction markets – and whether it is just gambling under a more cerebral guise – become more apparent.
To Dr Yifan Feng, assistant professor at the National University of Singapore Business School, these temperature-related markets illustrate “how easily platforms can turn even small uncertainties into tradeable contracts”.
“That demonstrates the flexibility of the prediction market idea, but it also reminds us that not every market is equally informative or meaningful,” he says.
Singapore was among the first in the world to ban leading prediction market platforms Polymarket and Kalshi. Polymarket was blocked by the Gambling Regulatory Authority in December 2024.
Singapore users are barred from Kalshi and Polymarket. When users in Singapore attempt to access both sites, they instead see a warning that they are trying to access an illegal gambling site.
Those who circumvent the Government’s blocking may be considered as having committed an offence under the Gambling Control Act, according to an April 17 statement from the Gambling Regulatory Authority, Infocomm Media Development Authority and the police. A person convicted of gambling with unlicensed providers is liable for a fine, a jail term or both.
Outside Singapore, however, prediction markets are reaching their cultural peak.
In the first three months of 2026, the volume traded on leading platforms Kalshi and Polymarket came in at nearly US$60 billion, which was more than the entire volume traded in all of 2025, according to CNBC.
A much-touted example of prediction markets’ value is that prediction markets estimated that then-US presidential candidate Donald Trump was more likely to win the 2024 presidential election, breaking with most polls at the time which said the race was too close to call.
Dr Jia Yanwei, an assistant professor at the Chinese University of Hong Kong who studies prediction markets, compares it to a “financially incentivised opinion poll”.
However, prediction markets have been marred by criticism over insider trading and the ethics of betting on political events.
An analysis by The New York Times found that just hours before the US and Israel launched an attack on Iran in March, bets which totalled US$855,000 were made on Polymarket correctly predicting the strike.
These bets were anomalous and suggestive of insider knowledge because they broke with the majority of Polymarket users at the time who predicted that a strike would not happen on that day.
In March, Kalshi was sued by users over failing to pay out more than US$54 million to those who bet that Iran’s Supreme Leader Ayatollah Ali Khamenei would leave office before March 1.
After he was killed in an air strike, Kalshi reimbursed the fees paid by all users in the market instead of paying out winnings.
Rife criticism which followed can partly explain why users are now turning towards betting on arguably lower-stakes items such as daily temperature.
Dr Jia believes that markets have formed around daily temperature because the event is publicly verifiable by the pre-specified settlement date and the event itself is not easy to manipulate.
“Our markets reflect accurate, unbiased and real-time probabilities for the events that matter most to you,” writes Polymarket’s website. “Markets seek truth.”
However, the turn towards markets like daily temperature, how often billionaire Elon Musk tweets over a period of time and whether there will soon be proof that aliens exist, show the inherent contradictions of platforms like Polymarket.
For many users, prediction markets are often less about truth-seeking and more about speculation and thrill.
Polymarket currently predicts a 22 per cent likelihood that the US will confirm aliens exist by Dec 31.
Dr Feng notes that while there is a conceptual distinction between prediction markets and gambling, this becomes more blurry in practice.
Prediction markets can be different from gambling when they genuinely aggregate information and produce useful forecasts, which represent informational value that goes beyond mere betting, says Dr Feng.
“That said, when the topic is mainly recreational, or the main attraction is speculative excitement rather than forecasting itself, the gap becomes narrower,” he adds.
In his view, this is ultimately why regulation and market design matter.
For instance, some of the earliest examples of prediction markets existed in academic institutions.
Dr Jia points to the Iowa Electronic Markets (IEM), not-for-profit prediction markets operated by the University of Iowa that allow academics to stake real money against their predictions for election outcomes or economic indicators. Established in 1988, the IEM frequently outperformed many leading polls when it came to predicting electoral outcomes.
The Iowa Electronic Markets are an early predecessor to modern prediction markets, which have opened the door to retail investors – something that Dr Jia believes would ideally improve accuracy and efficiency.
However, as these retail investors can have vastly different motivations than academic traders, one should interpret their predictions with a grain of salt.
“With the right guardrails, prediction markets can serve a genuine informational role,” says Dr Feng. “Without them, they can start to look less like forecasting tools and more like another form of speculative betting.”
Researchers are increasingly calling attention to the addictive design of prediction markets, which use gamified features like leaderboards, streak-based bonuses and countdown timers to exploit the same psychological mechanisms that encourage problem gambling behaviours.
Trivial markets are often surfaced to users through push notifications as they are about to resolve, creating a continuous stream of action, said US-based Baruch College researchers as part of a commentary for the academic journal Science in April.
Seemingly frivolous markets (such as those predicting the weather) serve a function for prediction market platforms, says Dr Chen Yi-Chun, professor of economics at the National University of Singapore and director of the Risk Management Institute.
“They are short-horizon, easy-to-settle contracts that can generate user activity and liquidity,” he adds.
All of this manifests in the subculture that has formed around prediction markets, where there is an undeniable sense of “fear of missing out” (FOMO) in the air.
Polymarket and Kalshi turn high-stakes political events – from election results to the outcome of court cases – into spectacle through a never-ending stream of social media posts that is not unlike the play-by-play coverage of sports commentators.
Social media platform X, where most discussion of prediction markets is concentrated, is flooded with content by users claiming that betting on the weather can be a get-rich-quick scheme, often by following the behaviour of popular traders – whose betting histories are public on Polymarket.
These hype-filled posts are out of touch with the reality experienced by most users.
One analysis of Polymarket user data from 2022 to 2025, by researchers from the ESSEC Business School, HEC Montreal and the University of Toronto, finds a striking feature of the platform: Out of 70 million trades (totalling over US$20 billion in volume), the top 1 per cent of users captured 84 per cent of all trading gains.
About 70 per cent of users recorded a loss, with most losing trivial amounts. The median amount lost was US$1.
For the majority of users, all of the hype appears to serve primarily as a means to get more on board to increase the amount at stake in prediction markets. In other words, increasing the potential winnings of the minority who already earn the most. - The Straits Times/ANN
