Tiger Research: Zuckerberg is starting to bet on prediction markets, while Asian countries still regard it as gambling

null Core takeaways

This article is written by Tiger Research. Prediction markets have grown into a mainstream industry, with monthly trading volume reaching $14 billion. Meta’s in-house “Arena” project also shows major tech companies’ recognition of it.

Its mechanism is very simple: if an event happens, the contract settles for $1; if it does not happen, it settles for $0. Therefore, its trading price is the real-time probability, and once the event ends, the result is confirmed by an oracle.

All of this is built on “skin in the game”: if participants make the wrong judgment, they lose money, which makes their information credible.

Western markets have incorporated prediction markets into the mainstream financial system, while limited participation in Asia is causing capital outflows, loss of information sovereignty, and lack of user protection.

Asia’s current task is not to block these markets, but to think about how to use this data responsibly within a regulated framework. Because avoiding the discussion effectively hands leadership over to overseas parties.

Prediction markets have found the product—market fit

For years, prediction markets largely stayed in the conceptual phase. Around 2020, the situation began to change: a few small projects started accumulating significant trading volume and gradually broke through regulatory barriers one by one, marking the industry’s formal maturation.

Since then, growth has accelerated. Currently, monthly trading volume is already above $14 billion, and the combined valuations of major platforms are about $40 billion.

Meta’s entry further proves it has moved beyond the early stage. The New York Times recently reported that Mark Zuckerberg personally led a team to develop a prediction market application called Arena. A large tech company investing such resources indicates that this industry has exited the experimental stage and built a proven business model.

Where did prediction markets originate?

Prediction markets are not a new concept. Before blockchain technology brought them to the mainstream and helped them form an industry, they were already used informally for decades in academia and finance.

Informal use

The term “prediction market” itself appeared later than its history. By the 1980s, this concept had various names—information markets, decision markets—until a 2004 economics paper fixed it as “prediction markets.”

But the underlying practice came much earlier than that name. The earliest form was political wagering on election outcomes. In 18th-century London coffeehouses, people bet on parliamentary scandals and prime minister changes, and the odds sometimes appeared in newspapers. In 19th-century New York, informal over-the-counter markets near Wall Street actively predicted presidential election results.

Academic use

Academia’s starting point was three economists at the University of Iowa in 1988. They were puzzled that polling failed to predict Jesse Jackson’s victory in the Michigan primary, so they designed a market where people could directly trade election outcomes. This later became the Iowa Electronic Markets (IEM).

In 1992 and 1993, the IEM received approval from the Commodity Futures Trading Commission (CFTC) for research use. Anyone who invested $5 could participate. From 1988 to 2004, the IEM outperformed traditional polls in about three-quarters of the time, becoming a “lab” for aggregating collective judgment into prices. Even so, there was no regulatory framework at the time that would allow it to operate as a public market.

Binary options

These early prediction markets are very similar to binary options in financial markets: contracts that pay out as a yes-or-no bet depending on whether the price crosses a specified threshold within a set time period. Their structure—settling for $1 if the event happens, otherwise $0—matches the logic of prediction markets exactly.

Binary options also entered regulated exchanges. Examples include the 2007 fixed-return options on the American Stock Exchange and 2008 S&P 500-based binary options on the Chicago Board Options Exchange. However, frequent fraud on offshore platforms led multiple major jurisdictions to ban retail sales of such products between 2017 and 2021. Even so, the basic structure of yes-or-no binary bets remains the logic foundation for how prediction markets operate to this day.

How do prediction markets trade today?

Today, prediction markets cover topics that are nearly limited only by imagination.

Sports events take up the largest share of trading volume. Thanks to continuous league and global schedules, the ongoing World Cup further boosts demand. Politics, geopolitics, and macroeconomics expand from indicators such as inflation data to private company valuation predictions—turning information itself into a tradable asset. Cryptocurrency and stock prices, along with some events driven by niche rumors, form a complete lineage from broad public interest to professional information needs.

Each contract settles in a binary yes-or-no manner. For example, for the 2028 Republican presidential nomination: if Vance is confirmed as the nominee, the contract that bets “yes” pays $1; otherwise, the contract that bets “no” pays $1.

The simplest way to understand this structure is to treat $1 as 100%. The contract pays $1 (100%) when the event happens, otherwise it pays $0, so the intermediate trading price naturally reflects probability. A 40-cent contract represents 40% of that dollar—meaning the market believes the event’s probability is 40%. The cent value can be read directly as a percentage (ignoring bid-ask spreads and trading costs).

Prices are formed through an order book, not by any central party deciding them. Buy orders (such as buying at 39 cents) and sell orders (such as selling at 40 cents) accumulate at each price level, and trades execute when the orders match. Prices (and implied probabilities) are generated in real time through the interplay of capital from many participants. Traders can also sell positions before expiry to lock in profit or cut losses—essentially converting beliefs about the event into money.

The result is recorded by an oracle. No matter how precisely the contract price reflects the market, after the event ends, someone still needs to determine “yes” or “no”; the oracle is the mechanism responsible for that judgment.

Oracles have two main ways of operating:

Decentralized oracles: proposers post a bond and submit a proposed result; if no one challenges it within the specified time, it becomes the final result. If challenges arise, it goes into a re-proposal process, and only after further challenges does it move to a vote.

Centralized: judgment criteria are set in advance. After the event ends, the exchange applies the official result directly and settles the market immediately. This fully hands the judgment power to a single exchange.

For example, on the Limitless platform, once the deadline passes, the result is finalized according to preset rules. The oracle service reports results by relaying real-world outcomes onto the blockchain. Most markets tracking crypto prices or stocks report automatically via Pyth Network, while sports or political custom markets are judged manually by the operations team within 24 to 72 hours.

At its core, a prediction market is an information system that compresses many participants’ views into a single number reflected by price, then determines whether the prediction is correct after the event ends according to preset rules.

The evolution of markets from games to information finance

Prediction markets have gone beyond simple wagering platforms and evolved into a core information-finance infrastructure—turning future uncertainty into real-time price information. The fundamental difference from traditional polls or expert forecasts is the “skin in the game” mechanism: participants use their own capital to take responsibility for their position.

In traditional approaches, expert mistakes have almost no reputational cost, and polls cannot filter out respondents’ indifference or strategic misreporting. Prediction market prices impose real costs for being wrong—wrong positions lose money—forcing participants to verify their beliefs with the most objective, up-to-date information and price them accordingly. This willingness to bear cost directly translates into market reliability.

This mechanism shows up in real data across multiple domains:

Accuracy in financial and monetary policy forecasting: A February 2026 study by a Federal Reserve economist explains why. Since 2022, prediction market interest-rate expectations ahead of Federal Open Market Committee meetings have been statistically highly consistent with actual outcomes, outperforming federal funds futures and Bloomberg consensus. The reason is that participants lose money immediately if they’re wrong, making them analyze available information more rigorously and price it accordingly.

Transparent probability estimates for political and election outcomes: In June 2026 local elections in South Korea, Polymarket correctly predicted 14 winners among 16 major cities and provinces. Where polls can only say places are “neck-and-neck,” prediction markets provide real-time probabilities bet in real money by participants—reflecting the combined judgment of many variables rather than a simple guess.

Response to market events and company valuation updates: In March 2026, when debates emerged about a cap on stablecoin interest income, the prediction market immediately priced the probability of Coinbase’s stock decline at 97.6%, treating it as a real-time risk indicator rather than a retrospective analysis. This shows participants’ sensitivity when their own capital is at risk. Academic research reached similar conclusions: a 2015 study of internal prediction markets for companies such as Google and Ford found that, compared with official forecasting models, prediction errors were reduced by as much as 25%, indicating that when insider knowledge combines with risk capital, forecasting accuracy improves.

Information asymmetry still limits outcomes. In a Venezuela case in January 2026, someone used confidential information for insider trading, exposing the real weakness. But this attempt to distort prices was identified and led to criminal prosecution, proving the market is designed to operate transparently and with accountability.

In areas where information is widely distributed, prediction markets are precise analytical tools; in areas where information is concentrated among a few, they are able to identify this kind of concentrated monitoring mechanism. Because participants’ funds are truly at risk, the prices generated by these markets constitute objective information for assessing the value of financial assets.

Prediction markets’ absence in Asian policy discussions

The nature and trajectory of prediction markets differ greatly across regulatory frameworks in different countries. In the US, court rulings have brought them into the regulated financial system, while in most Asian jurisdictions they are still viewed as a traditional gambling category.

In the US, lawsuits have resolved most regulatory uncertainties. The Commodity Futures Trading Commission tried to classify Kalshi’s election prediction contracts as gambling and sanction the platform, but courts ruled that election prediction is not a game of chance and regulators had no authority to ban it. This ruling changed the regulatory posture and became a decisive catalyst for traditional financial institutions, including ICE, Robinhood, and CME, to enter.

By contrast, in major Asian jurisdictions, the mainstream view still equates the binary settlement structure of prediction markets with traditional gambling. The dominant regulatory perspective is gambling control and public order rather than financial policy. Although approaches vary by country, prediction markets in the region are still mostly outside formal policy discussions, with India and Indonesia as exceptions.

This divergence ultimately comes down to whether regulators treat markets as financial innovation or as a social control issue.

Prediction markets are at a regulatory crossroads toward institutionalization

Prediction markets have become a core global financial and information infrastructure. There is now a significant gap between global trends and the rigid stances of Asian regulators. With technological and financial boundaries largely disappearing, attempts to confine new markets to old regulatory frameworks have inherent limitations. Current regulatory practices in major Asian jurisdictions have three key problems.

The first is the paradox of regulatory arbitrage

Prediction markets run on borderless digital networks. If a country blocks platforms or restricts access for users, it cannot eliminate underlying demand. Users will shift to unregulated offshore platforms, taking on greater risk. This leads to capital outflows from the jurisdiction, while regulators lose both market oversight power and related tax revenue, which in the long run weakens regional financial competitiveness.

The second is the loss of national information-infrastructure sovereignty

Prediction markets are advanced information infrastructure that transforms complex social problems into precise numerical estimates, rather than merely places for betting. Recent elections in Asia show prediction markets read public sentiment faster and more accurately than traditional polls. When they are excluded under the banner of regulation, the data that most reflects a society’s sentiment accumulates on foreign servers. The result is that foreign media and institutions understand local society more clearly than domestic analysts.

The third is the abandonment of user protection

Users are in a blind spot with no institutional safeguards. Policies that simply deny the market without thorough advance discussion only expose users to risk and push them out of the system.

The focus of discussion needs to shift completely.

The issue is no longer how to block this market, but how to use this data healthily within a legitimate framework. This shift in perspective requires dedicated research, but relevant discussion is still very limited.

In this area, Limitless Research is filling the gap by processing prediction data from Asian markets such as South Korea and Japan into information assets. More participants will be needed in the future to take on the role of building a healthy data ecosystem.

Regulation should not be a dam that blocks the flow of water, but a channel that guides it correctly.

What Asia needs now is not stricter enforcement, but to start forward-looking discussions to respond to this change. Pushing already-executed trades into the shadows is the worst policy. Ongoing efforts are needed to incorporate this into the mainstream framework through constructive discussion, to establish transparent oversight mechanisms, and to return the data generated during the process as national and social assets.