Prediction Markets Under Bias

Original title: What Most People Get Wrong About Prediction Markets

Original author: Jeff Park, Bitwise

Original compilation: Saoirse, Foresight News

Last week, two media organizations, Axios and MorePerfectUS (MPU), successively explained to the public what prediction markets are. Dan Primack of Axios tried to build a neutral space for multi-sided discussions with the founder of the Kalshi platform, even though his own stance is not hard to detect; while Trevor Hayes from the other outlet took a clear stance, deliberately amplifying contradictions, viewing prediction markets as a kind of social vulnerability.

To be frank, I agree with parts of both sides. I’ve spent years working at the intersection of Wall Street and the crypto industry, and I fully understand the public’s growing unease about the increasing intensification of over-financialization—this trend has already given rise to a gambling culture that is now seen as a public health crisis. But these journalists generally fall into a misconception: they reach conclusions too hastily, trace causes backward to find the original culprit, and mix together issues such as insider trading, online casinos, and gambling addiction into an overly simple, one-sided narrative.

But that is precisely the biggest misunderstanding the public has about prediction markets: setting aside the various harms caused by over-financialization from 0DTE expiry options, swap-type ETFs, and Meme stocks, prediction markets themselves should be recognized. They can grant individuals a high degree of autonomy in choosing, uncover the truth, and their decentralized nature in itself has legitimate value.

In what follows, I will break this problem down layer by layer.

The blurry boundary between investing and gambling depends only on whether participants’ strategies have positive expected value (+EV), and has nothing to do with whether the market itself is a deterministic mechanism or a randomness mechanism. In other words, it’s people who define the distinction, not the game itself.

Let’s break it down in detail. I noticed that in MPU’s coverage, Trevor Hayes often opens his arguments with a single assumed premise: “Since prediction markets obviously belong to gambling…,” as if this were an established fact that requires no justification. And that foundational assumption is exactly what needs to be re-examined.

Over the past twenty years, the most significant trend in finance has been the continuous erosion of the boundary between investing and gambling. The data speaks for itself:

• 60% of US stock trading volume comes from high-frequency trading, and this field has been formed into an oligopoly by Jane Street and Citadel;
• Passive ETFs account for more than 90% of total ETF assets under management (active investment strategies are only now slowly bouncing back);
• The average holding period for US stocks has shortened—from about 9 years in the mid-1970s to only about 6 months in 2025.

Meanwhile, over the past decade, daily trading volume in US stocks has more than tripled, and the driving force is still algorithmic trading. In addition, there is another irreversible trend: in 2025, the scale of retail trading surpasses $5 trillion, an increase of roughly 50% compared with 2023.

But why do very few financial commentators accuse stock trading itself of being gambling? The reason is that the public generally assumes that stock picking investing is not gambling, because people subconsciously believe it requires professional ability. This is crucial: people unfairly lump skill-based contests and pure probabilistic contests together under the broad label of gambling. For example, slot machines and poker are both called gambling, but they are worlds apart: slot machines are purely luck-based, with negative expected value; while poker relies on technical strategies and can completely achieve positive expected value.

To put it bluntly, the standard for distinguishing investing from gambling depends only on whether the strategy can achieve positive returns, and has nothing to do with the game itself—whether it’s a deterministic arbitrage game, a fixed-outcome pattern like a slot machine, or a random-fluctuation pattern like stock picking or poker.

Prediction markets are similar to poker: they are stochastic games with an embedded deterministic logic. Whether they count as investing or gambling is determined entirely by the participant: whether you’re a person with high autonomy and strong professional ability, or low autonomy and low cognitive level, or somewhere in between. This leads to a second question: if gambling is understood as human-driven speculative behavior, then how do these markets actually operate, and where does liquidity come from?

The other side of speculation is risk hedging (insurance).

All financial innovations, at the moment of their birth, are viewed as gambling. Early stock markets were rife with rampant insider trading, and in the futures markets, European dollars even became tools for politicians to conduct insider trading; today’s commodity trading also can’t be strictly defined as insider trading using traditional definitions—everything is like this. The root cause is that speculation and hedging are two sides of the same coin. This is a zero-sum game; the core is to complete the transfer of risk; and not all information is naturally generated by private entities.

This brings us to the most common challenge critics raise about prediction markets: some markets have only pure speculative attributes and cannot create value for society, so they shouldn’t exist in the first place. The example they most often cite is sports betting. In the public’s ingrained perception, sports are entertainment, and betting on entertainment has no social value.

But this view itself is wrong. Entertainment is social consumption by human beings, and arguably one of the core sources of human happiness. More importantly, entertainment itself is an economic activity with a two-sided market structure. The global sports industry generates more than $50 billion in annual revenue; together with the surrounding industrial chains such as media, equipment, apparel, and sports nutrition, the total scale is estimated to exceed $1 trillion. Take Nike as an example: it invests huge sponsorship money into teams and athletes, which in turn requires allocating capital and hedging risk based on event results and athletes’ performance. Simply because the United States hasn’t opened official, compliant markets for sports betting, people equate sports betting like that with a casino, completely ignoring its potential financial value.

The core value of derivatives is risk transfer. This is the underlying logic of all insurance products and asset securitization. And to achieve risk hedging, there must be speculators on the other side; in open and transparent markets with no administrative interference, this structure is irreplaceable. In fact, most problems with insurance systems occur because government interventions distort true market pricing. Insurance and securitization are also among the greatest financial innovations in human history for improving capital efficiency.

Yet one core question still can’t be avoided: how do we define whether something is a social harm or a financial service with practical value? How should we build a system to categorize events? Next, I will lay out the article’s final core argument.

Prediction markets differ from other derivatives in two key attributes: precision and limited expiration.

Let’s go back to market making fundamentals to understand this. Ordinary financial markets rely on a central limit order book to provide liquidity, and the underlying assets have perpetual value. Prediction markets are entirely different: once the corresponding event settles, market liquidity drops to zero directly, and both sides close positions and exit. A binary 0/1 settlement outcome makes conventional dynamic hedging strategies completely fail, posing a major challenge for professional market makers.

More importantly, prediction markets are odds-based markets rather than price-based markets. This means that within a 50% probability range, even small fluctuations generate much higher liquidity than fluctuations in an extreme probability range of 98%—in the latter case, for every point of change in odds, the settlement cost increases exponentially. Therefore, liquidity cannot be sustained by bid-ask spreads alone; traders of fixed-income derivatives understand this deeply (for example, a 10 basis point move at a benchmark rate of 4% versus a 10 basis point move at 0.5% are vastly different in meaning).

In summary, in event markets where the information gap is extremely large and participants hold absolute informational advantages, professional market makers almost never show up to provide liquidity. This also means that the scenario critics describe—“insiders harvest windfall profits using their information advantage”—has extremely limited profit potential in most situations. The market itself will naturally filter out the events that the public truly cares about.

For example, I know very well whether my next podcast will feature a Bitwise branded sweatshirt, but the corresponding prediction market will basically not generate any liquidity. One major concern that the public has about insider trading is that insiders will earn massive profits; but in reality, this isn’t the case. Events that are obscure and have no value inherently lack liquidity, and the market’s liquidity itself has already priced in the value of information. From this, a reasonable event grading system will naturally form.

So what is the true value of prediction markets—enough to cover their potential risks?

The precision mentioned earlier is its most precious trait. Today’s global finance is overly financialized: asset prices are increasingly influenced by capital flows and technical trends, drifting away from fundamentals and from facts themselves. Prediction markets are one of the few tools that can directly anchor prices to facts and remove unnecessary interference.

In the future, if you have a fundamental judgment that Tesla’s revenue will exceed expectations, instead of directly buying or selling Tesla stock (the stock price will still be affected by factors unrelated to the fundamentals, such as macro conditions, broader market movement, and capital), you might place a bet in the prediction market. If you want to forecast non-farm employment data, you don’t need to trade European dollar futures or stock index futures—just participate in the relevant prediction market directly. This precision attribute will truly reward in-depth research, professional judgment, and genuine informational advantages.

A large number of external voices criticize prediction markets as a way to extract losses from financially naive ordinary people, arguing that participants generally lose money and that this poses a social hazard. But the facts are exactly the opposite: prediction markets have the fairest mechanism, rewarding professional investors with informational advantages. And because it doesn’t involve a house platform taking fees, it’s completely different from Las Vegas casinos: casinos eject consistently profitable positive-expected-value players, while prediction markets welcome all participants who have informational advantages.

Citadel Securities and Charles Schwab have both announced that they will move into prediction market business. Are these giants exploiting vulnerable groups? Obviously not. They understand better than the public: speculation and hedging are two sides of the same coin, and one party’s risk exposure is the other party’s profit opportunity.

Why do authoritative media fear this true prediction market

(Note: Gray Lady refers to the “New York Times.” In its early days, the New York Times print edition was dominated by gray-toned paper, black-and-white layout, and very few color images; the page design was solemn and dark. Combined with a rigorous, conservative writing style, formal and weighty wording, and the steady temperament of a long-established authority media outlet, readers and people in the industry respectfully call it Gray Lady. Here, it broadly refers to long-standing mainstream American public-opinion benchmarks, the gatekeeping information mouthpiece for American elites, and the traditional large media that holds influence over the discourse.)

By now, you should understand that under reasonable regulation, prediction markets have enormous potential. As long as the returns outweigh the risks, issues like gambling addiction and negative social effects can all be addressed through solutions. But there is still one key question left: would insider trading involving major public events create an unfair situation where private monopolies profit?

This issue is very complex, and I will write a separate article to answer it in detail. For now, I want to share a piece of thinking—and a book I recently read: Ashley Rindsberg’s “The Gray Lady Winked.”

The book lays out the systematic dereliction of duty by this authoritative media over decades, and it is not the result of accidental mistakes: concealing Stalin’s Great Famine, glorifying Castro’s rise, pushing rumors about Iraq’s weapons of mass destruction on a large scale, and downplaying the risks of the resurgence of Nazi Germany. The New York Times has always relied on information channels, ideology, and the needs of institutional self-protection to distort the spread of truth.

Once you understand this book, you will see that media bias is not simply a matter of a left-versus-right stance dispute, but rather a deeper structural problem: top authoritative institutions actively manufacture social consensus, and then later whitewash their own reporting errors.

Returning to the original topic: Axios and MorePerfectUS are not industry neutral. This is also why, in the future, there will be more and more media outlets attacking prediction markets. But you need to be clear about this: the reasons they reject prediction markets are precisely the reasons you should support them.

Information already has a price, and there’s no need to debate this. I have always believed: the opposite of false information is never absolute truth; the opposite of false information is information under official control.

The real debate has never been about pricing information itself, but about who has the right to define information, who can profit from it, and whether information has already been monopolized and exploited before the public even knows about it.

When insiders hoard asymmetric information, making money is secondary; the more core issue is a struggle over power. Profit harvested by exploiting the public’s informational disadvantage is used to manipulate public opinion, manufacture false narratives, and hijack the entire truth dissemination system.

Therefore, the core of opposing insider trading has never been about economic efficiency—it has always been about equality in the right to access information: some people trade using exclusive information, while ordinary people can only access filtered and permitted information.

After you understand this layer, you won’t be pessimistic about prediction markets. You will only look at the world through a more precise and rational lens. That is also why I have always firmly believed this: supporting prediction markets is itself an idea with strong democratic value.