Antitrust Principles for Prediction Markets

Evan Miller is a Partner at Vinson & Elkins LLP. He focuses his practice on antitrust investigations and counseling with particular emphasis on emerging technologies. The views expressed in this commentary are solely his own and do not necessarily represent the policies or views of Vinson & Elkins LLP.

Introduction

Businesses are increasingly recognizing prediction market platforms such as Kalshi and Polymarket as sophisticated mechanisms for aggregating dispersed information and generating real-time probabilistic forecasts. [1] Hedge funds, political risk analysts, and AI developers have begun to integrate prediction market outputs into their systems. [2]

As these markets expand in scope and influence, a natural question arises as to whether concentration within various layers of prediction markets could confer monopoly power over “truth” as a downstream input. The idea of a “truth monopoly” is intentionally provocative and equally imprecise. Antitrust law does not recognize monopolization of “truth,” ideas, or beliefs as a cognizable offense. To the extent the concept has antitrust traction, it is most likely through the lens of Section 2 of the Sherman Act and a firm’s refusal to provide its competitors with access to an essential input.

There are two potential inputs at issue when thinking about prediction markets, each sitting at different layers of the prediction market technology stack.

First, there is the prediction data layer: the probabilistic signals produced by trading activity, typically reflected in contract prices and related metadata. This layer is informational. It improves with scale and liquidity, is easily reused across contexts, and may become embedded in downstream systems, including AI models. If the market coalesces around a particular feed, dominance can arise even absent exclusivity.

Second, there is the resolution layer: the mechanism that determines which outcome is deemed to have occurred for purposes of settling prediction market contracts. This layer is adjudicative. It converts reality into a settled contractual fact. If widely relied upon, resolution authority can resemble other verification infrastructure—financial benchmarks, clearing and settlement systems, certification bodies, and certain forms of standards administration.

The most significant antitrust risks often arise not from dominance over any specific layer in isolation, but from the interactions between interconnected layers. For example, where control over resolution is combined with leverage over data and downstream applications, antitrust risk may arise.

This commentary applies antitrust principles to these two layers of prediction markets. Importantly, it focuses on conditional risks rather than presumed outcomes. Prediction markets remain relatively early in their development, and it would be premature to assume concentration is inevitable or that any currently operating firm has engaged in exclusionary conduct. But as reliance grows, especially in institutional and AI contexts, market structure choices made today can shape contestability tomorrow.

This commentary proceeds as follows. Part I provides a primer on prediction markets and the role of resolution. Part II frames one aspect of the “truth monopoly” concern as a question about power over the resolution layer. Part III addresses the separate possibility of dominance at the prediction data layer, with particular attention to downstream reliance and AI. Part IV evaluates antitrust theories of harm that could arise from resolution power, data power, or their combination—including de facto standard-setting dynamics, raising rivals’ costs, and vertical leveraging. Part V cautions against overcorrection and argues for proportionate, evidence-driven engagement.

I. What Prediction Markets Do and Why Resolution Matters

Prediction markets are platforms that allow participants to trade contracts, the values of which depend on the outcome of a future event. Each contract typically pays a fixed amount if a specified event occurs, or nothing (or a different amount) if it does not. [3] The market price therefore approximates a collective assessment of probability.

For example, a contract that pays $1 if inflation exceeds a certain threshold may trade at $0.35, implying a probability of approximately 35%. Participants who believe the true probability is higher may buy; those who believe it is lower may sell or short. Through trading, dispersed information and beliefs are aggregated into a live price signal.

Unlike opinion polls or expert forecasts, prediction markets rely on financial incentives rather than self-reported beliefs. Participants put money at risk, which tends to reward accuracy and penalize error over time. This feature has led economists and policymakers to view prediction markets as potentially powerful tools for information aggregation in environments otherwise characterized by uncertainty and incomplete data. [4]

A defining feature of prediction markets is the resolution process—the mechanism by which the platform determines whether the event has occurred and settles contracts. [5] Resolution typically relies on predefined criteria and authoritative sources, such as official election results, government statistics, regulatory determinations, or widely recognized publications. [6] Resolution can be centralized or decentralized. A centralized resolution mechanism vests final adjudicative authority in a single entity. That entity determines whether the event occurred and resolves disputes. A decentralized resolution mechanism distributes adjudicative authority across token holders, stakers, or a decentralized oracle network. In decentralized resolution mechanisms, token holders vote on outcomes. Because contracts ultimately settle based on this determination, resolution is central to market integrity. Clear rules, reliable data sources, and transparent governance are essential to maintaining confidence.

Prediction markets operate under different legal and regulatory frameworks depending on jurisdiction and design. In the United States, certain platforms operate as regulated exchanges subject to oversight by the Commodity Futures Trading Commission. [7] Other prediction markets, including some operating internationally or on decentralized infrastructure, may not be subject to the same oversight. [8] Regulatory status can affect who may participate, what contracts may be listed, and how disputes are resolved.

II. The Resolution Layer as Input

As discussed in the Introduction, the phrase “truth monopoly” is legally imprecise. Antitrust law typically does not condemn a firm for influencing what people believe. To ground the concept in antitrust principles, it must be understood narrowly: as a concern about market power over a mechanism that is necessary to determine operative facts for market settlement.

A. Resolution Authority as a Critical Input

Prediction market resolution mechanisms do not merely disseminate information; they determine outcomes that have contractual, financial, and downstream economic consequences. That adjudicative function distinguishes them from entities engaged primarily in expressive activity (media outlets, publishers, or ordinary platforms for speech). A widely adopted resolution mechanism can become upstream infrastructure—without access to resolution, downstream trading and settlement cannot occur on comparable terms.

A useful analogy is credit rating agencies. Credit ratings do not merely inform investors; they determine regulatory treatment, capital requirements, investor eligibility, pricing, and liquidity across multiple markets. Antitrust authorities have long recognized that this adjudicative role can confer durable market power even in the absence of direct participation in downstream transactions. In 2021, the U.S. Department of Justice Antitrust Division expressed concern that S&P’s proposal “to automatically lower its ratings for assets in insurance company investment portfolios rated solely by S&P’s competitors” could raise “significant concerns that the Sherman Act has been—or will be—violated and warrant additional scrutiny by the Antitrust Division.” [9] Doing so, the DOJ argued, would disincentivize companies from using rating agencies other than S&P, raise barriers to entry and expansion, and insulate S&P from competition. [10] In 2023, the DOJ reiterated related concerns about the treatment of competitors in S&P’s rating methodologies. [11]

The relevance of this analogy is that control over a critical input can, under certain conditions, translate into durable market power. As in the S&P example, antitrust concern arises where control over resolution is used to disadvantage rivals or foreclose competition—either among prediction markets themselves, among resolution providers, or in downstream markets that depend on verified outcomes. Antitrust risk may arise, for example, if a resolution mechanism strikes an exclusive deal for data essential to settle contracts (e.g., league-approved sports data).

B. Distinguishing Resolution Power from “Speech” Narratives

Recent statements from the FTC and DOJ suggest a broad view of competitive harm that encompasses diversity and openness in content, not just price and output effects. In Children’s Health Defense v. WP Co., the Antitrust Division’s statement of interest urged courts to recognize that viewpoint competition can fall within the ambit of the Sherman Act. [12] In that same vein, FTC Chair Andrew Ferguson sent Apple a warning letter regarding allegations that Apple News “systematically promoted news articles from left-wing news outlets and suppressed news articles from more conservative publications.” [13] Ferguson warned that Apple’s conduct potentially violates Section 5 of the FTC Act’s prohibition on unfair or deceptive acts or practices. [14]

Prediction markets occupy this same conceptual space: they aggregate and elevate diverse assessments of future events, and the structure of their resolution mechanisms determines which viewpoints are recognized as authoritative. If a dominant resolution mechanism skews outcomes or forecloses access by oracles relying on alternative information sources, they could diminish informational diversity in ways analogous to the viewpoint discrimination concerns now on the agencies’ radars. This suggests that antitrust scrutiny of prediction markets may eventually encompass not only traditional competitive effects, but also the competitive dynamics of informational ecosystems and the availability of diverse prognostic signals.

Whether courts will formally adopt viewpoint competition as a standalone antitrust injury is unclear. But enforcement leadership’s focus on information diversity and competitive access to ideas positions prediction markets as a uniquely relevant class of intermediaries.

C. Why Network Effects Are Weaker at the Resolution Layer

Liquidity and participation improve the accuracy and stability of the prediction signal, but they do not necessarily make a single resolver intrinsically “better” at adjudicating what happened. Resolutions can be credible, auditable, and reliable at multiple venues, particularly where outcomes are verifiable from public sources and rules are transparent.

This does not eliminate the possibility of resolution concentration, but it does change the discussion from quantitative scale to qualitative characteristics of the resolution mechanism. In other words, the most plausible drivers of resolution-layer dominance may be credibility, conservatism, and finality—not pure scale effects. That distinction matters for antitrust analysis, because it bears on whether a resolution mechanism is truly non-replicable or instead contestable through entry, governance innovation, or interoperability.

III. A Separate Bottleneck: Prediction Market Data as an Upstream Input (Especially for AI)

As prediction markets mature, their competitive significance may revolve around their role as upstream suppliers of probabilistic data, not merely as standalone trading venues. This shift will accelerate as AI systems and enterprise risk tools incorporate market-set probabilities into their workflows.

Prediction market outputs differ from many other data feeds in three respects. First, they combine price, probability, and financial commitment, producing a signal that is quantitative and incentive-weighted. Second, they often include resolution-related metadata—information about event definitions, sources, timing, and settlement conditions—used by downstream systems for validation and auditability. Third, they are frequently consumed continuously rather than episodically, making them part of a continuous pipeline of live data rather than a static reference source.

These characteristics make concentration more plausible at the data layer than at the resolution layer. Prediction data exhibits the potential for strong scale-driven network effects: more participation improves the signal, a better signal attracts more reliance, and reliance draws further participation. As AI systems and enterprise tools train, calibrate, or benchmark against a particular feed, substitution may become costly. Switching may require retraining models, revalidating assumptions, rebuilding analytics pipelines, or accepting discontinuities in historical time series.

Similar dynamics exist in financial benchmarks, credit ratings, index licensing, and other informational intermediaries. In each case, concern did not arise because information was literally unavailable elsewhere, but because market participants coalesced around a particular source, rendering alternatives impractical for many use cases. Prediction market data could follow a similar trajectory—particularly if downstream reliance becomes one-directional rather than multi-sourced.

IV. Antitrust Risks and Theories of Harm: Resolution Power, Data Power, and Their Interaction

Antitrust law does not condemn success or scale; it intervenes where market power combines with conduct or structure that impairs competition on the merits. With that premise, several established frameworks may be relevant as prediction markets design and grow their platforms.

A. Bottleneck Input Control

If a particular resolution mechanism becomes effectively unavoidable for certain categories of contracts—or if a particular prediction feed becomes unavoidable for downstream users—regulators may assess whether rivals face practical barriers to entry or expansion. For resolution, the inquiry likely would focus on replicability, transparency, and whether credible alternatives can gain adoption. For data, the inquiry tends to focus on switching costs, path dependence, and whether rivals can offer functionally interchangeable signals at comparable cost.

B. Raising Rivals’ Costs Through Frictions Rather Than Denial

Even without outright refusal to deal, a dominant provider can impose technical, contractual, or economic frictions that disproportionately burden competitors or downstream users. In data-intensive markets, antitrust scrutiny often centers on indirect exclusion—access terms that are formally available but functionally discriminatory.

At the resolution layer, this could involve dispute processes, timing conventions, or definition choices that systematically advantage affiliated products. At the data layer, it could involve licensing structures, API constraints, latency, bundling, or data formats that entrench dependence.

C. De Facto Standard-Setting Without Formal Standards Bodies

Resolution definitions, identifiers, timing conventions, and dispute processes may become industry norms without a formal standards organization. Standardization can be efficiency-enhancing—it reduces transaction costs and supports interoperability—but control over standards has long been recognized as a potential source of market power when participation is not meaningfully open or when standards are insulated from competitive challenge.

This risk can arise at both layers: resolution conventions can become the “standard” form of adjudication, and data schemas can become the standard forecasting interface for downstream systems.

D. Vertical Leverage and the Compounding Risk of Layer Integration

The most salient concerns may arise if a platform combines (1) the adjudicative authority of resolution with (2) dominance over probabilistic data and (3) downstream products that depend on the same inputs. Scholars and enforcers have analyzed this combination of “gatekeeping” power and downstream leverage in the context of digital platforms. [15]

Vertical integration is not suspect per se and may generate efficiencies: faster settlement, better integrity controls, improved auditability, and more reliable products for users. But regulators may examine whether access to upstream inputs is neutral or whether the platform’s own services receive preferential treatment that distorts competition.

E. Countervailing Factors: Replicability, Interoperability, and Market Discipline

Several characteristics are likely to constrain both resolution and data dominance.

Replicability: Resolution mechanisms are often replicable in principle. Inputs—public sources, event definitions, methodologies—are typically available to all. Building trust is costly, but not necessarily dependent on exclusive rights to a scarce resource. Courts are generally reluctant to impose duties to deal where rivals can replicate an input, even if doing so is expensive.

Interoperability and multi-homing: Traders and downstream users can often access multiple platforms. APIs, standardized formats, and public settlement records can enable aggregation across sources. Where multi-sourcing is practical, no single platform becomes unavoidable, and competition shifts toward accuracy, reliability, and service quality.

Market discipline: Inaccurate or biased resolution creates reputational harm and may reduce liquidity. Because outcomes are often ultimately observable, platforms face ongoing pressure to preserve credibility and procedural integrity.

These constraints do not eliminate antitrust risk, but they counsel against assuming inevitability of market power and abuse of that power. They also suggest that enforcement, if warranted, should focus on specific conduct and practical foreclosure rather than abstract structural concerns.

V. The Case for Proportionate Engagement

History suggests that premature or overly prescriptive intervention can be harmful, especially in markets still forming and whose value depends on innovation, experimentation, and credibility.

A. Standardization as a Feature, Not a Bug

Some degree of standardization is inherent in markets that aspire to scale. Prediction markets require shared understandings of event definitions, settlement timing, and outcome verification to support liquidity and price discovery. Without coordination, markets risk fragmentation, arbitrage, and loss of confidence.

Antitrust doctrine recognizes this tension. The appropriate inquiry is not whether standardization exists, but whether it remains contestable. In other words, whether alternative approaches can emerge and gain acceptance over time.

B. The Limits of Structural Remedies in Informational Markets

Structural remedies, such as decentralization or shared oracles, would represent a significant departure from established antitrust practice, particularly in informational markets. U.S. courts are generally reluctant to impose duties to deal or structural separation absent clear evidence of exclusionary conduct and durable harm. [16]

In prediction markets, where trust and finality are central, structural overcorrection could weaken the qualities that make markets valuable—pushing activity toward less transparent or less regulated alternatives.

C. Antitrust as a Graduated Framework

Modern enforcement increasingly emphasizes graduated engagement. Agencies may begin with transparency, monitoring, and conduct-based inquiries, escalating only if competitive harm becomes evident and persistent. In prediction markets, this suggests scrutiny would most plausibly focus on observable signals: discriminatory access to resolution or data, exclusionary vertical integration, or insulation of de facto standards from competitive challenge. Absent such conduct, scale alone is unlikely to justify intervention.

D. Institutional Humility

Prediction markets sit at the intersection of finance, technology, and information governance—areas where regulatory overreach risks unintended consequences. Recognizing the limits of antitrust does not diminish its relevance. It underscores the value of evidence-based, proportionate responses that respect the dynamic nature of emerging markets.

Conclusion

The idea of a “truth monopoly” captures a genuine intuition: as prediction markets grow in influence, control over the processes that settle contracts can confer meaningful economic power. Translated into antitrust terms, however, the relevant risks are narrower. They arise from two distinct layers: the resolution layer, which adjudicates outcomes for settlement, and the prediction data layer, which supplies probabilistic signals that can become embedded in downstream systems, including AI.

These layers raise different competitive questions and are driven by different economic forces. Resolution authority resembles verification and coordination infrastructure; prediction data resembles other informational inputs prone to scale effects and path dependence. The most significant concerns may arise where these layers are combined and leveraged into downstream markets.

Antitrust’s role here is not to regulate truth, manage discourse, or preempt success. It is to preserve the conditions under which resolution infrastructure remains contestable and subject to competitive pressure—so that prediction markets can deliver their greatest value: decentralized information discovery, credible aggregation, and continual improvement over time.

[1] Justin Wolfers & Eric Zitzewitz, Prediction Markets, 18 J. Econ. Persp. 107, 108 (2004) (“In a truly efficient prediction market, the market price will be the best predictor of the event, and no combination of available polls or other information can be used to improve on the market-generated forecasts.”).

[2] To name a few: Kalshi has partnered with CNN to integrate Kalshi’s prediction market data into CNN’s reporting; Polymarket has a similar partnership with the Wall Street Journal; Google has integrated data from Kalshi and Polymarket into its AI-powered “Deep Search” feature to provide users with market-implied probabilities on economic and political events. Ryan Whitwam, Gemini Deep Research comes to Google Finance, backed by prediction market data, Ars Technica (Nov. 6, 2025, at 15:39 ET), https://arstechnica.com/google/2025/11/gemini-deep-research-comes-to-google-finance-backed-by-prediction-market-data/.

[3] See About Kalshi, Kalshi, https://kalshi.com/about.

[4] Jason Wingard, The Polymarket Effect: How Prediction Markets Are Beating The Experts, Forbes (Nov. 19, 2025, at 19:28 ET), https://www.forbes.com/sites/jasonwingard/2025/11/19/the-polymarket-effect-how-prediction-markets-are-beating-the-experts/.

[5] See How Are Prediction Markets Resolved?, Polymarket (Jan. 11, 2026), https://docs.polymarket.com/polymarket-learn/markets/how-are-markets-resolved; How UMA Works, UMA, https://uma.xyz/#how-it-works.

[6] For example, the outcome of “2028 Democratic presidential nominee” on Kalshi will be verified by democrats.org, the official website of the Democratic Party. 2028 Democratic presidential nominee, Kalshi, https://kalshi.com/markets/kxpresnomd/democratic-primary-winner/kxpresnomd-28.

[7] Kalshi became the first CFTC-regulated prediction markets platform in 2024 following a federal appeals court ruling that upheld Kalshi’s right to list contracts on political outcomes. KalshiEX LLC v. CFTC, 119 F.4th 58, 67 (D.C. Cir. 2024) (explaining that the CFTC had not provided sufficient evidence that such contracts would harm the public interest or the CFTC).

[8] Polymarket, for example, is “not effectively subject to CFTC rules,” as its largest exchange is domiciled offshore and blocks access to U.S.-based users. Karl E. Schneider & Rena S. Miller, Cong. Rsch. Serv., IF13187, Prediction Markets: Policy Issues for Congress (2026).

[9] Jonathan Kanter, Owen M. Kendler & Karina Lubell, Comments of the Antitrust Division of the United States Department of Justice Before S&P Global Inc. (Apr. 29, 2022), https://www.justice.gov/atr/page/file/1497956.

[10] Id.

[11] Owen M. Kendler & Karina Lubell, Comments of the Antitrust Division of the United States Department of Justice Before S&P Global Inc. (May 8, 2023), https://www.justice.gov/atr/media/1335201.

[12] See Statement of Interest of the United States at 7, Children’s Health Def. v. WP Co., No. 1:23-cv-2735 (D.D.C. July 11, 2025) (“[C]ontrolling precedent shows that the Sherman Act protects all forms of competition, including competition in information quality.”).

[13] Letter from Andrew N. Ferguson, Chairman, F.T.C, to Timothy Cook, Apple, Inc., Chief Executive Officer, Re: Potential FTC Act Violations Related to Suppressing or Promoting Featured News Articles for Political Reasons (Feb. 11, 2026), https://www.ftc.gov/system/files/ftc_gov/pdf/apple-news-warning-letter.pdf.

[14] Id.

[15] See, e.g., Herbert Hovenkamp, Antitrust and Platform Monopoly, 130 Yale L.J. 1952, 2014–2015 (2021); Lina Khan, Amazon’s Antitrust Paradox, 126 Yale L.J. 710, 731–732 (2017); H. Subcomm. on Antitrust, Comm. & Admin. Law, 117th Cong., Investigation of Competition in Digital Markets, at 28–33 (Comm. Print 2020).

[16] See, e.g., Fed. Trade Comm’n v. Qualcomm Inc., 969 F.3d 974, 994 (9th Cir. 2020) (reversing the lower court’s imposition of a duty to deal because it “ignores the Supreme Court’s … warning in Trinko that the Aspen Skiing exception should be applied only in rare circumstances”).