Prediction markets were designed to aggregate dispersed information into continuously updating prices, and by that measure, they have succeeded. What began as an academic experiment at the University of Iowa in 1988 has grown into a financial ecosystem worth billions, with over $3.7 billion wagered on the 2024 U.S. presidential election alone and real-time odds quoted alongside traditional polling by the world's leading financial press. But the same features that make these markets valuable, namely anonymity, offshore reach, and the absence of position limits, also make them uniquely susceptible to exploitation by those with privileged access to non-public information. This article examines how prediction markets evolved, how they compare to regulated derivatives, what the trading data surrounding recent geopolitical events reveals, and why the regulatory framework has yet to catch up
Introduction
On the morning of 28 February 2026, at 06:15 UTC, the United States launched Operation Epic Fury against Iran. The financial markets reacted immediately, as expected. What was less expected, and far harder to explain, is that Polymarket, the world's largest prediction market, had already moved hours earlier. Between 05:40 and 06:12 UTC, a small cluster of wallets initiated concentrated buy-side positions in the "U.S. strikes Iran" contract. By the time the first public reports emerged from Iranian state media at 06:27 UTC, those positions had already appreciated dramatically. The market did not react to the news – it had anticipated it.
This episode encapsulates the central tension that defines prediction markets today. At their most basic level, these platforms offer a structurally elegant mechanism: participants buy binary contracts that pay $1 if a specified event occurs and $0 otherwise. The transaction price can be read directly as a market-implied probability, updated in real time as new information flows in. In that sense, prediction markets perform a familiar and genuinely useful financial function: they aggregate dispersed information and convert it into continuously updating forecasts that, in liquid contracts, have been shown to outperform conventional polling and expert models alike.
But the same feature that makes them valuable, i.e., the financial incentive to incorporate information quickly, also creates a structural vulnerability. Once these contracts become liquid and reference elections, military actions, or policy decisions, they begin to resemble event-linked derivatives more than recreational wagers. And in a market where participants can trade anonymously, with no identity requirements, no position limits, and no regulatory surveillance, the line between superior forecasting and the monetization of material non-public information (“MNPI”) is impossible to draw from the outside.
This article traces that tension from origin to consequence. It examines how prediction markets evolved from a niche academic experiment into a multi-billion-dollar financial ecosystem, how they compare structurally to regulated derivatives markets, what the trading data surrounding major political and geopolitical events actually shows, and why the current regulatory framework has failed to keep pace. The conclusion is not that prediction markets should be abolished; their informational value is real. It is that they have grown too large, too liquid, and too connected to political and geopolitical decision-making to continue operating without a coherent cross-border oversight framework: one that matches the global footprint of these markets rather than stopping at the jurisdictional boundaries that sophisticated actors have proven adept at exploiting.
,
From Niche to Mainstream: The Rise of Prediction Markets
The first real-money prediction market of the modern era was not built by a financial institution. In 1988, economists at the University of Iowa launched the Iowa Electronic Markets (IEM), a platform designed explicitly to forecast U.S. presidential elections. The academic motivation was straightforward: elections produce binary, verifiable outcomes on a fixed date, making them structurally ideal objects for a contract market. Participants with private information, e.g., about local sentiment, campaign dynamics, or voter behaviour, would be induced by financial incentives to incorporate that information into prices, yielding a probability estimate superior to any individual forecast. This logic was empirically validated by Berg, Nelson & Rietz (2008), whose study of 964 contemporaneous polls across five presidential elections found that IEM forecasts outperformed the polls in 74% of cases, establishing prediction markets as a credible forecasting tool rather than a speculative novelty. Research by Wolfers and Zitzewitz (2006), published through the Federal Reserve Bank of San Francisco, further reinforced this claim by showing that similar mechanisms could be used to track market expectations for economic outcomes including CPI and FOMC decisions.
The IEM's growth was nonetheless deliberately circumscribed. The platform operated under CFTC no-action letters issued in 1992 and 1993, which exempted it from commodity exchange regulation on the condition that participation was capped at 2,000 traders and individual stakes limited to $500. This arrangement preserved the academic character of the market while preventing it from becoming the basis for commercial operation. The result was a structurally thin market that demonstrated the forecasting value of the mechanism but could not attract the liquidity necessary to scale. The broader regulatory logic compounded the problem: the Commodity Exchange Act did not clearly classify event contracts, leaving any for-profit platform in a grey zone between gambling regulation, securities law, and commodity oversight. The Irish-based platform Intrade illustrated the ceiling this created. Despite building genuine liquidity and a sophisticated user base through the mid-2000s, it was forced to block U.S. customers following a CFTC enforcement action in 2012 and shut down entirely in 2013.
The inflection point came from a deliberate technological circumvention. Polymarket, launched in 2020, built its platform on cryptocurrency infrastructure: participants deposit USDC stablecoins, trade on the Polygon blockchain, and access markets through a digital wallet rather than a conventional brokerage account. This architecture addressed three constraints that had kept prediction markets niche. The liquidity problem was resolved by enabling frictionless global participation with no position limits. The access problem – previously addressed by onerous Know Your Customer (“KYC”) and account registration requirements on compliant platforms – was resolved by displacing traditional, centralized account registration in favour of non-custodial digital wallets. Unlike a conventional account, where the platform “owns” the user’s data and funds in a private database, a non-custodial wallet allows the user to remain the sole gatekeeper of their assets and identity. In practice, the platform does not “create” an account for the user but rather simply interacts with a pre-existing digital signature. This structure therefore enables immediate market access without the gatekeeping of a traditional financial intermediary.
The regulatory constraint, on the other hand, was addressed by incorporating offshore, outside the direct reach of U.S. financial regulators. Volume followed rapidly. By October 2024, the platform was recording approximately $2.3 billion in monthly trading activity, with the "Presidential Election Winner 2024" contract alone generating $3.7 billion in total volume over its lifetime – a scale that placed Polymarket alongside, rather than apart from, conventional financial markets in terms of informational significance. CNN and the Financial Times began quoting its odds; its real-time probability estimates were treated as informational signals alongside traditional polling.
From Niche to Mainstream: The Rise of Prediction Markets
The first real-money prediction market of the modern era was not built by a financial institution. In 1988, economists at the University of Iowa launched the Iowa Electronic Markets (IEM), a platform designed explicitly to forecast U.S. presidential elections. The academic motivation was straightforward: elections produce binary, verifiable outcomes on a fixed date, making them structurally ideal objects for a contract market. Participants with private information, e.g., about local sentiment, campaign dynamics, or voter behaviour, would be induced by financial incentives to incorporate that information into prices, yielding a probability estimate superior to any individual forecast. This logic was empirically validated by Berg, Nelson & Rietz (2008), whose study of 964 contemporaneous polls across five presidential elections found that IEM forecasts outperformed the polls in 74% of cases, establishing prediction markets as a credible forecasting tool rather than a speculative novelty. Research by Wolfers and Zitzewitz (2006), published through the Federal Reserve Bank of San Francisco, further reinforced this claim by showing that similar mechanisms could be used to track market expectations for economic outcomes including CPI and FOMC decisions.
The IEM's growth was nonetheless deliberately circumscribed. The platform operated under CFTC no-action letters issued in 1992 and 1993, which exempted it from commodity exchange regulation on the condition that participation was capped at 2,000 traders and individual stakes limited to $500. This arrangement preserved the academic character of the market while preventing it from becoming the basis for commercial operation. The result was a structurally thin market that demonstrated the forecasting value of the mechanism but could not attract the liquidity necessary to scale. The broader regulatory logic compounded the problem: the Commodity Exchange Act did not clearly classify event contracts, leaving any for-profit platform in a grey zone between gambling regulation, securities law, and commodity oversight. The Irish-based platform Intrade illustrated the ceiling this created. Despite building genuine liquidity and a sophisticated user base through the mid-2000s, it was forced to block U.S. customers following a CFTC enforcement action in 2012 and shut down entirely in 2013.
The inflection point came from a deliberate technological circumvention. Polymarket, launched in 2020, built its platform on cryptocurrency infrastructure: participants deposit USDC stablecoins, trade on the Polygon blockchain, and access markets through a digital wallet rather than a conventional brokerage account. This architecture addressed three constraints that had kept prediction markets niche. The liquidity problem was resolved by enabling frictionless global participation with no position limits. The access problem – previously addressed by onerous Know Your Customer (“KYC”) and account registration requirements on compliant platforms – was resolved by displacing traditional, centralized account registration in favour of non-custodial digital wallets. Unlike a conventional account, where the platform “owns” the user’s data and funds in a private database, a non-custodial wallet allows the user to remain the sole gatekeeper of their assets and identity. In practice, the platform does not “create” an account for the user but rather simply interacts with a pre-existing digital signature. This structure therefore enables immediate market access without the gatekeeping of a traditional financial intermediary.
The regulatory constraint, on the other hand, was addressed by incorporating offshore, outside the direct reach of U.S. financial regulators. Volume followed rapidly. By October 2024, the platform was recording approximately $2.3 billion in monthly trading activity, with the "Presidential Election Winner 2024" contract alone generating $3.7 billion in total volume over its lifetime – a scale that placed Polymarket alongside, rather than apart from, conventional financial markets in terms of informational significance. CNN and the Financial Times began quoting its odds; its real-time probability estimates were treated as informational signals alongside traditional polling.
Figure 1 – Polymarket monthly trading volume growth, 2021–2026
Alongside the rapid expansion of offshore platforms, a structurally distinct model emerged within the U.S. regulatory perimeter: Kalshi. Launched in 2021, Kalshi is the largest CFTC-approved prediction market in the United States, classified as a “Designated Contract Market” under the same regulatory framework as the Chicago Mercantile Exchange. It is supported by institutional market makers including Susquehanna and is accessible to retail investors through platforms such as Robinhood and Webull. The economic function is identical to Polymarket's: both offer binary event contracts paying one dollar on resolution. The regulatory treatment, however, is different. Kalshi operates with mandatory KYC requirements, position reporting obligations, and active regulatory supervision.
While Polymarket received a CFTC Amended Order of Designation in November 2025, permitting U.S. operations through registered intermediaries, the platform's offshore version continues to operate outside this scope. It is precisely on this platform however that most of the significant suspected insider trades happened. For instance, the trades preceding Nicolas Maduro’s arrest in January 2026 and the Iran war bets of February 2026 were placed on the international exchange, not subject to any regulatory supervision, despite such events concerned actions initiated by the U.S. government.
American users retain access to the offshore platform via VPN, though Polymarket has progressively tightened access restrictions in recent months, making this route increasingly difficult for ordinary retail participants. The practical consequence is that those who continue to access the offshore platform are likely to be sophisticated professionals or politically connected actors, whose technical capabilities and material incentives are precisely consistent with the trading patterns documented. As an example, the single trader who placed bets hours before unannounced U.S. and Israeli strikes recorded a 93%-win rate across all wagers above $10,000.
Prediction Markets vs. Futures: A Structural Comparison
Following the CFTC’s 2025 designations, the regulatory landscape for prediction markets has transitioned from a period of ambiguity into a bifurcated, supervised regime. To understand this new environment, it is helpful to compare these instruments with traditional futures markets, instruments that have also long served as forward-looking contracts encoding collective beliefs about future states of the world.
Traditional futures trade on registered exchanges, use central counterparties for settlement, and reference continuous variables such as commodity prices, interest rates, or equity indices. Prediction markets, by contrast, reference discrete binary event outcomes and often operate via platform-managed or decentralized systems without a central clearinghouse. The risk architecture also differs in an important way: futures contracts are marked to market daily, with margin calls imposing continuous discipline on losing positions. Leverage is embedded by design, i.e., a trader holding a standard S&P futures contract controls a notional exposure that is a multiple of the margin deposited (typically 5-10% of notional), meaning that price movements amplify gains and losses relative to the capital committed. On a prediction market, by contrast, full collateralization is required on every position: a buyer of a "Yes" contract at $0.30 deposits exactly $0.30 per share and cannot lose more than that amount. This eliminates systemic counterparty exposure but also caps the return profile.
Beneath these institutional differences, however, both instruments generate continuously updated signals that function as probabilistic forecasts, and both move on the same underlying driver: information. A compelling illustration of this functional equivalence came on the night of the 2004 U.S. presidential election. Snowberg, Wolfers & Zitzewitz (2007), in a study published in the Quarterly Journal of Economics, showed that the Intrade contract pricing George W. Bush's re-election and the S&P 500 December futures contract tracked each other in near-perfect lockstep across twelve hours of vote counting. As flawed exit polls briefly showed John Kerry leading in key swing states, both the prediction market and the S&P future fell in tandem; as actual vote totals reversed that trend, both recovered together. The authors estimated that a Bush victory over Kerry was associated with an S&P 500 level roughly 1.6-2% higher, a degree of synchronicity that leaves little doubt about the informational overlap between the two market types.
Figure 2 – Intrade Bush re-election contract vs. S&P 500 futures, election night 2004 | Source: Snowberg, Wolfers & Zitzewitz (2007), QJE
This functional synchronicity makes the divergence in regulatory treatment all the more striking. Futures markets operate under anti-manipulation and insider trading doctrine developed continuously since the Commodity Exchange Act of 1936 and significantly strengthened by the Dodd-Frank Act of 2010. Position reporting requirements, large-trader surveillance, real-time exchange monitoring, and the formal obligations of designated market makers together create an environment in which anomalous pre-announcement trading is at minimum detectable and, as CFTC v. EOX Holdings (2018) demonstrated, prosecutable: in that case, a broker was charged with misusing MNPI in energy block trades precisely because the audit trail made the conduct visible.
While domestic platforms like Kalshi now operate within a defined regulatory perimeter, the broader landscape remains only partially addressed. For Kalshi, the CFTC's anti-fraud authority under the Commodity Exchange Act extends in principle: the online platform is subject to mandatory KYC, position reporting, anti-manipulation rules, and individual exposure capped at $7 million. But for offshore platforms like Polymarket, which throughout 2024 processed the overwhelming majority of global prediction market volume, none of these tools applied. The absence of mandatory KYC protocols meant that while every transaction was visible on the blockchain, the actor behind it remained anonymous. Without the ability to map public keys to real-world identities, the standard apparatus of financial surveillance was rendered functionally inoperable. If futures markets represented the standard against which event-driven price discovery should be regulated, prediction markets, as they operated through 2024, represented its deliberate inverse.
Even following the CFTC’s designation of Polymarket’s U.S. platform in November 2025, the enforcement gap identified here remains materially unclosed. This persistence reflects the continued operation of anonymous offshore liquidity pools on Polymarket’s international platform, which lies beyond the effective reach of mandatory KYC requirements and U.S. federal oversight.
The Suspicious Trades: What the Volume Data Shows
The regulatory gap described above would be a theoretical concern if prediction markets attracted only dispersed retail participants trading on publicly available information. However, the trading data from several high-profile events suggests the reality is different. The following two cases – selected because they involve documented timing anomalies, concentrated wallet activity, and subsequent official or journalistic scrutiny – illustrate a pattern that, in any regulated derivatives market, would trigger immediate surveillance.
Figure 3 – Signed order flow and VWAP, "U.S. strikes Iran" contract, 04:00–07:00 UTC on 28 February 2026
Operation Epic Fury, February 2026: The following analysis examines the Polymarket contract family on whether the United States would strike Iran. Polymarket listed a separate binary market for each possible strike date, so that each deadline corresponded to a distinct contract in which a "Yes" share paid $1 if a qualifying U.S. strike occurred by that date. In the absence of new information, the price of such a fixed-deadline contract would normally decline as expiry approached, since less time remained for the event to occur. The first pronounced increase in prices occurred on 27th February from 10:00 UTC, when contracts across all deadlines moved upward simultaneously. The contract expiring on 28th February briefly reached almost 30%, despite having less than 40 hours remaining until expiry – a move inconsistent with any plausible public information available at that time.
Figure 4 – Polymarket "U.S. strikes Iran" contract: implied probability across deadlines, February 2026
Operation Epic Fury formally began at 06:15 UTC on 28th February 2026. Examining trading activity in the preceding two hours reveals a striking concentration: only a small number of wallets were active before 06:12 UTC, and roughly 90% of the volume in this window came from four distinct wallets. The largest individual buy-side position was initiated at 05:40 UTC – 35 minutes before the operation commenced and 47 minutes before the first public report from Iranian state media. To estimate the economic significance of these trades, we apply a notional measure: multiplying the signed position size (the number of "Yes" contracts bought net of contracts sold) by the value-weighted average price (VWAP) at the time of execution. This yields the effective capital deployed in the direction of the outcome. On this basis, the largest single wallet invested approximately $40,000 in this window - implying a gross payoff of roughly $130,000 if the position was held to resolution, a return of approximately 225%. After 06:12 UTC, trading activity broadened sharply: the number of active wallets increased substantially, and the VWAP rose from 0.255 to 0.87 by 06:16 UTC, driven by purchases from more than 50 distinct wallets reacting to the same publicly observable signal that retail participants always had to wait for. The pre-06:12 cluster did not.
Figure 5 – Polymarket "Nobel Peace Prize: María Corina Machado" contract, VWAP and signed order flow, 9–10 October 2025
The 2025 Nobel Peace Prize: The second case concerns trading activity in the Polymarket market on whether María Corina Machado would win the Nobel Peace Prize. The prize was announced on 10th October 2025 at 11:00 CEST in Oslo. This setting is particularly relevant because the Nobel Peace Prize process is exceptionally secretive: the Committee does not disclose nominees or deliberations, and the underlying materials remain confidential for 50 years. Public forecasts did not place Machado among the leading favourites. PRIO's January 2025 shortlist ranked Sudan's Emergency Response Rooms first, and bookmaker odds on the morning of 9th October still had Donald Trump and Sudan's Emergency Response Rooms as joint favourites, with Machado absent from the principal contenders. Yet shortly after midnight Norwegian time on 9th October, Machado's Polymarket contract rose abruptly from 3.75% to 72.8%.
The price movement coincided with a sequence of large positive signed-volume spikes, indicating aggressive buy-side trading concentrated in the early hours of the morning – Plehours before any plausible public signal could have shifted probabilities so dramatically. Once repriced, the contract remained stable at roughly 70-80% throughout the day and overnight, moving to certainty only when the official announcement was made. This pattern is difficult to reconcile with ordinary public speculation: the prior probability assigned by expert forecasters was negligible, the timing of the initial move preceded any observable news, and the stability of the repriced contract over 30 hours suggests the buyer knew, rather than guessed. Press reports subsequently indicated that Norwegian Nobel officials had launched an investigation into whether confidential information had leaked into betting markets.
These two cases are not cherry-picked anomalies. A March 2026 study published on the Harvard Law School Forum on Corporate Governance authored by Joshua Mitts (Columbia Law School) and Guy Ofir (University of Haifa), analyzed over 93,000 distinct Polymarket markets and nearly 50,000 unique wallet addresses covering the period from February 2024 to February 2026. Across 210,718 suspicious wallet-market pairs, flagged traders achieved a 69.9%-win rate, a result exceeding the null distribution of random chance by more than 60 standard deviations under a permutation test. The authors estimate approximately $143 million in aggregate anomalous profit across the study period, a figure they describe as a conservative lower bound. The pattern is consistent: concentrated pre-event buy-side activity from a small number of wallets, a sharp price move in the hours before an announcement, followed by a broadening of activity once the news becomes public. In a regulated futures market, this sequence has historically triggered surveillance and enforcement action. In prediction markets, until recently, it could not, since there is no regulator within its jurisdiction, and no audit trail linking wallets to identities.
The U.S. Regulatory Gap: Jurisdiction Without Enforcement
The trading patterns documented above once stemmed primarily from a classification failure, a legal grey area; today, however, the problem has shifted: it is no longer one of definition, but of enforcement. While the CFTC's designation of domestic platforms under the Commodity Exchange Act has brought prediction markets meaningfully within the regulatory taxonomy, the more persistent problem is now one of reach: U.S. traders continue to access offshore platforms via VPNs, and the regulatory apparatus, even where it technically applies, lacks the identity infrastructure to act on what the blockchain makes visible.
Historically, securities law failed to apply because event contracts lack an issuer and do not represent claims on a business enterprise. Their value is not derived from managerial effort or corporate cash flows, but from the occurrence of an external event, placing them outside the core logic of the Securities Exchange Act. Gambling law, governed at the state level, remained a poor fit because there is no bookmaker setting odds or taking the other side of trades: prices emerge from continuous interaction between market participants, making these platforms structurally closer to exchanges than to casinos. Commodity law, governed by the Commodity Exchange Act, came closest – the CEA explicitly contemplated derivatives based on "events," and the CFTC previously took action against Polymarket, imposing a $1.4 million civil monetary penalty in January 2022 for offering off-exchange, event-based binary option contracts without proper registration. Nonetheless, from the early 2000s, when prediction markets first emerged, through to the early 2020s, event contracts were not consistently brought under a unified derivatives framework.
The Kalshi litigation of 2023-2024 represents the clearest attempt to resolve this ambiguity. In September 2023, the CFTC blocked Kalshi from listing certain U.S. election contracts, arguing they constituted "gaming" and were contrary to the public interest. Kalshi challenged that determination, and in September 2024 the U.S. District Court for the District of Columbia vacated the agency’s order. Although both the District Court and the D.C. Circuit issued temporary administrative stays during the ensuing appeal, the D.C. Circuit ultimately denied the CFTC’s request for a stay pending appeal in October 2024, allowing contracts to proceed while litigation continued.
The jurisdictional ambiguity was further complicated by the SEC-CFTC divide. If prediction market positions were classified as securities, the SEC would have authority; if they were derivatives written on non-security underlyings, the CFTC would. Neither agency formally claimed jurisdiction over the space as a whole, and the absence of a clear ownership – less a turf war than a shared blind spot – was itself a regulatory failure. The international comparison offers some perspective. The United Kingdom and much of the European Union classify prediction markets as gambling, which imposes tighter controls on retail access but largely ignores market integrity concerns such as insider trading and information asymmetry. The EU's MiCA framework introduced a unified structure for crypto-asset markets but does not address event contracts directly.
The November 2025 CFTC designation of Polymarket's U.S. platform marked a formal step toward regulatory normalisation, but it did not close the enforcement gap. Sophisticated actors continue to route around identity requirements via VPNs, accessing offshore liquidity pools with the same anonymity that characterised the pre-designation era. Without KYC obligations that extend meaningfully to international platforms, linking suspicious wallets to real-world identities remains structurally impossible, and the prohibition on insider trading, however clearly written, remains difficult to enforce.
The Road Ahead: A Regulatory Reckoning
Prediction markets have clearly demonstrated their value. During the 2024 U.S. election cycle, they aggregated vast amounts of dispersed information into continuously updating price signals, reacting faster than traditional polling and proving, in several instances, more accurate in capturing shifts in expectations. What began as a small-scale academic initiative at the University of Iowa has evolved into a multi-billion-dollar financial ecosystem attracting retail traders, institutional capital, and, as the evidence increasingly suggests, participants with proximity to political and geopolitical decision-making. That informational efficiency is real, and any regulatory framework worth designing should preserve it.
But scale changes the stakes. The transition from niche to mainstream has not been accompanied by a corresponding evolution in oversight. The result is a market in which the same features that generate informational value, such as speed, anonymity, accessibility, and the absence of position limits, also generate ideal conditions for the exploitation of non-public information. The trading patterns surrounding Operation Epic Fury and the 2025 Nobel Peace Prize are not isolated curiosities. They are consistent with a broader pattern identified by Mitts & Ofir (2026) across more than $143 million in anomalous profits: concentrated pre-announcement activity, precise timing, and repeated abnormal returns from a small number of actors whose performance cannot be explained by superior public analysis alone.
In regulated derivatives markets, such behaviour would trigger surveillance, investigation, and potentially enforcement action. In prediction markets, it has largely remained unaddressed – not because the activity is different, but because the existing tools and jurisdiction granted to regulators have yet to be fully enforced against offshore bypasses. Whether this is characterised as a policy choice or an institutional failure, the effect is the same: a market that increasingly resembles a regulated financial instrument has been allowed to operate, in its offshore form, without the safeguards that such instruments require.
The Commodity Exchange Act already provides a ready-made structure. Ensuring the full effectiveness of that framework now requires a more uniform application of existing standards to bridge the “grey zone” still present in many markets. The longer that decision is deferred, the larger the markets will grow, the more sophisticated the actors will become, and the more the absence of oversight will begin to look less like a gap and more like a feature.
By: Moritz Luther, Giuseppe Marcarelli, Stefaniya Yakubovskaya
SOURCES
- Berg, J.E., Nelson, F.D. & Rietz, T.A. (2008). "Prediction Market Accuracy in the Long Run." International Journal of Forecasting, 24, 283–298.
- CNN
- CFTC
- CFTC v. EOX Holdings, L.L.C., Case No. 1:18-cv-08890 (S.D.N.Y. 2018).
- Dune Analytics
- Mitts, J. & Ofir, G. (2026). "From Iran to Taylor Swift: Informed Trading in Prediction Markets." Harvard Law School Forum on Corporate Governance, 25 March 2026.
- Polymarket
- Snowberg, E., Wolfers, J. & Zitzewitz, E. (2007). "Partisan Impacts on the Economy: Evidence from Prediction Markets and Close Elections." Quarterly Journal of Economics, 122(2), 807–829.
- Wikipedia
- Wolfers, J. & Zitzewitz, E. (2006). "Five Open Questions About Prediction Markets." Federal Reserve Bank of San Francisco Working Paper 2006-06.