Prediction markets make a bold claim: that they are engines of truth discovery. Proponents argue that by aggregating collective judgment, prediction markets generate forecasts that are far more valuable than traditional gambling or sports betting. That claim has merit. But a prediction market’s output is not merely a well of raw, collective intelligence.
Identifying Integrity in Prediction Markets
Instead, the output is information filtered through architecture that determines who participates, how contracts are drafted, what evidence counts, how disputes are resolved, and which participants possess structural advantages. Kalshi and Polymarket operate under very different governing models. Kalshi relies on detailed contract rules and centralized resolution. Polymarket uses a decentralized, oracle-based resolution process.
Each model is prone to different integrity risks. Kalshi’s model can favor rigid contractual formalism that produces results at odds with common-sense interpretations of an event. Polymarket’s model allows greater interpretive discretion, with ultimate outcomes potentially influenced by economically interested token holders. That concern isn’t just theoretical. A recent analysis cited by The Economist found that the five largest UMA wallets controlled about 40 percent of votes, while another analysis identified Polymarket traders who also participated in UMA votes resolving markets in which they held positions.
Scrutiny of the representation gap has intensified: 67% of profits on Polymarket reportedly flowed to just 0.1% of accounts. That doesn’t guarantee that anyone is guilty of manipulation or fraud; concentrated profits are common in many markets. But how broadly distributed is the collective intelligence these markets claim to harness? How different is what’s true from what’s featured on the operator’s website and social media? The gap between representation and reality is being seized on by those who want to regulate strictly or ban prediction market products.
Bad actors are only part of the problem. Market rules can also create gaps between what a contract appears to ask and how it is ultimately decided. The integrity and legitimacy of prediction markets are now on trial.
A 60 Minutes episode highlighted scrutiny of long-shot military contracts by the Anti-Corruption Data Collective, which reported that contracts with implied odds below 35 percent paid off about 52 percent of the time. By comparison, long-shot sports bets pay only about 7 percent of the time. That suggests persistent informational advantages, even if it doesn’t establish illegality. When long shots win about half the time, information asymmetries are conspicuous, raising the question of whether ordinary consumers are trading against participants with superior information.
When Metrics Manipulate & Distort
In a robust market, attempts at manipulation should be self-correcting. A trader who pushes a price away from reality creates an opportunity for better-informed traders to push back and profit. The historical record largely supports this argument. But that defense assumes that only the price is being manipulated, rather than the event itself, or the evidence used to settle a contract.
The Spotify incident illustrates the problem. Kalshi relied on Spotify’s public chart to settle its contract, but Spotify later determined that more than 500,000 of the streams elevating the chart were artificial. It’s clear that once a public ranking determines who wins a wager, participants acquire an interest in influencing it.
A similar vulnerability arises when contracts are settled from specific news reports. The 60 Minutes broadcast described how Polymarket bettors pressured Times of Israel reporter Emanuel Fabian to change his account of an Iranian missile strike. These messages escalated into death threats against Fabian and his family. When a news report settles a wager, the report itself can be the object of pressure that should discomfort a free society.
This raises two basic design questions: how independent is the settlement evidence from the people betting on it? And who gets to decide what counts as the event having happened? Settlement is not merely an administrative task, but a governance function. Ambiguous contracts can force an operator or oracle to choose which facts, definitions and sources control. By contrast, overly detailed rules and rigid rule interpretations can conflict with common-sense understandings of whether an event occurred. In either case, sophisticated traders may be predicting how the rules will be applied as much as whether the underlying event will occur.
Military events raise special concerns. Knowledge of a military operation is often distributed among planners, analysts, commanders, contractors, and others. A sudden, confident position can itself become information, potentially signaling that an operation is imminent, negotiations are advancing, or escalation is likely. Conversely, a participant may place trades intended to create a false impression of informed conviction.
The consequences are more profound than unfair advantage: they could expose highly sensitive information or give credibility to a signal that is manufactured by an adversary. A market-integrity issue now becomes a national-security concern. Providers therefore must consider whether the contracts they offer create incentives to influence events, pressure sources, expose private information, or manufacture misleading signals. There is also a more basic boundary question: whether some contracts involving military action, death or other consequential events create incentives or risks that make them inappropriate markets in the first place.
Minding the Gap
Operators of prediction markets must realize they are playing a different game than other providers of digital products and services. Outsiders routinely scrutinize trading records, blockchain data, win rates, account patterns, and other anomalies. That visibility changes the risk calculus.
Trust gaps will be priced through new restrictions, higher compliance costs, lost partners and customers, insurance and litigation risk, and ultimately, lower enterprise value. When outsiders measure persistent gaps, they shape the public narrative, regulatory priorities, commercial relationships, and company value. When anomalies and gaps are defined and publicized by outsiders, they determine the timing, framing, and language of the debate, forcing operators to defend themselves on someone else’s terms.
AI is intensifying these clashing dynamics of innovation and fraud by enabling fabricated identities, coordinated activity, and automated trading while improving surveillance and anomaly detection. This spotlight extends beyond individual users and transactions to illuminate underlying market patterns and trust signals.
There is an understandable industry objection to deep self-examination at a time of rapid market growth; if we inventory our weaknesses or air our soiled laundry, it will be used against us. That’s a real risk. But self-scrutiny doesn’t mean public disclosure. Operators can investigate, correct, and monitor problems without inviting greater scrutiny or providing a roadmap for future abuse. Declining this effort is the riskier bet.
The practical choice is not between complete disclosure and an ostrich-in-the-sand approach. It’s between mapping the gaps on your own terms or allowing others to define the debate and set the agenda.
A Roadmap to Trust in Prediction Markets
Operators must identify and measure their own gaps first. This is not merely a compliance, risk management or public relations exercise. Mapping a market’s promises with its operation and results is the first step towards rebuilding trust, sustainable long-term growth and freedom to innovate. Start with these questions:
- Can a participant materially influence the outcome they’re betting on?
- Does anyone hold privileged operational knowledge of the event? Would the platform know if they traded on it?
- How much of the market’s activity is synthetic, and does that activity corrupt the price signal itself rather than merely skim profit from it?
- What behavior does the product’s design reward, independent of what its marketing claims?
- Why should participants trust this market and its signals?
Many gaps can be narrowed through better detection, monitoring, product design, and safeguards. The goal is not to eliminate or disclose every vulnerability, but to address what can be addressed and make deliberate decisions about what to communicate with whom.
Participants assume that markets are monitored, misconduct is investigated, and meaningful sanctions follow serious abuses. Regularly testing and evaluating fairness, integrity, and trust should be part of an operator’s “practical license” to operate.
Operators that identify and narrow trust gaps before others do may enjoy a lasting advantage as these markets mature. How operators respond to controversies and regulatory pressures will help determine whether prediction markets earn and deserve the public’s trust and become sustainable institutions embedded within society’s information and decision-making.
