Many claim that prediction markets demonstrate the wisdom of crowds, and accurately tell us the future. This assumption has led them to become widespread in governments, newsrooms, and AI systems, treated as a valuable source of knowledge. But in new research, Yasaman Rohanifar, Syed Ishtiaque Ahmed, and Sharifa Sultana argue this is a catastrophic misreading. Behind prediction markets' democratic facade lies a process of prediction laundering—one in which anonymous whales, political insiders, and coordinated capital manufacture consensus while the platforms ensure the public never sees the machinery behind it.

The rules of public knowledge are changing. For decades, when citizens, policymakers, and financial institutions needed to gauge the likelihood of a major world event, they turned to traditional pillars of authority such as intelligence briefings, journalist reports, statistical pollsters, and panels of academic experts. Today, that legacy informational infrastructure is being bypassed by raw financial mechanisms instead. The trading prices of decentralized prediction platforms like Polymarket have evolved from niche internet curiosities into the new global norm for establishing political and social facts.

The scale of this trend is massive, and it is accelerating. Leading global financial networks have officially integrated live prediction market feeds into their flagship terminal frameworks, treating speculative contract fluctuations with the same institutional seriousness as sovereign bond yields or corporate equity streams. At the same time, major media conglomerates are restructuring their reporting around these metrics, routinely citing market odds as definitive proof of public sentiment rather than relying on traditional polling data. Perhaps most significantly, prediction market data has migrated into the automated computational systems that govern daily digital life. Conversational artificial intelligence tools and large language model search engines now scrape these market percentages in real time, serving them directly to millions of everyday users as the objective truth regarding complex geopolitical conflicts and domestic policy decisions. 

This institutional adoption is driven by the premise that financial markets possess a unique capacity to aggregate dispersed human knowledge efficiently. Proponents argue that because traders must put their money where their mouth is, backing their assertions with capital, the resulting price signal acts as a more honest distillation of global opinion. This view is not unfounded—the Iowa Electronic Markets consistently outperformed major polling aggregators in forecasting US presidential elections across multiple cycles. Controlled experiments have demonstrated that prediction markets can outperform both individual experts and conventional survey instruments when the conditions are right.

However, treating the speculative percentages produced by platforms like Polymarket as neutral information carries a risk. As public debate intensifies over the expansion of commercial predictive speculation into new international jurisdictions, critics have increasingly challenged their perceived neutrality. Prediction market platforms function less like democratic public squares and more like predatory, unregulated financial spaces that can actively incentivize the manipulation of public opinion and insider trading.

Behind the authoritative digital facade lies a complex social and financial machine. The apparently clean, unbiased numbers consumed by media networks and AI models are the output of a multi-stage pipeline we might think of as “prediction laundering”: a structural process by which highly strategic financial bets are systematically scrubbed of their context to present a mirage of objective truth to the external world.

So how does the laundry cycle work? There are four distinct stages of how a prediction is manufactured.