Market manipulation in prediction markets is far more complex than it appears on the surface. This article highlights a core paradox: although historically successful manipulation of election markets has been rare, in an era where AI-generated public opinion and social media amplify signals, even brief price distortions can trigger a crisis of trust.

The key lies in liquidity. Low-liquidity markets are indeed more susceptible to manipulation—research by Rhode and Strumpf has confirmed that large unilateral trades in high-liquidity environments are quickly arbitraged back, but in thin markets, such distortions can persist. This means not all prediction markets are equal.

From a data perspective, the real threat is not that manipulation can change election outcomes, but that it can undermine the perception of the market itself. Once the public can no longer distinguish whether price fluctuations are driven by genuine information or simply capital games, the value of the market as an information aggregation tool collapses entirely. This is especially deadly now—traditional polls have failed amid AI noise, and we urgently need a mechanism to integrate dispersed expectations.

The solutions point in three directions: media should set liquidity thresholds and only report active markets; platforms need to establish manipulation detection systems and transparency disclosures; regulators should incorporate market manipulation into existing anti-manipulation laws. It may seem like a patchwork, but it is sufficiently practical at this stage. The true test will come after 2024.