There is a number that should focus minds in Washington right now, and it is not the one on any contract. Across Kalshi and Polymarket combined, more than 100 accounts have been flagged to authorities this year on suspicion of insider trading — Kalshi reporting over 50 referrals, Polymarket referring more than 90 accounts to regulators in the United States and abroad.
The overlap between those two figures is not public. Some of those 90 Polymarket referrals may be the same individuals Kalshi identified; some may not. What the numbers establish, taken together, is that the platforms are no longer treating surveillance as a reputational gesture. They are filing cases.
This matters structurally in a way that the headline count alone does not capture. Prediction markets have spent three years arguing, in courts and in Congressional testimony and in public filings, that they function like financial markets and should be regulated as such. The CFTC review of event contracts now reportedly in motion is the direct consequence of that argument succeeding. But financial markets come with a cost: the scrutiny that applies to the New York Stock Exchange now applies here too. You cannot claim the legitimacy of a regulated market and then treat your order book as ungoverned territory. The referral numbers are the platforms demonstrating they understand this.
I have seen this transition before, in earlier instrument classes that achieved regulatory recognition and then discovered recognition brought obligations their compliance functions were not sized for. The question is not whether prediction markets can flag suspicious activity — they clearly can. It is whether the downstream infrastructure exists to act on what they flag.
That infrastructure is the part I am not convinced is there yet. A referral to the CFTC or a foreign authority is not a prosecution. The CFTC's enforcement division is not large, and event contracts are still novel enough that the legal theories around insider trading in thin prediction markets have not been tested in court in any systematic way. What counts as material non-public information when the underlying event is a political decision, a regulatory ruling, or a military action? These are genuinely unsettled questions, and the answer matters enormously for how seriously a would-be manipulator needs to take a referral.
JPMorgan's decision to exit Polymarket's banking relationship while reportedly remaining open to an IPO role tells you something about how sophisticated institutions are reading this moment. The compliance risk of holding the account is real and immediate. The upside from an eventual public offering is speculative and distant. That is a rational calculation. It is also a signal that the regulatory normalisation of prediction markets is happening unevenly — fast enough to create liability, not yet fast enough to create clarity.
