The Cop Left the Building: CFTC Shrinks 24% While Prediction Markets Hit $60B

#predictionmarkets #fintech #regulation #datascience

Issue #7 · April 27, 2026 · By John Leslie

The agency responsible for policing prediction markets has lost a quarter of its workforce. Meanwhile, the industry just crossed $60 billion in trades this year. Brazil banned every prediction market platform overnight. And the economist who invented the theoretical framework behind prediction markets argues that the insider trading everyone is panicking about is actually the entire point. Welcome to the most confusing week in the industry's short history.

The Cop Left the Building

The Commodity Futures Trading Commission -- the federal agency that regulates prediction markets -- has shrunk to its smallest size in 15 years. Workforce is down 24% since January 2025. That's not attrition. That's a policy choice.

This is happening precisely as the agency's jurisdiction has become the most contested terrain in financial regulation. More than two dozen state-level lawsuits allege that Polymarket and Kalshi are unlicensed gambling operations. Three states tried to ban the platforms outright last week; the DOJ sued all three, arguing federal preemption. The CFTC is supposed to referee this fight. It now has fewer referees than at any point since 2011.

The volume numbers make the staffing collapse almost comical. Kalshi and Polymarket have processed $60 billion in trades so far in 2026. In March alone, Kalshi handled $13 billion and Polymarket roughly $10 billion. For context, the entire CFTC-regulated prediction market had less than $1 billion in annual volume as recently as 2023. We're talking about a 60x increase in three years with a 24% decrease in oversight capacity.

Our take: Underfunding the regulator is short-term bullish and long-term dangerous. Right now, it means platforms operate with minimal oversight -- good for innovation, bad for legitimacy. The risk is that a major scandal forces Congress to overcorrect.

Brazil Bans Everything

Brazil became the first major economy to ban prediction market platforms entirely. As of today, Polymarket, Kalshi, and 27 other platforms are blocked in the country. The Brazilian government classified all of them as illegal betting operations.

This isn't a small market. Brazil has the fifth-largest internet population globally and a massive sports betting culture. Polymarket had meaningful traffic from Brazilian users. That's gone overnight.

Our take: Brazil's ban is a setback for global market depth but reinforces the US as the center of gravity. If you're building in this space, your regulatory strategy is now a US-first strategy by default.

Robin Hanson: Insider Trading Is the Point

Fortune published a remarkable interview this week with Robin Hanson, the George Mason economist whose 1990s work essentially invented the theoretical framework for prediction markets. His argument: the insider trading that everyone is panicking about is exactly what makes prediction markets valuable.

The logic: prediction markets aggregate dispersed information into accurate prices. When a soldier bets on a military operation he has classified knowledge about, that information moves the price closer to truth. The market becomes more accurate, not less.

Our take: Hanson is directionally right and practically wrong. In a frictionless theory, insider information makes markets more efficient. In the real world, it creates perverse incentives. The right answer is the one Kalshi is already implementing: ban the trades that create bad incentives (politicians on their own races), allow the rest.

Hormuz: "Reopening Is Impossible"

Iran's parliament speaker said this week that "reopening the Strait of Hormuz is impossible as long as the US blockade is in place." Ship traffic remains at roughly 5 vessels per day versus 140 pre-crisis. Bloomberg described it as "maritime trench warfare."

Current prices:

April normalization: 0.4% YES -- locked in
May normalization: 36.5% YES -- well below our entry at 69.5%
June normalization: 55.5% YES -- still overpriced in our view

Original Research: How Calibrated Are These Markets?

We built an interactive calibration tracker analyzing 7,661 resolved Polymarket markets. Key findings:

Excellent at the extremes. Events priced at 0-5% almost never happen (0.1% actual). Events at 95-100% always happen.
Systematic underpricing in the middle. Markets at 65-75% resolved YES 96.5% of the time.
Overall Brier score of 0.025 -- roughly 10x better than random guessing.

Explore the data yourself: polymarket-calibration.vercel.app