Study weighs prediction markets and AI against traditional polling
Prediction-market volumes and attention are climbing ahead of the U.S. midterms, with platforms including Polymarket and Kalshi drawing renewed focus. Brazil's 2026 first-round presidential election became a live test case: Polymarket's pricing correctly forecast Senator Flávio Bolsonaro's lead over President Luiz Inácio Lula da Silva when traditional polls had not. Brazil first-round markets on Polymarket reached $165 million in trading volume, according to the Polymarket Builders Program rankings. A WBUR segment separately examined whether surging election markets could influence outcomes. An academic study also weighed prediction markets and AI simulations as alternatives to traditional polling.
The Brazil result gives prediction-market operators their most visible accuracy win to date against traditional polling, and they will weaponize it in every regulatory and investor conversation ahead. Polymarket and Kalshi now have a concrete case to cite when state attorneys general call their products gambling: traders collectively outperformed professional pollsters on a live national election. That narrative matters because the $3 million statehouse lobbying push and pending CFTC preemption rules both turn on whether lawmakers see these venues as information markets or as betting.
If platforms can stack two or three more election cycles where they beat polls, the political cover they gain may outrun the legal risk. For traders, the upside is validation; the risk is that expected-accuracy pricing compresses the edge that early contrarian bets used to capture. The platforms' real fight is no longer against polls — it is against state gambling codes that do not care who guessed right.