Midterm election prediction markets are extremely vulnerable to manipulation, ACDC warns
A single bet under the permitted personal campaign contribution would move the price of 97% of live congressional election markets 5 cents or more. Where this has already happened, most markets did not self-correct to the previous probability.
New analysis from the non-profit research, investigations and advocacy group the Anti-Corruption Data Collective (ACDC) finds that prediction markets on the upcoming midterm elections are extremely vulnerable to deliberate price moving.
ACDC examined 18,780 markets related to the 2026 U.S. congressional midterm elections on Polymarket, Polymarket US and Kalshi, and the live order book of the 11,011 markets still active on 1 September on the three platforms.
The analysis revealed that in 97% of live congressional markets, $3,500 — the maximum allowed personal campaign contribution — will buy a price shift of five cents or more. Ninety-four percent of markets would move ten cents following a single bet under $1,000.
Many markets are extremely cheap to move. As of 1 September, on 64% of markets on Polymarket and Polymarket US, a user would only need to place an order of around $1 to shift the price five cents, without necessarily finding a counterparty or fulfilling the actual trade. On Kalshi, 63% of markets cost $25 or less to move the price five cents.
Moved prices stick, and build
On Polymarket, where wallet-level transaction data is available, at least one wallet has bet the amount needed to move a price by five cents in 83% of 3,206 active congressional markets.
ACDC found hundreds of 2026 Congressional midterm election markets on Polymarket in which a price move by a single wallet remained at its new level for 24 hours. These shifts typically lasted around four days. ACDC also identified dozens of cases when the price not only stuck, but built momentum, leading to continued shifts in the same direction. Only about half of the price-moves that stuck or built momentum were ultimately in the correct direction.
This vulnerability is a function of an explosion in midterm markets that has outstripped apparent demand from users. Research has found that thin prediction markets are susceptible to manipulation. The three major platforms have already opened 40 times as many markets on the midterms as were available for the entire 2024 cycle. While the top senate races in Texas, Michigan and Maine have seen tens of millions of dollars in trading, under $1000 has been bet in more than three quarters of midterm election markets. Markets on underdog candidates that trade at under 5 cents are the cheapest and easiest to move. It typically costs just $23 to double an underdog’s implied probability of winning
ACDC’s briefing calls for greater caution from the media, which routinely cites prediction market prices with reference to only one platform and without important information about recent price shifts and the wallets behind them.
The Texas Republican Primary: A Case Study
Even in the most traded midterm race in the country, the Texas Senate race, a single wallet moved the price of a candidate 11 cents with a bet of $1240 in December 2025. Soon after, Newsweek published an article based on the new Polymarket price. Later, in February, larger bets pushed the same candidate’s price down and one of his opponents’ up. Occurring during a gap between external polling information, these price shifts may have produced a reaction by automated wallets in other, related markets on the same race. The case study demonstrates that absent reliable external data, markets may actually amplify attempts at manipulation, ACDC said.
Michelle Kendler-Kretsch, Research Fellow, Anti-Corruption Data Collective, said:
“Prediction market platforms like to say that the profit motive of their users leads to better accuracy than polls, and that prices self-correct when someone tries to move them. But American elections attract millions of dollars in donations that are not driven by a short-term profit motive. The conditions for accuracy are not being met in markets about the November midterms and primaries. The data shows that this is not a theoretical risk, but a real vulnerability. Platforms need to recognize this and only open markets where there is sufficient demand and attention for the wisdom of the crowd to apply.”
David Szakonyi, Co-Founder, Anti-Corruption Data Collective, said:
“Major prediction market companies are receiving financial backing from members of the Trump family, deference from Trump-appointed regulators and encouragement from the White House itself. Meanwhile, they are flooding the zone with a huge number of thinly traded electoral markets. With just a few dollars, anyone can create the perception that a candidate has widespread momentum. Congress needs to get to grips with the risks that prediction markets pose to the integrity of our elections. The ease of manipulating markets is opening the door to a new type of election misinformation.“
***ENDS***
NOTES
Download the full briefing: “Gambling on Democracy — Briefing 2: Price Vulnerabilities in Prediction Markets”
ACDC will continue to update its analysis while diving deeper into election integrity risks related to prediction markets. Visit https://acdatacollective.org/work/gambling-on-democracy for updated data and downloads.
About ACDC
The Anti-Corruption Data Collective (ACDC) leverages public and private data to expose transnational corruption, illicit financial activity and corporate opacity. Our research and investigations seek to document and reduce the harms that corruption causes to human security, democratic participation and environmental sustainability: to people, politics and planet.
Media Contact
Anti-Corruption Data Collective
team@acdatacollective.org
www.acdatacollective.org
Gambling on Democracy — Briefing 2: Price Vulnerabilities in Prediction Markets