Based on the millions of dollars betted on the market, Spencer Pratt ( Spencer Pratt ) seems to be in a strong position to win the primary elections for Los Angeles Mayor at the beginning of June.

Platt was not originally the expected candidate for mayor. This former reality show star had never held any elected public office, but after a wildfire in Pacific Palisades (Palisades) destroyed his home, he turned his criticism of the city's response into a campaign slogan.

Through prediction markets that include Kalshi, Polymarket U.S., and Polymarket International, speculators flocked to this election. The related transaction volume exceeded 47 million US dollars, making it the most active mayoral election in 2026. On the eve of the primary elections in early June, one market indicated that Pratt had a 78% chance of advancing to the November runoff.

However, the predictions about the market were completely wrong, and Pratt ended up in third place.

However, Platt's supporters used this failed speculation to question the fairness of the election. In response, Dean Logan, the registrar from the Los Angeles County Clerk's Office, stated in a declaration to CNBC: "When people regard the predictions of the forecasting markets as more authoritative than the actual votes that are being counted, these forecasting markets only exacerbate this chaos."

The market's failure to accurately predict the election has also raised the possibility that the market may have been manipulated, making certain candidates appear to have more support than they actually did. Exclusive reporting from CNBC indicates that this risk indeed exists.

The discrepancy between polling surveys, speculation-driven prediction markets, and the final results has become increasingly significant as election betting has become a prominent feature in elections. Odds and betting platforms do not use the methods of traditional political polls, therefore they cannot replace them.

Polymarket and Kalshi are two major predictive market platforms. They allow people to speculate on almost anything – from who will be the champion in the NBA in 2027, to whether the United States will confirm the existence of aliens.

According to data from the Washington-based non-profit organization Anti - Corruption Data Collective ( ACDC ), the betting amounts on contracts related to the 2026 congressional elections on these platforms have exceeded three times those of the 2024 elections. People are also increasingly using these platforms as a source of information. Regarding the mid-term election market, Kalshi indicates that the ratio of viewers to actual traders reaches 5 to 1.

Identify suspicious transactions

Despite the active contract trading and high liquidity in Pratt's election, Kalshi still detected and disclosed a price surge that occurred during these transactions in a case study published in July. Experts explain that a price surge refers to a brief but significant increase in price, which can make one outcome seem more likely to occur. Price surges can be caused by various reasons; for example, new information may change traders' expectations, or a single trader may buy a large number of contracts, driving up the contract price as cheaper contracts are bought out.

Kalshi indicates that on May 20th, a trader spent $1.25 million on the platform to buy a "yes" contract stating that Pratt would become the next mayor of Los Angeles. One of these transactions caused the price of the contract for Pratt's victory to jump from 29 cents to 45 cents, suggesting that the trader believed Pratt's chances of winning had increased from 29% to 45%. However, according to the case study in Kalshi, within 9 seconds, the price returned to its previous level, and the market corrected itself.

Kalshi, the law enforcement supervisor responsible for monitoring abnormal activities on the platform, said, "You can be completely assured that there will be rational traders in the market willing to take on these transactions and pull the price back to its proper level." He also added, "By the end of the day, the price was even lower than it was at the beginning."

However, analysis by ACDC on CNBC shows that during the Los Angeles mayoral election period, there were multiple price spikes for Kalshi and Polymarket International, which increased Platt's chances of victory. This analysis did not take into account transactions involving Polymarket U.S.

The data collected by ACDC confirms the findings of Kalshi. In that case study, Kalshi only provided data within about 30 minutes after large orders were placed. However, the organization found that if the time window was extended to 3 hours after the price spike, the price would be 3 cents higher than before, and this level would persist for about two hours. In the long run, prices would eventually return to normal. But research by ACDC suggests that the longer the price increase persists, the greater the risk that this information is interpreted as market sentiment.

Moreover, as shown by the research of ACDC, election contracts are vulnerable to sudden price spikes – and that was precisely the case in Pratt's election.

The organization found that on Polymarket International and Kalshi, among 167 markets in the congressional market with a value of over $100,000 each, there were 38 markets – approximately 23% – that experienced price spikes causing the contract prices to fluctuate by at least 30 cents.

One example is the contract on Kalshi regarding whether Juliana Stratton ( Juliana Stratton ) would become the Democratic nominee for the Illinois Senate. ACDC states that there were two particularly notable purchases on March 3rd – whereas before the election day, the median price of orders was only 12 dollars. One buyer spent 5,000 dollars to purchase a contract supporting this long-time public official's campaign, pushing the price up from 47 cents to 79 cents. Three seconds later, the price returned to normal at 47 cents; but about a minute later, after another 3,000-dollar contract was purchased, the price soared again to 99 cents. Ultimately, the price fell back to 52 cents, which is 5 cents higher than the starting point – this means that one or more people spent 8,000 dollars and, in the process, pushed up the price to make Stratton's victory seem more likely.

Two weeks later, Straton won the Democratic primary election. Straton refused to comment on the matter, but a spokesperson from her office provided CNBC with a poll released on the same day by Public Policy Polling, which showed Straton in the lead.

Kalshi The spokesperson stated in a declaration to CNBC: "This is quite common on Kalshi, and the market quickly corrected to a price close to fair value."

ACDC It is impossible to know the motivations behind these bets. While making a profit might be a factor, it is not sufficient. In a predictive market, participants can buy and sell contracts with values ranging from 0 to 1 dollar, where the price reflects the traders' perceived likelihood of a certain event occurring. The higher the contract price, the smaller the potential return. Winning contracts are settled at 1 dollar, while losing contracts become worthless. Therefore, a trader who buys for 99 cents can only expect a profit of 1 cent if their prediction comes true, and the transaction fees charged by the predictive market will also erode this profit.

ACDC Researcher Michelle Kendler - Kretsch said, "If theoretically, the accuracy of market forecasting is attributed to price motives, then the lack of such motives raises a question: what real goals do those who engage in such transactions intend to achieve for themselves?"

Affecting the election results might be the answer.

A new study funded by liberal advocacy groups in New York, Brennan Center, for, and Justice, shows that if the forecast market is optimistic about a certain candidate, voters are 10 to 20 percentage points more likely to think that candidate will win, regardless of what the polls say. The survey was conducted in September and covered 1,994 registered voters.

The data shared by ACDC and CNBC also shows that in markets with fewer transactions, it is even easier and less costly to change odds. In the 18,000 active market sessions on Polymarket International, Polymarket U.S., and Kalshi, 88% of the markets can move the price by 10 cents with a bet of less than $1000.

Cracking down on market manipulation

Nevertheless, the market predictions of Kalshi are often correct. The company stated in a website article: "In the U.S. elections since 2024, when Kalshi's market indicates a certain candidate as the winner, that candidate has approximately a 90% chance of actually winning." In addition, The Washington Post reviewed 268 candidates in this year's primaries; these candidates had a 70% to 80% chance of winning on Kalshi or Polymarket at any point within two months before the primaries, and three-quarters of them ultimately won.

Kalshi indicates an effort to alleviate concerns about election distortion from the outside world. Earlier this month, the company added new transparency features to its mid-term election market, helping users understand how much confidence to have in the odds of certain contracts. These features include adding labels to markets with very little trading activity and displaying real-time information streams for individual transactions. Polymarket has also launched its own dedicated monitoring system for elections, "used to detect attempts to manipulate the election market, distort probabilities, or create a superficial market consensus."

Kalshi told CNBC that the company uses algorithms to generate alerts based on parameters of abnormal trading activities, including suspicious transaction times and volumes, as well as price spikes. However, it is not yet clear whether Kalshi's internal processes issued any alerts for transactions related to Pratt; the company has not responded to CNBC's request for clarification.

Kalshi and Polymarket U.S. are both regulated by the U.S. Commodity Futures Trading Commission (CFTC), which is an independent federal agency whose rules are designed to prevent market manipulation and insider trading. Transaction prices and activities are publicly visible; at least in the case of Kalshi, the order books on its derivatives exchanges are also public, allowing people to report and flag any suspicious transactions to the exchanges. In addition, Kalshi and Polymarket U.S. collect "Know Your Customer" information, including full names and social security numbers.

However, the outside world has no way of knowing who is behind each transaction.

Polymarket, the head of global intelligence and investigations, Shana Bautista said: "I used to work at FBI and have also worked in the field of centralized finance outside of the blockchain. If you want to obtain information about a certain case or transaction, you must obtain a subpoena or search warrant in order to get that information."

In April, Polymarket International announced a partnership with blockchain analytics company Chainalysis, which helps this blockchain-based platform track suspicious transactions on its international platforms.

Chainalysis Senior Product Manager Adam Hart demonstrated to CNBC the data collected by the company for each Polymarket trader. For example, Polymarket can see the number of transactions, total trading amount, the proportion of the trader's bets relative to their overall activity, as well as their total profit and loss.

Hart introduced to CNBC a case of collaboration between his team and Polymarket: An American soldier, Gannon Ken Van Dyke, was prosecuted for allegedly using confidential information to bet on the removal of Venezuelan President Nicolas Maduro and the entry of U.S. troops into Venezuela. Van Dyke pleaded not guilty to the charges.

Hart said that this wallet was flagged because it placed bets with exceptionally high returns and never suffered any losses. Van Dyke's identity was eventually confirmed, in part because he hardly concealed it at all: the funds in his cryptocurrency wallet came from a crypto exchange that required users to verify their identities. Hart also mentioned that Van Dyke transferred about $400,000 in profits to a mainstream exchange in the United States as well.

Of course, transactions that seem suspicious may simply be bets placed by shrewd speculators or enthusiastic supporters. This is the case with the anonymous “big player” from France, Theo, who made a bet in support of Donald Trump during the 2024 elections and earned approximately 85 million US dollars. Blockchain analysis firm Bubblemaps is one of the organizations that tracked bets totaling around 60 million US dollars by Theo; these bets were spread across at least 10 Polymarket accounts. Bubblemaps traced these activities back to two cryptocurrency exchange recharge wallets owned by Theo. Theo stated that his betting decisions were based on his own assessments of neighborhood opinion polls as well as telephone contacts with swing states.

Engineering Integrity for Elections

However, this internal investigation did not reassure Jared DeMarinis. As the manager of the Maryland State Board of Elections, DeMarinis is convinced that the only way to protect the integrity of elections is to completely ban betting on election results.

In July, he requested the State Attorney's Office to investigate whether market predictions violated state laws that prohibit betting on election results. A few months later, in September, he prohibited certain individuals and election staff—including election judges and members of the State Election Commission—from participating in election market transactions.

DeMarinis said, "We employed 26,000 election judges, who are the staff at the polling stations. They have access to information before the official counting and tallying of votes begins. They will be aware of certain things. All of this is problematic because it relates to the legality and integrity of the electoral process."

According to data from the Pew Research Center, more than half of the states have laws that make electoral betting illegal in all or some cases. The prediction market believes that these contracts should be regulated by CFTC rather than by individual states, as they are classified as swap contracts and fall under the jurisdiction of federal law. CFTC also supports this view.

However, these laws may not apply to federally regulated predictive markets. With the support of New Jersey and dozens of other states, stakeholders have requested the Supreme Court to rule on whether the rules of CFTC are sufficient to address concerns regarding elections, as well as whether each state has additional legal standing to file lawsuits.

Meanwhile, prediction markets continue to intensify their efforts in launching more election-related products.

Kendler-Kretsch said, "We don't have much information regarding the accuracy of market predictions. When a transaction does not unfold as per the election results, it is not about causing people to doubt the market predictions and their accuracy, but rather about causing them to doubt the elections themselves."

—— Ananya Chetia contributed to this report.

Revelation: CNBC has a business relationship with Kalshi, which includes customer acquisition cooperation and a small amount of equity investment.