30:07
This is a machine. If we want to burn this town to the ground- The race for mayor of Los Angeles was unconventional in every way. You can do it, Spencer. Let him in. Former TV reality show star Spencer Pratt versus a city council member and the incumbent mayor in the June primary. I know what the consequences of failed leadership are. That's why I'm running for mayor. AI-generated campaign ads and prediction markets buzzing with activity. In fact, it is the highest-traded mayoral race in 2026, surpassing $47 million in bets,
a highly liquid market that theoretically should reflect the will of the people. Raman overtook reality star turned wannabe politician Spencer Pratt. Except it didn't. The prediction markets not only got it wrong leading up to the election, but during the race, there were numerous unusual instances of price spikes on Kalshi and Polymarket International that pushed the odds in Pratt's favor, according to an exclusive analysis for CNBC by the Anti-Corruption Data Collective, or ACDC, a nonprofit that brings together journalists, academics, policy experts and data scientists who work to expose corruption.
Michelle Kendler-Kretsch is a researcher at ACDC. I feel extremely worried that we don't have enough information, we don't have enough data in order to truly know what may happen come November. In the L.A. Elections, the discrepancy between the prediction market odds and the actual outcome fueled voter doubt about the integrity of the election itself, a precursor of what could come this November. Polymarket and Kalshi are the two largest prediction market platforms. They let people trade on almost anything, from what team will be the 2027 NBA champion to will the U.S. confirm that aliens exist to who will win the upcoming midterm elections.
Users can buy and sell contracts valued between 1 cent and 99 cents. The higher the price, the more likely traders think the event will happen, but the lower the payout. For instance, if a trader buys a contract for 80 cents, that means they think there's an 80% probability the event will happen, but only a 20-cent payout. I'll check the live results on Kalshi. Already on these platforms, the amount wagered on contracts related to the 2026 congressional elections has more than tripled total bets on the 2024 races, and people are increasingly using the platforms as a source of information.
For midterm election markets specifically, Kalshi reports viewers outnumber actual traders by a 5-1 ratio. And as election wagers become more and more part of the national discourse, the concerns are growing about the impact the odds can have on voter behavior and whether those odds are vulnerable to manipulation. The stakes are high. Prediction market odds can be fuel for a campaign, a favorable mention in a news article, or even a pitch for a campaign contribution. Jared Demarinis is a state administrator for the Maryland State Board of Elections, which is overseeing dozens of contests on the ballot this Election Day.
My job is to make sure the right person was elected. How worried are you that there will be questions about the validity of any of those races because of the prediction markets? I think that the prediction markets will amplify heightened scrutiny as well as apprehension regarding the current election landscape. There's distrust. Right.
Can I just get one picture? I'm sorry. I only got into the race because nobody else was going to run. On the eve of the June primary, Kalshi gave Pratt a 78% chance of advancing to November's runoff. Ultimately, he finished third and did not advance. Despite it being a heavily traded liquid market, ACDC found anomalies on Kalshi. Across May, every time the market said, 'Okay, let's drop the price, there's negative news,' there was also part of the market that said, 'Hang on, let's increase this price.'
Could it be that Pratt's campaign was just better at countering negative news? Ultimately, all of it's possible. We can't ascribe intent to any trading behavior. All we can say is, overall, for the dips, we saw recoveries and we didn't see that in the other candidates. On Polymarket International and Kalshi, ACDC also found price spikes, temporary but large price jumps that can make outcomes look more likely to happen. Price spikes can happen for a variety of reasons, such as new information, changing trader expectations, or a trader buying a large number of contracts and driving up the prices as cheaper contracts get bought up. In the Pratt race,
the group found in one series of trades on Kalshi $56,000 in contracts were purchased at once, ranging from 73 cents to 99 cents, temporarily pushing the price up to 99 cents. With every tick higher, the potential payout on those contracts went lower. All in all, the trader or traders spent $56,000 and lost out on about $8,000 in potential profit. So then there's a question of why would a trader do something like this? We simply don't know. But spiking to 99, losing that profit margin, there's many questions on why someone would do that. So to you that's a red flag. A price spike doesn't have a profit motive. And if the theory is that prediction markets are accurate because of a price motive,
it begs the question of what type of goal does someone making a trade like that really have for themselves? A spokesperson for Kalshi told CNBC, 'Mispriced markets represent free money sitting on the table, and traders are financially incentivized to correct that mispricing quickly.' In a statement, the L.A. County Clerk's office wrote that 'During the June primary, we saw how changes in candidate standings generated significant online criticism, speculation, and questions about the integrity of the process. Prediction markets can add to that confusion when people treat their projections as more authoritative than the actual votes being counted.'
ACDC says election contracts can be vulnerable to unusual price spikes, similar to those in the Pratt race, and that it identified price spikes happening in midterm races on Polymarket International and Kalshi. In data shared exclusively with CNBC, ACDC found that of 167 congressional markets worth more than $100,000, 38, or about 23%, experienced price spikes, moving the contract price 30 cents or more. One example: Will Juliana Stratton be the Democratic nominee for the Senate in Illinois? On March 3rd, a buyer spent $5,000 on contracts, taking the price from 47 cents to 79 cents.
The price normalized after three seconds, going back down to 47 cents. But about one minute later, it shot back up to 99 cents after $3,000 of additional contracts were purchased. The price eventually traded back at 52 cents, 5 cents more than where it started. That means a person or persons spent $8,000, resulting in a push up in price that made a Stratton victory look more likely. Two weeks later, Stratton won the Democratic primary. In a statement to CNBC, a spokesperson for Kalshi wrote, "As is common on Kalshi, the market corrected quickly to a price near fair value." Stratton declined to comment
on this story, but a spokesperson from her office pointed to a poll published the same day as this price spike that showed Stratton leading. Data shows moving the odds may be easier and cheaper than you might think, especially when liquidity is thin, meaning not many wagers are placed. Across more than 18,000 active congressional markets on Polymarket International, Polymarket U.S. and Kalshi, 88% could have their prices moved by 10 cents with a single bet under $1,000. In August, Kalshi published a study that analyzed more
than 2 million markets. It found deeper participation may sharpen prices rather than distort them, but even markets with as few as 20 traders can be accurate. In a statement to CNBC, a Polymarket spokesperson wrote, "Traders will recognize an opportunity to turn a profit off of an artificial boost. There is an incentive to correct, which is the entire point of markets and why we are more accurate than traditional polling." When you look at actual trading behavior, a manipulator and a really enthusiastic voter will look exactly the same, right? They're putting lots of money on a candidate.
They're moving the price. What we're relying on is the platforms to be looking for and monitoring manipulation. Kalshi gave us an exclusive inside look at how it combats suspicious trading from its daily morning meeting, where teams discuss unusual trades, to the dashboards it uses to identify those trades. Prediction markets are now forecasting a Democratic majority. CNBC has a commercial partnership with Kalshi and uses the company's market data across its TV, digital and subscription platforms. Robert Denault is the head of enforcement at Kalshi. He's tasked with tracking unusual activity on the platform. Are you worried, especially as we approach the midterms, about the incentive of influence, of influencing outcomes, influencing voter behavior?
We watch election markets with a particular care at Kalshi, not just because we have to from a regulatory perspective, but because there's such a unique characteristic and role that these markets are going to play in terms of how people ingest information about elections. In fact, Kalshi says its track record shows how accurate its markets are. Across U.S. elections since 2024, when Kalshi's markets favored a candidate to win, that candidate went on to win roughly nine times out of ten. And even if there are price spikes for candidates, whether caused by enthusiastic backers or bad actors, the market will self-correct.
We saw that happen in the L.A. mayor's race. We did see a person who spent a great deal of money in a single day, over $1 million, on Spencer Pratt, a candidate for mayor there. The price corrected in about nine seconds, because other traders see that as an opportunity. You can absolutely guarantee there are rational traders in that market willing to take those trades and bring that price right back down to where it belongs. And by the end of the day, his price was lower than when it began. Kalshi published a case study on the Pratt race. The title is called Price Distortion in L.A.
Mayor's Race Corrects Within Seconds. Based on your findings, what actually happened? Was that true? Didn't it correct within seconds? In the snapshot that they show, seconds later, someone else is paying a reasonable price for the contract. But what we see afterwards, Kalshi tells us that it's the same trader who's continuing. We can't see that. But this is Kalshi is telling us. They continue to buy and buy, and that actually elevates the price when you look at not a half-hour window, but a three-hour window. So they left out the three-hour look.
They were looking at what happened immediately after. And that's the vulnerability in price shifts. So take that three-hour window. It's up three cents. That's a 3% chance. This is on whether Spencer Pratt wins the November election. But a trader could look at that and say, 'Spencer Pratt's doing well on November. I think he'll also do well in the runoff. I think he'll also do well on advancing.' Pratt did not respond to CNBC's requests for comment.
There's a Wild West sort of perception of predictions markets. How much of that is true in your view? It's untrue. So much of this information is public. On derivatives exchanges, the entire order book is available to people to look at, and they're able to actually escalate and flag trades to exchanges themselves. Kalshi is federally regulated by the Commodities Futures Trading Commission, meaning it must follow rules designed to prevent market manipulation and insider trading. Trading prices and activity are publicly visible, and Kalshi internally collects and enforces "Know your Customer" or KYC information. But the outside world has no insight into who is behind each trade.
We're putting a lot of trust in Kalshi to use their KYC system to find whether that's insider trading, whether that's possible market manipulation, all of that we're entrusting into Kalshi because we don't have the user-specific information. And in an effort to increase transparency, Kalshi added new features to its midterm election markets to show market liquidity, accuracy forecasts and real-time individual trades, while Polymarket U.S. is a CFTC-regulated platform available to Americans and has the same KYC requirements as Kalshi, its bigger, more popular platform is Polymarket International. Its users are not required to provide any identifying information, but Chana Bautista, the head of global intelligence and investigations at Polymarket,
says the international platform is inherently transparent because the time, the size and the wallet of every trade is publicly visible on the blockchain. The international platform does not require users to submit "Know Your Customer" sort of information, so identifying features. Wouldn't it be a lot easier if the platform required that? I would argue honestly the opposite. So for reference, I formerly worked at the FBI and in centralized finance off the blockchain. If you want to get information on a case or a
certain transaction, you have to serve either a subpoena or a search warrant to be able to get that information. Isn't it difficult, though, to attach a wallet to an individual? You would think, but no. About 95% of the volume on our platform is attributed to known KYC sources. In order to participate or gain funds on the blockchain, most of the time and most of the cases that we see originate from on-chain exchanges or centralized exchanges. That enables us to continue and work collaboratively with the private sector in this regard, too. In April, Polymarket International announced a partnership with blockchain analytics firm Chainalysis,
which helps the platform track suspicious trading. Adam Hart is a senior product manager at Chainalysis. He gave CNBC an exclusive look at some of the data the firm sees about each trader. We can see the total number of trades. How many buys are they making? How many sells are they making? How much profit and loss are they making? These are all kind of key factors that feed into the larger model that decides is this normal or is this anomalous? Hart also walked us through an investigation that his team worked on in collaboration with Polymarket.
Do not block his way, please. The case of Gannon Ken Van Dyke, the U.S. soldier the DOJ charged with using classified information to make bets that Venezuelan President Nicolas Maduro would be removed from office and that U.S. forces would enter Venezuela. Van Dyke has pled not guilty. This wallet would have been flagged by our system for making highly, highly profitable bets and never losing. And then if we investigate their activity, we can kind of see they received their funds from basically two wallets that were both funded out of a mainstream, fully compliant U.S. exchange. And then the winnings, the $400,000 or so of winnings plus the initial investment, was cashed out also to that same mainstream exchange.
So this is a very straightforward case because there's no effort to hide anything. Yes. Olivia is the deputy chief legal officer at Polymarket. Part of her role includes working with regulators on cases involving suspected insider trading. In the soldier case, in April 2026, the DOJ indicted the soldier for insider trading on Polymarket. According to the indictment, the unusual trading activity was flagged by journalists and social media. Did Polymarket miss it? So I can't speak to any particular cases, but we are regularly identifying activity in this case and it's been recognized by the prosecutors in this case.
We provided information to them. We flagged it affirmatively. So our systems did not miss it. Even though it was social media and media reports that were attributed to finding the suspicious activity. Our systems did detect this trading activity. How high of a priority for you is suspicious activity related to the midterm elections. Very high priority. One of the interesting pieces about election markets is that you could anticipate potential activity that looks different from insider trading, for example, trading that might be designed to skew the odds to look a particular way for a
particular candidate, for it to seem like a candidate's winning or losing. So we have to design controls to account for that as well. And even trades that may look suspicious could just be the wagers of a sharp trader or enthusiastic backer. That was the case of Theo, or the so-called French whale, who made about $85 million betting on contracts favoring Donald Trump in the 2024 elections. Bubblemaps is one of the on-chain analytic firms to trace Theo's approximately $60 million in wagers across at least ten Polymarket accounts.
The activity pushed up Trump's odds, but Bubblemaps traced the activity to two crypto exchange's deposit wallets owned by Theo, who said he made the bets based on his own neighbor polling and calls to battleground states. Still, for Demarinis, the only way to protect election integrity is an outright ban. In July, he asked the Office of the State Prosecutor to investigate whether prediction markets violate state laws prohibiting wagering on election outcomes. In September, he also prohibited election staff from trading in election markets. We employ 26,000 election judges that are poll workers, right?
They have access to information right before it's getting tallied officially and counted. They'll know certain things. This is all problematic because it goes to the legitimacy and integrity of the electoral process. More than half of all states have laws making election betting illegal in some or all circumstances, according to Pew Research Center. But there are questions about whether these laws apply to prediction markets. Polymarket U.S. and Kalshi argue they are federally-regulated financial exchanges governed by the CFTC, and buying a contract is akin to trading.
States argue they are unlicensed gambling entities and participants are betting. New Jersey, backed by dozens of states, has asked the Supreme Court to weigh in. In the meantime, more and more bets continue to be placed, increasing the potential influence on the elections themselves. So if there was manipulation of the odds and it had an impact on the public, we won't know that until afterwards, until it's all done. And that's inherently what the risk is here. We're doing something new. We don't have a lot of information on whether prediction markets are accurate. And what concerns us is when it doesn't go the way of an election, instead of sowing doubt in prediction markets
and their accuracy, it's sowing doubt in the election itself.
