# Can We Trust Election Prediction Markets?

An episode of [Big Take](https://invisiblenetwork.ai/podcast/1578096201), 29 Jul 2026, 19 min.

Audio: https://podtrac.com/pts/redirect.mp3/tracking.swap.fm/track/UVBrz8bN8aM2Xe47PEPu/traffic.omny.fm/d/clips/e73c998e-6e60-432f-8610-ae210140c5b1/825d4e29-b616-46f4-afd7-ae2b0013005c/09387b14-ff8c-4237-b3c1-b496015934ee/audio.mp3?utm_source=Podcast&in_playlist=8b1dd624-a026-43e9-8b57-ae2b00130066

With the US midterms fast approaching, voters and the media are increasingly looking to a new forecasting tool: prediction markets. The leading platforms, Kalshi and Polymarket, have been touted as a truth serum for political forecasting — arenas for mass betting that reveal what people really think when they’ve got money on the line. But a new Bloomberg investigation has found evidence that these markets are vulnerable to manipulation. On today’s Big Take podcast, host Sarah Holder and Bloomberg investigative reporter Liam Vaughan dig into what can happen when prediction markets mix with politics — and how cheap and easy it can be to sway them. Read more: How a Few Hundred Dollars Could Manipulate Election Prediction Markets We have a special Bloomberg subscription offer for podcast listeners at Bloomberg.com/podcastoffer. Hosted by Sarah Holder; Produced by Victor Swezey and David Fox; Reported by Liam Vaughan; Edited by Tracey Samuelson. Fact-checking by Julia Press; Engineering by Alex Sugiura. Senior Producer: Naomi Shavin; Deputy Executive Producer: Julia Weaver. Executive Producer: Nicole Beemsterboer. See omnystudio.com/listener for privacy information.

## Transcript

[0:02] Speaker 1: Bloomberg Audio Studios, podcasts, radio news. When Matt Mahon officially
[0:10] Speaker 1: entered the race for governor of California in late January,
[0:14] Speaker 1: he wasn't exactly a household name. Mahan was the mayor
[0:17] Speaker 1: of San Jose, a centrist with ties to Silicon Valley.
[0:22] Speaker 1: He was pulling around four percent, But on the morning
[0:25] Speaker 1: Mahan announced his candidacy, he got a boost from an
[0:29] Speaker 1: unexpected place.
[0:30] Speaker 2: On poly Market, which is the pre eminent political prediction
[0:34] Speaker 2: market in the world. This suddenly a trade for forty
[0:38] Speaker 2: four thousand dollars that comes into that market to back
[0:43] Speaker 2: Matt Mahon's candidacy.
[0:45] Speaker 1: That's Liam Vaughn, a reporter and editor on Bloomberg's investigations
[0:49] Speaker 1: team based in London. He's been covering prediction markets, which
[0:53] Speaker 1: allow users to bet on the outcomes of real world events.
[0:57] Speaker 2: Suddenly his odds go from the low teens all the
[1:00] Speaker 2: way up to ninety six percent, which this huge spike.
[1:05] Speaker 1: If you bet ninety six cents that Mahann would win,
[1:07] Speaker 1: you'd get paid out a dollar if he did. But
[1:10] Speaker 1: before that big bet, you only needed to bet about
[1:13] Speaker 1: twelve cents for the same payout.
[1:15] Speaker 2: And essentially the price fluctuates between zero and one hundred,
[1:21] Speaker 2: and that reflects the apparent odds, you know, according to
[1:25] Speaker 2: that market of an event happening. So when Matt mahanzods
[1:28] Speaker 2: go from like twelve percent to ninety six percent, at
[1:31] Speaker 2: least for a few moments, what that market's saying is
[1:35] Speaker 2: there's a ninety six percent chance that he's going to
[1:39] Speaker 2: be the governor of America's largest state.
[1:42] Speaker 1: Quickly, other traders on polymarket saw an opportunity to make money.
[1:47] Speaker 1: They started taking the other side of the bet, wagering
[1:50] Speaker 1: that Mahon would lose. That started pushing his odds back down.
[1:54] Speaker 2: But the interesting thing about this market is it doesn't
[1:56] Speaker 2: bring it back anywhere close to where it started. It
[1:59] Speaker 2: goes to thirty six percent, which is still very high
[2:02] Speaker 2: for this guy who hasn't got much name recognition in
[2:05] Speaker 2: a market full of, like, you know, very big big names.
[2:09] Speaker 2: And so immediately you start to see press coverage. For example,
[2:14] Speaker 2: in the New York Post, there's a piece that runs
[2:16] Speaker 2: that afternoon that says the smart money is backing Matt
[2:20] Speaker 2: Mahon for California governor.
[2:24] Speaker 1: Mahon did not respond to requests for comment. Ultimately, his
[2:28] Speaker 1: polymarket bump wasn't enough to win him the primary.
[2:31] Speaker 2: But eventually kind of gravity caught up. Other candidates with
[2:35] Speaker 2: more money and more name recognition prevailed, and ultimately Mahan
[2:39] Speaker 2: kind of quietly exited stage.
[2:41] Speaker 1: Left, but the forty four thousand dollars question remained. Who
[2:47] Speaker 1: had made the bet?
[2:48] Speaker 2: We don't know very much, and I think that that's
[2:50] Speaker 2: telling in and of itself because poly market is crypto based,
[2:54] Speaker 2: so the authorities don't know who it is, and unless
[2:56] Speaker 2: this individual comes forward, nobody else knows who it is.
[3:01] Speaker 1: Whoever it was had shaped the conversation around a major
[3:04] Speaker 1: primary race with one big polymarket bet, and with the
[3:08] Speaker 1: US midterms fast approaching, it's just one example of how
[3:12] Speaker 1: prediction markets could be gained in elections.
[3:15] Speaker 2: What's fascinating to me about this incident is that it
[3:18] Speaker 2: highlights the inherent vulnerability of some of these markets to manipulation.
[3:24] Speaker 2: And so the questions that I was sort of keen
[3:26] Speaker 2: to explore were like, well, how is that able to happen?
[3:30] Speaker 2: What safeguards are in place to stop it happening? And
[3:33] Speaker 2: as we go into the midterms, and you know, elections globally,
[3:37] Speaker 2: is this something that we're going to see much more.
[3:39] Speaker 1: Of I'm Starah Holder, And this is the big take
[3:46] Speaker 1: from Bloomberg News today on the show the dangers of
[3:50] Speaker 1: mixing prediction markets with politics. Until relatively recently, on politics
[4:01] Speaker 1: wasn't allowed in the US.
[4:03] Speaker 2: Previously, there was only a couple of forums that were
[4:06] Speaker 2: allowing political gambling, and they were associated with universities, and
[4:11] Speaker 2: so it was, you know, studied like can these things
[4:14] Speaker 2: reveal good information? And are they useful forecasting tools?
[4:17] Speaker 1: For years, that restriction on political betting was enforced by
[4:21] Speaker 1: the Commodity Futures Training Commission, the federal agency responsible for
[4:25] Speaker 1: regulating derivatives markets in the US. But in twenty twenty three, CALCI,
[4:30] Speaker 1: the largest prediction market platform in the US and polymarket's
[4:34] Speaker 1: chief competitor, pushed for those restrictions to be loosen.
[4:38] Speaker 2: The CFTC undertakes a review, and ultimately it concludes, as
[4:43] Speaker 2: its predecessors had, that this is a bad idea, that
[4:46] Speaker 2: it's against the public interest, that it's going to potentially
[4:49] Speaker 2: encourage people to either try and make money from democracy
[4:53] Speaker 2: or try and influence the outcome of elections, and therefore
[4:57] Speaker 2: it should be banned. That then goes before a judge,
[5:00] Speaker 2: but ultimately the judge decided betting on politics is not
[5:04] Speaker 2: gambling or it's not gaming as it's written in this statute.
[5:08] Speaker 2: So it's a kind of nuanced argument, but ultimately it
[5:11] Speaker 2: means that Kushi and its competitors are allowed for the
[5:15] Speaker 2: time being at least to allow mass scale gambling on
[5:20] Speaker 2: US politics.
[5:22] Speaker 1: That opened the floodgates right before the US presidential election.
[5:27] Speaker 2: And so in the presidential election of twenty twenty four,
[5:30] Speaker 2: I think it's like four billion dollars is traded. And
[5:34] Speaker 2: when both Kushi and Polymarket predict that Trump is going
[5:40] Speaker 2: to get in and the polls again are sort of
[5:43] Speaker 2: prevaricating in some sets of tie and the prediction markets
[5:47] Speaker 2: perform much better in that question. That kind of adds
[5:49] Speaker 2: to this sense that this is a good thing and
[5:53] Speaker 2: this is something that should be embraced.
[5:54] Speaker 1: To prediction markets and their boosters. This outcome seem to
[5:58] Speaker 1: confirm something important that by harnessing the wisdom of the crowds,
[6:02] Speaker 1: crowds with real money on the line, these markets can
[6:06] Speaker 1: predict the future more accurately than traditional polling can. What
[6:10] Speaker 1: happens next when the Trump administration takes office. How does
[6:14] Speaker 1: the regulatory approach to these markets change.
[6:16] Speaker 2: I mean it's a complete one eighty. Essentially, the Trump
[6:19] Speaker 2: administration comes in, they put their own people in charge
[6:22] Speaker 2: of the various regulators, and at the CFTC they get
[6:25] Speaker 2: rid of all the Democrats and essentially put in some
[6:30] Speaker 2: conservative commissioners who are predisposed to and are supporters of
[6:35] Speaker 2: prediction markets. You know, essentially you see a real kind
[6:39] Speaker 2: of shift in the administration's attitude towards these markets, and
[6:44] Speaker 2: as a result, they explode.
[6:46] Speaker 1: In December twenty twenty four, the monthly trading volume on
[6:50] Speaker 1: Calshi was two hundred and twenty six million dollars. A
[6:53] Speaker 1: year later, it was six point four billion dollars. At
[6:58] Speaker 1: this point, how do Calshi and pol Market's approaches to
[7:01] Speaker 1: political wagering differ.
[7:03] Speaker 2: Yeah, they're kind of the yin and yang of approaches.
[7:06] Speaker 2: CAW she has spent a huge amount of time and
[7:10] Speaker 2: energy trying to position itself as the official market. It
[7:15] Speaker 2: does KYC know your customer, So anyone that signs up
[7:19] Speaker 2: has to provide details of who they are, where they live,
[7:23] Speaker 2: and so anything that goes wrong or any sort of
[7:26] Speaker 2: attempt at manipulation cow she should in theory know who's
[7:30] Speaker 2: behind that poly market is crypto first, and you don't
[7:35] Speaker 2: need to provide your bank details, and essentially you could
[7:38] Speaker 2: do it all anonymously. You could see via sort of
[7:41] Speaker 2: a nu miracle or a kind of user name what
[7:46] Speaker 2: that account is up to on a kind of rolling basis,
[7:49] Speaker 2: but you just don't know who it is.
[7:52] Speaker 1: So Liam there was another moment that really marked the
[7:54] Speaker 1: entrance of prediction markets into sort of the polling mainstream.
[7:57] Speaker 1: After the election, after the shifted regular reapproach, Polymarket and
[8:02] Speaker 1: Calshi made a string of deals with major news outlets.
[8:07] Speaker 2: Yes, so I would describe it as a kind of
[8:09] Speaker 2: feeding frenzy of deal making between Calshi and poly Market
[8:14] Speaker 2: and the mainstream media companies. So calsh have done deals
[8:18] Speaker 2: with CNN, with CNBC, with Fox and several others. Poly
[8:24] Speaker 2: Market have done deals with the Wall Street Journal, with
[8:27] Speaker 2: Dow Jones X Substack. So there was a real strong
[8:32] Speaker 2: financial incentive for some of these media companies to be
[8:36] Speaker 2: citing these prediction market odds. And that's the kind of
[8:40] Speaker 2: contrast with poles because even though they carry polls, they're
[8:43] Speaker 2: not getting paid by the polls to carry them.
[8:46] Speaker 1: We don't know the details of all of these deals,
[8:49] Speaker 1: but in the agreement between CNBC and cal she, CNBC
[8:53] Speaker 1: can earn a fee every time someone signs up for
[8:56] Speaker 1: Calshi from its website.
[8:57] Speaker 2: The other criticism which we get into in the p
[9:00] Speaker 2: is that quite often, because these markets aren't very big
[9:04] Speaker 2: outside of the major political events, they're easy to manipulate,
[9:09] Speaker 2: and yet they're still cited regularly, just to give one example,
[9:13] Speaker 2: but they're far from alone. But CNN has got the
[9:15] Speaker 2: show called The Odds, and they will regularly use Calshy
[9:20] Speaker 2: Odds as a sort of frame to talk about current
[9:22] Speaker 2: political events. But often they cite markets that are very small.
[9:28] Speaker 2: So you know, there's a discussion around who's going to
[9:30] Speaker 2: be the next Speaker of the House, and we looked
[9:33] Speaker 2: and at the time it had like a total of
[9:35] Speaker 2: sixteen thousand betting across its lifetime. So sort of talking
[9:40] Speaker 2: about a market where every day there's like, you know,
[9:42] Speaker 2: a couple of hundred or few hundred amounts being traded.
[9:45] Speaker 2: So the question is like, is that a market that's
[9:47] Speaker 2: robust enough to really be cited by mainstream media organizations.
[9:54] Speaker 2: And there's some people that argue no, absolutely not.
[9:58] Speaker 1: Yeah, I mean it's like the mat Mahannah example, the
[10:00] Speaker 1: Matt mahanon market was vulnerable to being moved by this
[10:03] Speaker 1: one big bet. Some major outlets ran that story about
[10:06] Speaker 1: Mahan's polymarket odds and the governor's race without additional context.
[10:12] Speaker 1: Does that happen a lot?
[10:13] Speaker 2: I mean, it happens more than you'd think. Definitely, And
[10:15] Speaker 2: it's not helped by the fact that because they've done
[10:18] Speaker 2: these deals, they're sort of obliged to consistently refer to
[10:22] Speaker 2: these polls.
[10:23] Speaker 1: CNN says that quote, prediction markets offer just one source
[10:27] Speaker 1: of data that journalists can use in telling a story unquote,
[10:30] Speaker 1: and that CNN staff are prohibited from trading A Calshi
[10:34] Speaker 1: spokesperson says that its media partners have quote full control
[10:38] Speaker 1: over what and how they report, and we don't tell
[10:41] Speaker 1: them which markets to cite unquote. CNN and Calshi both
[10:45] Speaker 1: declined to comment on how much CALCI pays CNN to
[10:47] Speaker 1: cite its data. Bloomberg LP, the parent of Bloomberg News,
[10:51] Speaker 1: provides Polymarket and Calshi data as part of its broader
[10:54] Speaker 1: data offering for clients. What do prediction markets themselves say
[10:59] Speaker 1: about their utility as forecasters of election results and these
[11:04] Speaker 1: risks that they.
[11:05] Speaker 2: Pose so in terms of their ability to forecast events,
[11:11] Speaker 2: Calci and poly Market point to research that suggests that
[11:15] Speaker 2: they are consistently very good. They are at pains to
[11:19] Speaker 2: say that they should be viewed as additive to existing
[11:23] Speaker 2: forms of forecasting, you know, whether that's polls or reporting
[11:28] Speaker 2: or whatever else. In terms of the issues that we
[11:33] Speaker 2: raise in the piece around the vulnerability of these markets
[11:36] Speaker 2: to manipulation, Cauci would say, we are not the same
[11:40] Speaker 2: as poly market, and you know, we take our responsibilities
[11:45] Speaker 2: to identify our customers and monitor what they're doing very carefully,
[11:51] Speaker 2: and if there is anything untoward going on, then we
[11:55] Speaker 2: will find it. Poly Market would say a couple of things.
[11:58] Speaker 2: One is that in terms of market manipulation, these markets
[12:03] Speaker 2: have a sort of degree of ability to self correct,
[12:06] Speaker 2: which is demonstrated, and so even if there are attempts
[12:09] Speaker 2: at manipulation, they'll be quickly remedied. And so they would say,
[12:14] Speaker 2: I think that the Mayhan example is testament to that
[12:17] Speaker 2: because ultimately, whilst his odds did move, they did also
[12:21] Speaker 2: come back as the market kind of responded to that.
[12:24] Speaker 1: Coming up, how prediction markets can be used to help
[12:28] Speaker 1: spread misinformation and so doubt in election results around the
[12:39] Speaker 1: time Matt Mahon was surging on Polymarket. In the California
[12:43] Speaker 1: Governor's race, another candidate was riding his own wave of
[12:46] Speaker 1: prediction market height in the Los Angeles mayoral primary, Republican
[12:51] Speaker 1: former reality TV star Spencer Pratt Coultrey.
[12:54] Speaker 2: Prediction markets are showing a fifty nine percent chance at
[12:57] Speaker 2: Spencer prep and it's just second place in.
[12:58] Speaker 1: The first round of the LA mayoral Actually.
[13:02] Speaker 2: What if he clears fifty one percent tonight, cal She's
[13:04] Speaker 2: saying he's going to go.
[13:06] Speaker 1: When the votes finally rolled in, Pratt came in third,
[13:08] Speaker 1: failing to qualify for the November runoff. But not everyone
[13:12] Speaker 1: was willing to accept that the prediction markets had been wrong.
[13:16] Speaker 2: You saw lots of right wing supporters of Pratt coming
[13:21] Speaker 2: forward and saying, this is ridiculous, the race was rigged,
[13:25] Speaker 2: like there must be an issue here.
[13:26] Speaker 1: Some of the posts came from influencers who were being
[13:29] Speaker 1: paid by Calshi and Polymarket to cite their arts. According
[13:33] Speaker 1: to Semaphore, after paid influencers for Calshi questioned the integrity
[13:37] Speaker 1: of the LA mayoral election on x CALG asked them
[13:41] Speaker 1: to take the post down.
[13:42] Speaker 2: So Trump wrote on truth Social this is an outrage. Essentially,
[13:46] Speaker 2: Spencer was robbed and he's kind of citing how well
[13:50] Speaker 2: he was doing in the race up until that point.
[13:52] Speaker 2: And so the concern is that ultimately these markets kind
[13:56] Speaker 2: of present themselves as truth machines, but they can also
[13:59] Speaker 2: be utilize by people that are looking to sort of
[14:02] Speaker 2: spread mistruths.
[14:04] Speaker 1: In the past year, the platforms have missed the mark
[14:07] Speaker 1: in high profile races like in Argentina, India, and New
[14:11] Speaker 1: York City, and one election in Romania showed they can
[14:15] Speaker 1: also be vulnerable to potential influence from bad actors.
[14:19] Speaker 2: There was an election in twenty twenty four that was
[14:21] Speaker 2: actually annulled because there was found to have been Russian
[14:24] Speaker 2: disinformation campaign that was enacted on TikTok and YouTube. It
[14:29] Speaker 2: was so bad that they ended up canceling the election.
[14:31] Speaker 2: They held another election, and that election was subject to
[14:34] Speaker 2: a lot of speculation around whether the polymarket numbers for
[14:40] Speaker 2: the kind of nationalist candidate who was supported by the
[14:43] Speaker 2: Kremlin were being manipulated because there was such a big
[14:47] Speaker 2: divergence between what the polls said and what the odds
[14:51] Speaker 2: were were on this political candidate, and there were lots
[14:53] Speaker 2: of people that kind of suggested that this was like
[14:56] Speaker 2: Russia cottoning on to the potential opportunity to mess with
[15:01] Speaker 2: people's perceptions of reality by targeting prediction markets alongside TikTok
[15:07] Speaker 2: or YouTube or other forms of disinformation.
[15:11] Speaker 1: Of disinformation campaign Yeah, a firm Bloomberg asked to view
[15:16] Speaker 1: the trades found that a few recently created accounts we're
[15:19] Speaker 1: driving a lot of the trading. Liam says, maybe that
[15:23] Speaker 1: was a coincidence, but for supporters of the right wing
[15:26] Speaker 1: candidate George Simeon, the bets helped create the expectation that
[15:31] Speaker 1: he'd win.
[15:32] Speaker 2: When this guy Simeon lost again, you had all these
[15:35] Speaker 2: people coming forward and saying, look, this is ridiculous on
[15:37] Speaker 2: product market. He was, you know, a dead sir, and
[15:40] Speaker 2: now you expect us to believe that he lost. So
[15:42] Speaker 2: it's almost like this damage either way, even if ultimately
[15:45] Speaker 2: it doesn't succeed, or even if ultimately it's not manipulation,
[15:48] Speaker 2: if everyone is suggesting it, then it kind of degrades
[15:52] Speaker 2: the sanctity and faith in the democratic process.
[15:56] Speaker 1: I'm wondering, given all this potential for manipulation and just
[16:00] Speaker 1: the rise of these prediction markets importance and reach, do
[16:04] Speaker 1: you see the risks of this becoming more severe moving forward.
[16:07] Speaker 1: Are we at an inflection point?
[16:09] Speaker 2: I think so. I think that essentially, you can have
[16:12] Speaker 2: a situation where something is quite cheap to move can
[16:17] Speaker 2: have a huge amount of traction, both in social media
[16:20] Speaker 2: and in the media, and so I do think that
[16:22] Speaker 2: the risks are increasing that people will try and manipulate
[16:26] Speaker 2: these markets for kind of political ends. Having said that,
[16:30] Speaker 2: I do think there's fairly easy solutions to some of
[16:35] Speaker 2: these issues. A lot of the issues around poly market
[16:38] Speaker 2: could be fairly easily remedied if it kind of emulated
[16:41] Speaker 2: its rival Calshie and started getting the identification of its users.
[16:47] Speaker 1: Another thing that could minimize the risk of manipulation would
[16:51] Speaker 1: actually be to increase the number of people placing bets,
[16:54] Speaker 1: boosting the trading volume on the platforms. More gambling carries
[16:59] Speaker 1: its own downside, but more heavily traded markets are harder
[17:03] Speaker 1: and more expensive to manipulate.
[17:05] Speaker 2: I think that if you talk to Calchian poly Market,
[17:07] Speaker 2: they kind of hope that one day everyone's going to
[17:10] Speaker 2: be sitting there watching television betting on everything, and everyone's
[17:13] Speaker 2: going to want to bet on the you know, local
[17:15] Speaker 2: mayor race in you know, in Sacramento. I'm not sure
[17:20] Speaker 2: that that's going to happen.
[17:23] Speaker 1: We are the midterms are coming up in the US.
[17:26] Speaker 1: We're going to be seeing a lot more polls, a
[17:28] Speaker 1: lot more prediction market election odds being reported. How should
[17:33] Speaker 1: the public and the media be reading these results?
[17:37] Speaker 2: What the media companies can do? And this is something
[17:39] Speaker 2: that a professor at Stanford called Andy Hall. He's done
[17:42] Speaker 2: a lot of work on and essentially he sort of
[17:45] Speaker 2: suggests a charter for the media where they only cite
[17:49] Speaker 2: markets of a certain size. They just do some basic
[17:53] Speaker 2: interrogation of like is this price being determined by one
[17:56] Speaker 2: or two wallets or is there a real divert number
[18:00] Speaker 2: of trades that have come to this price. He also suggests,
[18:05] Speaker 2: rather than taking like a price at a set moment
[18:08] Speaker 2: in time, which can be pushed around, that you actually
[18:10] Speaker 2: take some kind of average over the course of a day.
[18:13] Speaker 2: So it's different things that the media companies could do
[18:15] Speaker 2: fairly easily. They wouldn't really undermine their use of these things,
[18:18] Speaker 2: but would make the odds that they're citing more robust.
[18:28] Speaker 1: This is the Big Take from Bloomberg News. I'm Sarah
[18:30] Speaker 1: Holder to get more from the Big Take and unlimited
[18:33] Speaker 1: access to all of bloomberg dot Com. Subscribe today at
[18:37] Speaker 1: bloomberg dot com slash podcast offer. If you liked this episode,
[18:41] Speaker 1: make sure to subscribe and review The Big Take Wherever
[18:44] Speaker 1: you listen to podcasts. It helps people find the show.
[18:48] Speaker 1: Thanks for listening. We'll be back tomorrow

Shows are listed from Apple Podcasts' charts and directory, and Podcast Index where it is used. Every show and episode belongs to its publisher and plays from the publisher's own feed; this site hosts none of them.