There’s a serious problem with the idea of betting trading on everything, and Kalshi is finding this out in real-time.
the company is grappling with several integrity concerns that raise questions about fairness and/or erode trust in what are being billed as “truth machines”:
Insider trading
Market settlement
Is it P2P
From STTP’s point of view, these are issues the industry needs to resolve, and the first step is scrapping the idea the idea of betting on everything, which opens Pandora’s box.
Insider Trading
On the insider trading front: The latest example, but far from the only one (the Pope, the Nobel Prize, a Google insider, Coinbase CEO, Israeli airstrikes, Nicholas Maduro, and Chris Hayes are just some of the others) was a hypothetical posed by CNBC to Kalshi CEO Tarek Mansour in what started off as a puff piece interview that quickly turned serious.
Despite recently announcing a zero-tolerance policy to insider trading and beefed up protocols, Mansour danced around questions from CNBC’s Andrew Ross Sorkin and Becky Quick about potential insider trading concerns:
Like everything with prediction markets, the debate seems to be around definitions of words, with everyone arguing from different starting points. But lost in the messy debate is the simple fact that markets susceptible to manipulation shouldn’t be offered.
As I wrote on Wednesday (see final sentence):
“Essentially, Mansour tried to draw a line between insider information (data within a company) and material non-public information, which are usually used interchangeably, with the only difference being the origin of the information.
“Mansour seems to be saying (Kalshi can correct me if I’m wrong) that the origin matters, and MNPI, as put forth in the hypothetical Bad Bunny dancer question, isn’t insider trading, which it technically isn’t. The issue is, all insider trading is MNPI, but not all MNPI is insider trading.
“MNPI is prohibited by the CFTC, under its anti-fraud and anti-manipulation rules, which prohibits trading on MNPI in breach of a pre-existing duty, such as a confidentiality agreement, employment contract, or nondisclosure obligation owed to the production team, artist, or event organizers.
“What Mansour is claiming — that someone with non-public information isn’t technically insider trading — is correct, but they are violating the rules that insider trading falls under.
“Further, the Commodity Futures Trading Commission (CFTC) has rules that effectively prohibit the listing of markets that are readily susceptible to manipulation.”
In a nutshell, Insider information and material non-public information covered by NDAs and other agreements/obligations are prohibited, but so is offering a market that is susceptible to manipulation through material non-public information that isn’t prohibited.
Then there is the, “insider trading is good, actually” crowd.
As I wrote in January in the wake of the Maduro story:
The existence of insider trading at prediction markets is somehow a controversial topic, and it looks like things are coming to a head. Major outlets like The Wall Street Journal, NPR, Slate, and Politico are reporting on what looks to be insider trading, this time from Polymarket’s ‘will Nicholas Maduro be captured before the end of January’ market.
As NPR reported: “On Polymarket, a popular site for making bets on the outcome of real world events, a user wagered $32,000 that Maduro would be toppled by the end of January hours before Trump ordered the operation. When it was clear the U.S. had captured Maduro, the trader made more than $400,000 in profit.”
There is also the memification of prediction markets, as the market on ‘will Jesus return before 2027’ has seen a recent surge, and we now know why:
This 2024 piece is a solid explainer on the issue with these secondary markets:
Upshot #1
If the public thinks the markets are rigged, prediction markets will struggle to sustain mainstream acceptance as credible “truth machines,” no matter how often they repeat their practiced talking points.
Repeated instances of apparent insider trading will fuel skepticism that these platforms are fair for everyday users. Without robust safeguards against manipulation (starting with not offering easily manipulated markets), even if it’s not “technically” insider trading, the industry risks being dismissed as just another speculative gimmick.
Market Settlement
On the market settlement front: We have several examples from the recent Super Bowl to point at, including Jason Kelce and Mark Wahlberg, but most notably, Cardi B’s, ‘did she or didn’t she perform during the halftime show’ (full story here).
The market turned into a fiasco that highlights how even seemingly straightforward event contracts can devolve into chaos when market settlement rules aren’t clearly defined.
Again, everything with prediction markets boils down to definitions— what counts as “performing”? Singing? Dancing? Mouthing lyrics? And you better be paying close attention to the resolution rules.
As I noted on Wednesday, Kalshi eventually resolved the market, noting on its website that, “Due to ambiguity over whether or not Cardi B’s attendance at the 2026 Super Bowl halftime show constituted a qualifying ‘performance’, Kalshi is invoking Rule 6.3(c) to settle this market to the last traded price before trading was paused. Those prices are $0.26 for Yes holders and $0.74 for No holders.”
FWIW, Polymarket paid out “yes” bettors at $1 and “no” bettors at $0.
The CFTC’s anti-manipulation rules are supposed to prevent this, but when platforms have sole discretion in gray areas, things get really hazy, really fast.
And then there are the straight up errors:
Upshot #2
Ambiguous market settlement rules turn what should be objective “yes/no” outcomes into subjective disputes. Unclear definitions invite chaos and accusations of bias (as I’ll mention in the next section), especially when platforms invoke discretionary rules to resolve trades.
This not only violates the spirit of CFTC anti-manipulation guidelines but also undermines user confidence in fair payouts. event resolution must be clear and transparent, or prediction markets risk devolving into glorified three-card Monte games rather than reliable forecasting tools.
Peer-to-Peer
Kalshi bills itself as a peer-to-peer platform, but if you look under the hood, it starts to resemble a traditional sportsbook. And once again, the debate involves definitions, with Kalshi saying its market makers (institutional or individual) are simply peers. But what makes it P2P? No house edge? User-vs-user matching?
Here’s why people aren’t necessarily buying Kalshi and other prediction markets’ claims of being peer-to-peer, which come with a lot of technicalities that may erode trust when laid on the table.
Margin Trading
Reports indicate Kalshi and other prediction markets are seeking CFTC approval for margin trading, which would let big players (mostly institutional market makers) bet without fully collateralizing positions, blurring the line between peer-to-peer trading and something closer to a leveraged bookmaker.
As Joe Brennan Jr. put it:
A side note to this, as InGame reported, Kalshi is also ready to offer sportsbooks rebates if they lay-off risk at Kalshi and meet certain thresholds:
“The rebate program will be available to “entities that currently offer sportsbook services” who purchase at least 300,000 event contracts for hedging purposes in a single order. An order of 300,000 event contracts will always pay out $300,000 in combined stakes and winnings if they win.
“The rule filing says that the rebate will apply to both taker fees and request-for-quote (RFQ) fees, which are primarily used for parlays. However, the minimum order size may limit the ability to hedge all parlays.”
Institutional Market Makers
Let’s start with Kalshi Trading, a separate company that provides liquidity and acts as a market maker, buying and selling contracts to keep markets on Kalshi Exchange fluid. From the Kalshi website:
Kalshi Trading is one such institutional market maker, and critics argue this muddies the “peer-to-peer” purity, especially with powerhouse firms like Susquehanna International Group onboard as market makers.
And then there is the recent news that Jump Trading has acquired small equity stakes in Kalshi and Polymarket in exchange for providing institutional-grade liquidity. Jump Trading’s involvement — a firm with deep Wall Street roots and experience in equities, futures, and crypto — signals that these platforms are maturing into a legitimate asset class, attracting traditional finance players. I don’t believe retail traders would call a market peer-to-peer when Kalshi Trading, Susquehanna, and Jump Trading are on the other side of the market betting, as Joe Brennan Jr. would say, '“on credit.”
Upshot #3
As Kalshi and others court Wall Street, they may be losing the trust of retail bettors who see trading against powerhouse financial institutions who are potentially trading on margin and receiving rebates as a bit one-sided.
“At its core, prediction markets appeal to our ego: It’s a gamified way to put your money where your mouth is. But here’s the rub: most people aren’t actually good at it, and peer-to-peer versions of this are known for rapid churn, as casual users overestimate their edge, leading to quick losses…
“Retail customers will be chewed up and spit out, so you will need a never-ending supply.
“Thus far, betting exchanges haven’t solved the P2P puzzle, and the current argument is, ‘this time it’s different.’ We shall see.
“Historical Example: Following Chris Moneymaker’s 2003 World Series of Poker win, millions of casual players flocked to sites like PartyPoker and PokerStars, and hundreds of online poker sites were created overnight. Some were sold early for staggering amounts, and many just flat failed, as most poker players overestimated their skills, losing money quickly to professionals, resulting in massive churn.”
Excellent post. An additional problem with settlement ambiguity is that it prevents cross-market arbitrage, so information doesn't flow as easily from one market to another.
I did this briefly for a living as part of my job. Specifying, not paid but Metaculus "fake internet points" markets. We didn't have any answer to the problem either. I feel like specifying and managing prediction contracts is the most sorely missed operational competency at most all of these places. UMA, etc are a fiasco.