The $117 Million Lesson

For years, the sports-betting industry’s open secret has been the parlay. Books love them because the math works overwhelmingly in the house’s favor. Bettors love them because the payouts look like jackpots. The result is a product where the casino’s hold rate balloons from the 4–5% range on a standard moneyline to 25%, 30%, sometimes north of 40% on a four- or five-leg ticket. It is the single most profitable product in the sportsbook ecosystem — more than half of sportsbook operator gross revenue flows from parlays — and it is profitable for the same reason scratch tickets are profitable. The hope, when prediction markets like Kalshi entered the sports space, was that exchange-style pricing would change this dynamic. No house. No vig baked into the line. Peer-to-peer matching that, in theory, should drive the cost of action toward zero. And on straight contracts, that promise has largely held. Parlays, however, are a different animal — and the numbers now make that brutally clear.

The Data

According to data stored for public access by Kalshi’s blockchain partner Dune, retail bettors on Kalshi’s app and website lost more than $117 million placing parlay bets between January 1 and April 30 of this year, with at least $35 million of that flowing directly to the exchange operator as estimated fees. On April 26 alone, parlay builders dropped $4.3 million. Overall, bettors risked roughly $800 million constructing custom multi-leg wagers over those four months. Sportico

The headline figure is the implied hold: for every $100 wagered, bettors lost about $15. That is not exchange pricing. A 15% effective take rate on a parlay product is squarely in the range of what traditional books extract on the same kind of wager. For context, the exact sportsbook parlay hold varies by month, often exceeding 18% in the U.S., and jumping well above 20% for single-game and live parlays. Kalshi’s number is lower — but the gap is narrower than the platform’s peer-to-peer branding implies, and the structural reason it could be that high at all lies in how Kalshi actually builds the parlay product.

The RFQ Problem: Why “Exchange” Is the Wrong Word for Parlays

The exchange model that makes Kalshi competitive on single-event contracts does not extend to parlays. Kalshi pioneered the use of a Request for Quote (RFQ) market structure to facilitate custom parlays. With RFQs, historically used as a financial tool for institutions to privately buy and sell large-scale positions to manage risk, it is impossible for people to offer parlay odds from the retail app as they can for single-event wagers. In-app users can only request take-it-or-leave-it odds set externally, often by large financial entities and professional bettors.

The party setting those odds on the other side is frequently not another sports fan with an opinion. Kalshi contracts institutional market makers — with Pennsylvania-based Susquehanna International, a trading and technology firm, often taking the “No” position on these wagers. The fee arrangements these makers operate under are not visible to retail users: institutional market makers often sign private deals with Kalshi that give them fee rebates in exchange for providing massive liquidity to the platform.

The practical consequence is that the “taker” in Dune’s data is almost always the retail customer, and this is by design. Retail bettors are, by design, always the ones accepting the terms. The professionals set the price. Research from University College Dublin found that retail takers on Kalshi’s RFQ combo system lose more often than not, with the institutional market makers on the other side earning money consistently. This is not peer-to-peer wagering. It is retail bettors requesting odds from firms with better models, faster data, and more capital — firms that, unlike a traditional sportsbook, face no regulatory pressure to display a posted line.

The Growth Trajectory: Engineered, Not Organic

Parlays did not naturally find their audience on Kalshi. They were cultivated. When parlays were quietly introduced as a beta in September 2025, they represented less than 3% of Kalshi’s total exchange volume. By April 2026, that figure had climbed to about 22%. The growth tracks how aggressively Kalshi has pushed the product.

Over the past month, the company has populated pages with suggested multi-leg wagers, borrowing a page from how traditional sportsbooks nudge bettors toward higher-margin products. One featured parlay for a recent Timberwolves-Spurs playoff game had 18 legs and offered the faint possibility of turning $1 into $185. A few thousand bettors combined to lose $48,000 on that single suggestion.

This playbook is familiar to anyone who has watched DraftKings or FanDuel evolve. You build the exchange product to earn credibility with sophisticated users, then monetize the mass-market user base with the highest-margin product you have. The difference here is that Kalshi does it while maintaining the rhetorical posture of a financial exchange. Rather than calling itself gambling, Kalshi refers to prediction markets as a new financial asset class. “People can make money on what they know, actually monetize their knowledge, monetize their hobby, because everyone is an expert on something,” Kalshi co-founder and COO Luana Lopes Lara told CBS News. That framing is defensible for straight contracts on liquid markets. It collapses entirely when applied to an 18-leg parlay priced by Susquehanna.

The Data Suppression Sequel

What happened after the Sportico report is itself instructive. Dune, a company that provides tools to track prediction market betting activity, removed two Kalshi datasets from public view on May 15 — less than 48 hours after Sportico used them to analyze how much retail bettors were losing on parlays. Dune later claimed the decision was purely its own and had been planned in advance, not made in response to the reporting. Regardless of the cause, the effect is notable: Dune said it restored the once-free datasets — but only for its highest-paying customers, limiting the datasets to its top-tier Enterprise plan costing roughly $40,000 annually for most users. The two Kalshi datasets that remain free do not reveal user losses.

Research infrastructure that enables independent analysis of whether a retail product is harming its users should not be behind a $40,000 paywall. The timing of the restriction, whatever its cause, means the kind of analysis Sportico performed in May will be significantly harder to replicate going forward. That matters for bettors trying to assess the product’s true cost.

The Contagion: Robinhood, Polymarket, DraftKings

The RFQ parlay model Kalshi pioneered is now proliferating across the prediction market ecosystem. Robinhood will broker its multi-event bets using the Request for Quote (RFQ) system on Kalshi’s exchange, according to Robinhood’s senior director of futures and prediction markets. Robinhood hopes the high engagement level for parlays seen at traditional sportsbooks carries over to its exchange format. Polymarket has filed with the CFTC to introduce its own version, which it refers to as “Combinatoric Athletic Outcome Contracts.” DraftKings also launched parlays on its prediction market platform earlier this month.

The incentive structure here is self-reinforcing. Parlay products generate consistent, predictable revenue from retail users. They are easy to market. Platforms that do not offer them cede engagement to platforms that do. Once Kalshi demonstrated both the technical feasibility and the financial upside, the rest of the market followed. The only people for whom this development is unambiguously bad are the retail bettors at the end of the RFQ chain.

What Serious Bettors Should Take From This

The lesson has not changed, but the Kalshi data gives it new quantitative grounding. Single-market mispricing remains the only legitimate source of edge in sports betting. Parlays — on any platform, through any mechanism — multiply cost rather than multiply opportunity. The compounding of vig across legs is mathematically identical whether the vig is explicit (a traditional sportsbook’s juice) or implicit (an institutional market maker’s spread embedded in a take-it-or-leave-it RFQ price).

The structural advantage of Kalshi combos over sportsbooks — no compounding house vig — is real, but it does not automatically translate into a trader advantage. You have eliminated one source of edge erosion but introduced another: an institutional counterparty who may price the combo more accurately than you can. The net effect, as the $117 million figure demonstrates, is that retail parlay bettors still lose at a rate indistinguishable from what they lose at traditional books.

The venue is new. The math is not. And now, the data has been priced out of reach.