Six Days Between a Billion-Dollar Bet and a Volume Bombshell

Paradigm led a $1 billion investment into Kalshi on December 2, 2025, with participation from Sequoia, Andreessen Horowitz, Meritech Capital, IVP, ARK Invest, Anthos Capital, CapitalG, and Y Combinator. The Series E valued Kalshi at $11 billion — double its valuation from just sixty days prior.

Six days after that round closed, Paradigm's data researcher Storm Slivkoff published findings that effectively cut in half the reported trading volume of Kalshi's chief rival, Polymarket. Polymarket's data team pushed back almost immediately, saying they had flagged the same "bug" to Paradigm two months earlier in a private conversation.

Where past cases of inflated gambling figures took investigators years of wiretaps and informants to unravel, this one took a single researcher armed with a transaction simulator and six days separating a venture round from a technical blog post.

01Two Companies, Two Playbooks

Prediction markets carrying Wall Street backing and NYSE partnerships appeared credible right up until their headline numbers stopped holding together.

Shayne Coplan founded Polymarket on the Gnosis Conditional Token Framework. Peter Thiel's Founders Fund backed its early rounds. Later, ICE — the parent company of the New York Stock Exchange, put $2 billion into the platform on October 7, 2025, valuing it at $9 billion post-money. A month afterward, on November 13, Coplan rang the NYSE opening bell. Around the same time, UFC and Zuffa Boxing signed on as official prediction-market partners, and ICE took on distribution of Polymarket's data feeds — casting the platform as market infrastructure rather than a betting site.

The pitch behind Polymarket, as with prediction markets generally, was decentralization and neutrality: a protocol that lets users bet against one another without itself taking a side. Its headline statistic was "$25 billion in lifetime trading volume" — a figure that, as of this writing, still reflects the pre-correction data described below and has not been updated at the source.

Kalshi built its story around regulation instead. The exchange received Designated Contract Market status from the CFTC on November 3, 2020 but didn't open to the public until July 2021 — meaning it operated as a regulated entity from day one. Its backers — Sequoia Capital, Andreessen Horowitz, Capital G, and ARK Invest — are the type of institutional money typically associated with compliance-first bets. Kalshi raised $300 million at a $5 billion valuation in October 2025, then Paradigm led its $1 billion Series E at an $11 billion valuation on December 2 — doubling in under sixty days. That same day, CNN announced a partnership with Kalshi to display its real-time probabilities during live broadcasts with ticker integration. Paradigm co-founder Matt Huang described the investment as follows: "We see this as an uncapped cultural and economic phenomenon, similar to how we felt about crypto a decade ago." Nine days later, on December 11, Kalshi joined a newly formed Prediction Market Coalition alongside Crypto.com and other operators — a coordinated lobbying effort ahead of anticipated regulatory battles.

Each company chose a different route toward the same claim of legitimacy. Polymarket went offshore after a $1.4 million CFTC fine in 2022, operated in crypto-native fashion without KYC, then bought a CFTC-licensed exchange, QCX, for $112 million in July 2025 to re-enter the US market. Kalshi instead spent years courting the CFTC directly, positioning itself as the compliant alternative.

Both companies reported comparable November 2025 volumes on paper: Polymarket at $3.74 billion in monthly spot volume, Kalshi at $5.8 billion. And both, it turned out, were quietly moving toward the same business model. Polymarket was reportedly preparing an internal market-making desk, while Kalshi already ran one: its affiliate, Kalshi Trading LLC, actively trades on the platform, a practice that has drawn a class-action lawsuit alleging the supposedly neutral exchange bets against its own users.

02The Double-Counting Problem

Polymarket's on-chain data, it turns out, systematically overstated trading volume — not through fabricated trades, but through a structural quirk in how its smart contracts log activity.

Every trade on the platform emits two separate OrderFilled events: one for the Maker, the party whose resting order gets filled, and one for the Taker, the party executing against it. These are two records of a single transaction, not two distinct trades. Third-party analytics platforms — Dune Analytics, DefiLlama, and Blockworks among them — had been summing both events, effectively counting every $100 trade as $200 of volume.

A single transaction illustrates the mechanism: 0xbf47fbf1bc113a7ec50a1103921265ba5d8fbe6dfb4d12a1c78c61c8fdb195bf. The actual trade moved $4.13 into YES tokens. The contract emitted two OrderFilled events of $4.13 each — one maker-side, one taker-side. Dashboards summed them and reported $8.26 in cashflow volume. The distortion applied to both notional volume (contracts traded) and cashflow volume (USD value at time of trade), across every trade, every market, every day.

The scale of the correction is significant. November 2025 volume, publicly reported at $3.7 billion, may actually have been closer to $1.9 billion according to Paradigm's analysis — a figure DefiLlama has since adopted. Lifetime volume, shown by dashboards at roughly $25 billion, appears closer to $13.5 billion in reality. Notably, Polymarket's own website had been displaying the correct, taker-only figures the whole time — the same methodology Kalshi uses internally. The discrepancy lived entirely in third-party tools, not in Polymarket's own reporting.

This wasn't a new observation. Blockworks' Dan Smith had raised the same question back in August 2024, asking why dashboards summed OrderFilled events without separating maker and taker sides. It circulated among crypto data analysts in private chats; Polymarket's team reportedly acknowledged the methodology issue at the time, and the conversation didn't go further. According to Polymarket's Primo Data account, Paradigm was part of those same chats and therefore aware of the issue months before publishing its own research. Meanwhile, public dashboards kept showing the inflated figures, and media coverage of Polymarket's fundraising continued to cite the "$25 billion" number.

What changed was that Storm Slivkoff built a transaction simulator modeling all eight of Polymarket's trade types, audited the contracts' event-emission logic, and examined data invariants across both the CTF Exchange and NegRisk contracts. His conclusion attributed the confusion to "interacting layers of complexity," where eight distinct trade types generate redundant event streams that standard block explorers are "not sufficient" to disentangle. He was also careful to note, in a follow-up post, that the investigation concerned only third-party dashboards, not Polymarket's own systems.

03A Question of Timing

The sequence of events raises an obvious question: was this diligence, or was it competitive positioning?

Paradigm's $1 billion investment on December 2 doubled Kalshi's valuation to $11 billion, with Matt Huang calling it "an uncapped cultural and economic phenomenon, similar to how we felt about crypto a decade ago." The same day, Kalshi announced its CNN tie-up. Around that same window, Polymarket was reportedly deep in its own fundraising talks — Bloomberg had reported back in October that the company was seeking $12 to $15 billion in a new round, with press coverage still citing the $25 billion volume figure.

Then, on December 8, 2025, Storm Slivkoff posted his findings on Twitter/X: "Found a pretty major data bug. it turns out almost every major dashboard has been double-counting Polymarket volume (not related to wash trading). this is because Polymarket's onchain data contains redundant representations of each trade." He linked to the full write-up on Paradigm's site — transaction simulator, contract audits, data-invariant analysis, describing it as six months of work published on a Sunday morning. Matt Huang amplified the post minutes later: "Polymarket data bug: volumes are double-counted in most public data. Interesting find in diligence from notnotstorm" — notably framing it as a diligence finding rather than a public-interest disclosure.

Within hours, Primo Data responded on behalf of Polymarket: "This is not how prediction markets report volume, including your portfolio company Kalshi. To be clear: Our site does not double count volume. We show notional taker volume (same as Kalshi). Dan Smith (Blockworks) asked this question back in October in a crypto data group chat we're both in, and you saw my response over 2 months ago." If accurate, this would mean Paradigm had known about the methodology issue since October, watched the inflated figures persist publicly for two months, then published six days after investing $1 billion in Polymarket's direct competitor — timed to coincide with Polymarket's own fundraising push. In the aftermath, DefiLlama, Allium Labs, and Blockworks confirmed they were updating their Polymarket dashboards to remove the double-counting after validating Slivkoff's findings.

Reaction in the data community was mixed. Will Sheehan of Parsec Finance called it "a bit like a hit piece when it's just data being hard and Polymarket's contracts being open/onchain." Nick Preszler of Melee Markets pointed to a related quirk: "If a user buys $10 worth of contracts at .1c each, they are risking $10, but get credited for $10,000 of volume because they have purchased 10,000 contracts." Dragonfly's Hildobby pushed back on the novelty of the finding: "Every half decent dashboard has had this accounted for since at the very least 2024. See dune tables with matched orders attached for example." Whatever its merits, the research landed just six days after the investment that gave Paradigm a direct financial stake in Polymarket's decline.

04Becoming the Counterparty

Both platforms marketed themselves as neutral venues connecting buyers and sellers — neither advertised that it was also building a desk to trade against those same users.

On December 4, 2025, Bloomberg reported that Polymarket was hiring for an internal market-making team that would trade against customers on its own exchange, with access to platform data such as order flow and user behavior. Harry Crane, a statistics professor at Rutgers, argued the move blurred the line between a prediction market and a sportsbook — and then some: "At a sportsbook it is well understood that the book is the counterparty, and will use whatever information it can to get the edge over its customers. Exchanges are supposed to be different."

Kalshi had already gone further. Its internal desk, Kalshi Trading, is already active on the platform, and in November 2025 it drew a nationwide class-action lawsuit alleging the desk sets betting lines that disadvantage customers who believed they were trading against other users, not a data-advantaged in-house market maker.

Harry Crane also pointed to NoVig, another prediction-market platform, as a cautionary precedent: once its internal desk ended up on the losing side of trades, the platform voided the winning bets rather than pay out.

The broader pattern of platforms quietly becoming market participants isn't new; a similar dynamic played out in 20th-century Las Vegas, where casinos with clean corporate ownership on paper were, in several documented cases, actually run by organized-crime figures behind the scenes. In one well-known example, Allen Glick owned four Las Vegas casinos through Argent Corporation while Frank "Lefty" Rosenthal managed operations and Tony "The Ant" Spilotro handled enforcement behind him. When the FBI eventually unwound the arrangement, Glick made a deal and avoided prison while the operators behind him were convicted.

05Regulation as a Competitive Moat

Regulatory status, rather than protecting users, increasingly functions as a purchasable or negotiable advantage in this market.

Kalshi's CFTC approval in 2020 to offer event contracts became its core competitive pitch: federally approved, while its rival operated offshore. That positioning gained further weight through media partnerships — the CNN deal announced December 2 and a similar integration with CNBC put Kalshi's real-time odds alongside stock tickers and economic data in mainstream financial coverage.

Those partnerships coincided with a dramatic market-share shift. In December 2024, Polymarket held roughly 95% of prediction-market volume; by December 2025, Kalshi held about 78%. Part of that shift, however, reflects distribution rather than organic adoption of a new platform: Kalshi's incoming CFO acknowledged on the company's Q3 earnings call that a large portion of its volume originates from Robinhood, which processed 2.5 billion prediction-market contracts in October alone. Webull added Kalshi integration in February, and PrizePicks brought Kalshi contracts to 38 states in November — regulatory approval opened doors to retail-app partnerships that an offshore platform couldn't access.

Crypto wallets pursued the opposite strategy: MetaMask integrated Polymarket into its mobile app in December, and Trust Wallet launched a Predictions hub planning to add both platforms, with a combined user base cited at 220 million.

Locked out of US retail partnerships by its 2022 CFTC enforcement action, Polymarket instead purchased its way to compliance: acquiring QCX, a CFTC-licensed exchange, for $112 million in July 2025, followed by a CFTC no-action letter in September 2025 that cleared a path back into the US market through a closed beta.

Beyond regulatory status, Polymarket also gained a corporate stamp of approval. ICE's $2 billion investment functioned as much as reputational transfer as capital. ICE CEO Jeffrey Sprecher framed it this way: "Our investment blends ICE, the owner of the New York Stock Exchange, which was founded in 1792, with a forward-thinking, revolutionary company pioneering change within the Decentralized Finance space." Media integrations reinforced the same effect: placing Kalshi's odds next to S&P 500 tickers on CNN and CNBC frames betting activity as a form of price discovery rather than gambling — a dynamic with its own historical antecedent, since, as one Kansas City organized-crime figure, Nick Civella, reportedly put it, political access requires cash, and that cash historically flowed out of Las Vegas casinos.

06Positioning for an Exit

One reading of these moves is that both companies — and their venture backers — are building toward an eventual sale to traditional financial institutions rather than a long-term standalone business.

There's precedent for this in Las Vegas history. Howard Hughes bought several Nevada casinos in the late 1960s for an estimated $300 million, and Nevada officials presented his ownership as a cleanup of the industry's image; his entry helped prompt the 1967 Corporate Gaming Act, which opened the door for publicly traded corporations to own casinos. Hughes reportedly lost money on most of these properties even as day-to-day operations changed little — but the change in ownership structure was what mattered to Wall Street. By the 1980s, Hilton and MGM had built major casino portfolios, and Hughes's estate ultimately profited from selling his properties to corporate buyers such as MGM.

The same logic could apply to prediction markets now. ICE's stake in Polymarket, per Sprecher's own framing, pairs a 232-year-old exchange lineage with a crypto-native platform — and Coplan's NYSE bell-ringing served as much as visibility for institutional audiences as ceremony. Kalshi's CNN and CNBC placements achieve something similar by putting its odds beside bond yields and currency data in mainstream financial coverage. Read against that backdrop, Matt Huang's "uncapped cultural and economic phenomenon" language could describe positioning for a future sale — potentially to sovereign wealth funds or pension allocators — rather than simply enthusiasm for the product category. Whether Polymarket's disputed volume figure lands at $25 billion or $13.5 billion may matter less, in that scenario, than what an eventual institutional buyer is willing to pay once the platforms have accumulated regulatory approval, media distribution, and demonstrated market-making revenue.

07A Structural Blind Spot

How did a discrepancy on the order of $11.5 billion — the gap between the $25 billion headline figure and the roughly $13.5 billion adjusted one — go unchallenged by the industry's own analytics infrastructure for so long?

Part of the answer is structural rather than deliberate. Blockchain analytics platforms typically review open-source dashboard code for whether it runs correctly, not necessarily whether its underlying logic measures the right thing. Once a dashboard is live, subsequent analysts tend to check their numbers against existing ones rather than against first principles — so agreement between dashboards hardened into an assumption of accuracy, even after Dan Smith flagged the underlying issue in August 2024. With volume figures rising and headlines favorable, there was little incentive to interrogate whether the inputs were being double-counted.

The episode is a reminder that these dashboards function as aggregators of self-reported on-chain logic rather than independent auditors — a distinction that matters when the same figures are cited in fundraising materials and press coverage.

Taken together, the sequence — a $1 billion investment in Kalshi, followed six days later by research that halved Polymarket's reported volume during its own fundraising window, set against both companies quietly building trading desks that compete with their own users while racing for regulatory cover — suggests that the competition between Kalshi and Polymarket is no longer only about accuracy or transparency. It is also about which platform ends up best positioned, on paper, for its next round or its eventual sale.