Cointelegraph is committed to providing independent, high-quality journalism across the crypto, blockchain, AI, and fintech industries.

All news, reviews, and analyses are produced with full journalistic independence and integrity. For more details on our standards and processes, please read our Editorial Policy.

“We view this as the first of several rulemakings the SEC will undertake to provide regulatory certainty for crypto assets after the Senate failed before the August recess to advance the Clarity Act on crypto market structure,” TD Cowen analyst Jaret Seiberg wrote in a client note sent after the SEC’s notice.

The proposal is expected to give a path to crypto firms enabling them to raise capital for projects without triggering SEC registration requirements, and the businesses are also expected to be provided an exit path for getting clear of the agency’s jurisdiction when they’re not engaged in hands-on management of the projects anymore.

Before this, Atkins and the agency had rolled through a lengthy series of crypto policy statements meant to clarify its regulatory position on digital assets, but the staff statements have little long-term durability. A formal rulemaking would be more difficult to reverse in the future.

But the rule will likely take further months to develop and finalize. This first stage will come with a comment period — typically two or three months — and be followed by a potentially lengthy rewrite.

Reg Crypto would join some of the other significant steps the agency has taken or is still working on to foster the U.S. crypto industry. One of the major moves was a joint stance with the Commodity Futures Trading Commission on a “taxonomy” to define how they view various crypto assets and which jurisdictions they belong in. The agency is also still working on its tokenized securities approach, which Atkins routinely mentions as one of the SEC’s marquee crypto maneuvers.

Trump Media said it plans to revamp its digital asset treasury strategy after unrealized losses on crypto and securities helped push the company to a $238 million net loss in the second quarter.

The company reported $190.4 millio n in unrealized losses across its digital assets, pledged digital assets and equity securities in its Q2 earnings release on Monday.

Trump Media said the new framework is intended to preserve its long-term digital asset exposure while managing volatility and improving the productivity of its balance sheet.

Trump Media is the publicly traded company behind Truth Social, Truth+ and financial services brand Truth.Fi. The company is tied to US President Donald Trump, who is the sole beneficiary of a trust that held about 41.1% of Trump Media’s voting power as of Feb. 25, according to its latest annual report.

Related: Trump Media sells Wall Street low-latency access to Trump posts

Its Q2 filing shows the company is already using options to manage Bitcoin volatility and generate premium income, while deploying some BTC through lending and other yield-generating arrangements. 

The company also said it plans to direct more resources toward Truth Social, Truth+ and other parts of its media business as part of a broader shift in how it allocates capital.

Trump Media boosts Bitcoin holdings after Q2

Trump Media’s Bitcoin holdings were little changed during the second quarter before the company stepped up its direct Bitcoin exposure in July.

As of June 30, Trump Media held 9,477.16 Bitcoin, down from 9,542.16 BTC at the end of the previous quarter.

Separately, the company had pledged 2,077.34 BTC as collateral for its options strategy. Of its reported holdings, 4,260.73 BTC was serving as collateral for convertible notes. 

Related: Strategy turns 1,690 BTC into $108.6M STRC buyback

In July, the company sold Bitcoin-related securities worth $159.6 million and used the proceeds to purchase Bitcoin.

By July 31, Trump Media reported holding approximately 14,139 BTC, including pledged Bitcoin, worth about $890.5 million at the time.

Trump Media flags risks from Bitcoin yield strategy

Trump Media also warned that its efforts to earn additional income from its Bitcoin carry counterparty credit risk and the potential loss of its assets.

The company said it has deployed a portion of its Bitcoin holdings to third parties through lending, placement and other yield-generating arrangements, which it described as relatively new strategies.

Some of those counterparties may not be rated by major credit rating agencies and could default during market downturns, liquidity crises or other financial distress.

If an arrangement is unsecured, the company said it may be unable to recover its Bitcoin if a counterparty becomes insolvent. Trump Media is also limited in its ability to sell or pledge Bitcoin while it is deployed, while counterparties may use those assets at their discretion. 

Magazine: Bitcoin will never fall below $60K again: Nansen founder

Trump Media says it will overhaul how it manages its digital-asset treasury after crypto and securities losses contributed to a $238 million net loss in the second quarter. In its Q2 earnings materials, the company attributed $190.4 million in unrealized losses to its mix of digital assets, pledged digital assets, and equity securities, while describing a plan aimed at keeping long-term Bitcoin exposure but reducing balance-sheet volatility.

The publicly traded company—best known as the parent of social platforms Truth Social and Truth+ and the financial services brand Truth.Fi—linked the strategy shift to the need for a more resilient framework. The company noted that the changes are meant to improve the “productivity” of its balance sheet without abandoning its core digital-asset positioning.

Key takeaways

• Trump Media reported $238 million net loss in Q2, with $190.4 million tied to unrealized losses across digital assets, pledged digital assets, and equity securities.
• A new treasury framework is planned to preserve long-term Bitcoin exposure while managing volatility and improving capital efficiency.
• Bitcoin use in hedging and yield activity is already in place, including options-based volatility management and deployments of some BTC to third parties.
• The company increased direct Bitcoin exposure in July, moving from 9,477.16 BTC at quarter-end to about 14,139 BTC by July 31, including pledged BTC.
• Counterparty and liquidity risks are explicitly flagged for Bitcoin-yield strategies, including default risk and limits on selling or pledging deployed BTC.

Why Trump Media is changing its digital-asset plan

Trump Media’s shift comes as investors focus on how publicly traded firms balance crypto exposure with the accounting swings that unrealized losses can create. In its second-quarter reporting, the company said its existing digital-asset and securities positions generated significant unrealized markdowns. Those losses, it said, were part of what drove the quarter’s large net loss figure.

Rather than retreating from Bitcoin, Trump Media emphasized that the revamp is intended to “preserve” long-term exposure while addressing volatility and making the balance sheet work more efficiently. The company also said it plans to direct more resources toward Truth Social, Truth+, and other media operations, framing the treasury shift as part of a broader capital allocation change.

Because the group is tied to former U.S. President Donald Trump, the broader context matters for market watchers. The filing notes that a trust holding roughly 41.1% of Trump Media’s voting power as of Feb. 25 remains the sole beneficiary of Trump Media voting power, according to the company’s latest annual report.

What the Q2 filing says about Bitcoin strategy

Trump Media’s Q2 documentation indicates it is not treating Bitcoin purely as a long-term spot holding. Instead, the company described a framework that already includes options to manage Bitcoin volatility and generate premium income. It also reported using part of its BTC in lending and other yield-style arrangements.

As of June 30, Trump Media held 9,477.16 Bitcoin, down slightly from 9,542.16 BTC at the end of the prior quarter. Separately, it reported pledging 2,077.34 BTC as collateral for its options strategy. The company also said 4,260.73 BTC was serving as collateral for convertible notes.

That structure shows a balancing act: maintaining Bitcoin exposure while ring-fencing assets for derivatives and financing obligations. It also highlights how pledged collateral can constrain a company’s flexibility during drawdowns or liquidity events.

July: Bitcoin-related sales followed by increased BTC exposure

While the second quarter itself left Trump Media’s direct Bitcoin holding relatively stable, the company later stepped up its Bitcoin exposure in July. By July 31, Trump Media said it held approximately 14,139 BTC, including pledged Bitcoin, which it valued at about $890.5 million at the time of reporting.

The path to that increase was tied to an intermediate step: the company said it sold Bitcoin-related securities worth $159.6 million in July and used the proceeds to purchase Bitcoin. This matters because it suggests the company viewed those securities as a temporary component in its capital deployment rather than a permanent replacement for direct BTC exposure.

For readers tracking how non-traditional crypto entrants manage treasury assets, the key takeaway is that Trump Media’s exposure management appears active rather than passive. The company is also maintaining a portfolio where some Bitcoin remains tied up—through pledges and other arrangements—while the headline BTC totals can rise through incremental purchases.

Risks Trump Media says it faces with BTC yield activities

Trump Media’s filings do not just outline how it earns additional income; they also provide a clear warning about the trade-offs. The company stated it deployed a portion of its Bitcoin holdings to third parties via lending, placement, and other yield-generating arrangements, describing these as relatively new strategies.

According to the company, some counterparties may not be rated by major credit rating agencies. That increases the risk that counterparties could default during periods such as market downturns, liquidity crises, or other financial stress.

Trump Media also warned that if an arrangement is unsecured, it may be unable to recover its Bitcoin in the event a counterparty becomes insolvent. It added that its ability to sell or pledge Bitcoin can be limited while assets are deployed, and that counterparties may use the assets at their discretion.

These disclosures are especially relevant when paired with the company’s decision to revamp its treasury strategy. The new framework is positioned as a way to maintain long-term exposure and reduce volatility, but the filings indicate the risk is not only market-driven. It is also operational and credit-driven—tied to whether deployed Bitcoin is recoverable and how counterparties behave under stress.

In other words, the company is trying to enhance balance-sheet performance while accepting that yield-style BTC deployments can introduce new failure modes that typical spot holding does not.

What investors should watch next

Trump Media has flagged both accounting volatility from unrealized losses and credit/liquidity risk from its Bitcoin-yield counterparties. Going forward, investors will likely focus on how the company implements its revamped treasury framework—particularly whether it changes the share of Bitcoin deployed to third parties versus retained as pledged collateral or held directly, and how those choices affect reported results in subsequent quarters.

Bitcoin’s BTC$64,080.96 elite holders are quietly loading up again.

The number of wallets holding at least 10,000 BTC has climbed back to 90, a six-month high, according to analytics firm Santiment. Over the past eight weeks alone, the count of these “whale” wallets has risen by six, a 7.1% increase.

The move builds on a broader accumulation trend first flagged four days ago. Since July 29, wallets in the 10–10,000 BTC range (whales and sharks) have accumulated BTC worth $1.5 billion. Santiment noted at the time that the pattern of larger players accumulating while smaller holders sell raised the odds of a move above $70,000 versus a drop below $60,000.

Meanwhile, “micro” wallets have been steadily shrinking throughout August. Santiment links this divergence to two recent catalysts of uncertainty and doubt: The Coldcard hardware-wallet exploit, which drained roughly $120 million worth of bitcoin, and continued delays to the U.S. Clarity Act, the long-awaited crypto market-structure bill that the Senate has now pushed to September.

These trends point to a classic supply rotation, with coins shifting from smaller holders into the wallets of the largest investors, the so-called “strong hands.”

Trump Media and Technology Group terminated its planned Cronos treasury venture with Crypto.com and Yorkville on August 7. The retreat signals that the corporate crypto treasury thesis, built on the premise that holding tokens is a balance sheet strategy, is running into the same saturation problem it was supposed to solve.

On August 7, 2026, Trump Media interim CEO Kevin McGurn told Axios that the company had terminated the CRO Strategy venture, the services agreement, and the digital asset product suite that accompanied the original deal. The companies also abandoned a related prediction market integration. Yorkville Acquisition Corp, the blank check vehicle created to take the treasury venture public, agreed to the termination. McGurn cited saturation. The treasury sector, he said, had become crowded.

The deal between Trump Media and Crypto.com was announced with the kind of numbers that attract headlines. A $6.42 billion treasury play. A branded entity called Trump Media Group CRO Strategy. A SPAC vehicle to take it public. The premise was simple: acquire CRO tokens at scale, hold them on a public balance sheet, and let market appreciation create shareholder value. The strategy echoed what MicroStrategy had done with Bitcoin, but applied to a token that was orders of magnitude smaller in market capitalization, less liquid, and tied to a single exchange ecosystem. The termination, coming less than a year after the announcement, is not just a deal falling apart. It is the first major public admission that the corporate crypto treasury model may have structural limits that its promoters did not advertise.

What the CRO Strategy deal was supposed to build

The original announcement in August 2025 outlined a multi-layered financial structure. Trump Media would purchase approximately $105 million in CRO tokens, acquiring 684.4 million tokens at a price of roughly $0.153 per token. Crypto.com would purchase $50 million in Trump Media shares, creating a reciprocal ownership link between the media company and the exchange. The two parties would then form Trump Media Group CRO Strategy, a dedicated entity designed to function as a publicly traded digital asset treasury.

Yorkville Acquisition Corp, a blank check company, would serve as the public listing vehicle. The SEC filing described the venture as the first and largest publicly traded CRO treasury firm, a designation that was accurate only because no one else had attempted the same structure with that specific token.

The deal also included a services agreement covering digital asset product development and a prediction market integration called Truth Predict, which would embed betting features inside Truth Social using Crypto.com Derivatives North America as the infrastructure provider.

The combined structure tied Trump Media to CRO price performance in multiple ways: through direct token holdings, through the treasury entity, through the SPAC listing, and through the product integrations that assumed continued engagement with the Cronos blockchain ecosystem. The multi-layered exposure was presented as strategic depth at the time of the announcement. In retrospect, it created a situation where CRO price weakness propagated through every component of the deal simultaneously, amplifying the downside in a way that a simpler structure would not have.

The valuation environment that produced the deal was also unusual. At the time of the announcement, corporate crypto treasuries were attracting premium multiples from investors who treated token accumulation as a growth strategy. Trump Media’s management appears to have structured the CRO Strategy venture to capture that premium. When the premium disappeared and the valuation environment shifted, the economic rationale for the deal evaporated alongside it.

Why the parties walked away

McGurn framed the termination around focus and saturation rather than failure. “We wanted to get focused,” Axios reported him saying. He characterized the crypto treasury sector as overcrowded, arguing that the competitive dynamics had shifted since the deal was announced.

The saturation claim has some factual basis. Between the second half of 2025 and mid 2026, more than 30 public companies announced crypto treasury strategies of various sizes. Most were modeled on the MicroStrategy template: issue equity or convertible debt, buy Bitcoin, and report the holdings as a core balance sheet item. The Trump Media venture stood apart because it targeted CRO rather than Bitcoin, but the underlying logic was the same.

What McGurn did not say is arguably more revealing than what he did. He did not claim the deal was restructured or paused. He confirmed mutual termination, meaning all three parties agreed that the venture no longer served their interests. The services agreement was also terminated. The prediction market integration was downgraded from a full product to a marketing arrangement.

McGurn added that staking CRO had become less central for Crypto.com itself, making a split logical for both sides. That detail matters. If the exchange that issues a token is de-emphasizing it, the case for a third party to build a treasury around it weakens considerably.

The $406 million loss that changed the math

The termination did not occur in a vacuum. Trump Media reported a $406 million quarterly loss earlier in 2026, driven primarily by markdowns on its digital asset holdings. The loss was not a realized trading loss in the traditional sense. It reflected the accounting treatment of crypto assets under fair value rules, where price declines flow directly through the income statement.

For a company with limited operating revenue, a nine figure markdown on token holdings is not a rounding error. It is a thesis test. The crypto treasury model assumes that holding tokens creates long term value for shareholders. But the same model forces the company to report price declines as losses, creating quarterly volatility that public market investors tend to punish.

MicroStrategy, the original corporate Bitcoin treasury, has navigated this problem by leaning into it. The company rebranded as Strategy, made Bitcoin accumulation its primary corporate identity, and attracted a shareholder base that understood and accepted the volatility. Trump Media was not positioned to make the same bet. Its core business is a social media platform. Its shareholders purchased the stock for reasons that included but were not limited to the crypto treasury thesis. When the token holdings produced a loss that exceeded the company’s operating revenue by orders of magnitude, the strategic case for continuing became harder to make.

The regulatory environment added complexity. While McGurn explicitly denied that regulatory pressure drove the termination, the broader landscape has shifted. The GENIUS Act and related legislative efforts have introduced new compliance requirements for companies holding digital assets at scale. Whether those requirements directly affected the CRO Strategy venture or simply increased the cost of maintaining it is an open question that the termination announcement did not address.

The accounting treatment itself deserves closer examination. Under ASC 820, crypto assets held at fair value require quarterly mark to market adjustments. When token prices rise, the company reports unrealized gains that flow through its income statement, inflating earnings in ways that may attract shareholders who mistake accounting gains for operating performance. When prices fall, the reverse occurs, and the company must explain why its core business lost hundreds of millions of dollars in a quarter where nothing operationally changed. The asymmetry between the two scenarios creates a ratchet effect: the company captures enthusiasm during price increases but faces existential questions during declines. For a company like Trump Media, whose operating revenue from Truth Social was in the low tens of millions annually, a $406 million markdown made the treasury strategy the dominant item on the income statement, overshadowing every other aspect of the business.

Truth Predict and the prediction market retreat

The CRO Strategy termination was not the only product that got scaled back. Truth Predict, the prediction market feature that Trump Media unveiled in October 2025, was also restructured.

The original plan embedded betting features inside Truth Social, powered by Crypto.com Derivatives North America. Users would trade prediction market contracts without leaving the social media platform. The integration assumed that Truth Social’s user base represented a captive audience for event contracts on politics, sports, and financial outcomes.

Under the new arrangement, the companies will pursue a marketing partnership rather than a full product integration. Crypto.com will promote its prediction products to Truth Social users, but Trump Media will not operate the back end infrastructure. McGurn argued that the space was already crowded with prediction market operators and that running infrastructure offered little return compared to simply distributing someone else’s product.

The reasoning reveals a broader shift in how Trump Media sees itself. The company is moving away from operating crypto infrastructure and toward licensing its brand and audience as distribution assets. That pivot is visible in its API business, which now serves roughly 10 customers (up from about five), most of whom are high frequency trading firms that feed Truth Social data into algorithmic strategies. McGurn said the company is also courting large language model developers and prediction market platforms as API customers.

The transition from infrastructure operator to data licensor is a meaningful strategic shift. It reduces Trump Media’s exposure to the operational risks of running crypto products while creating revenue streams that do not depend on token price performance. Whether the data licensing business can generate enough revenue to replace the value that was supposed to come from the treasury venture remains to be seen.

The prediction market landscape that existed when Truth Predict was conceived looked different from the one that emerged by mid 2026. At the time of announcement, Polymarket was the dominant player and the only U.S. focused platform with meaningful liquidity. By August 2026, Kalshi had expanded into event contracts for political and financial outcomes, Robinhood had added prediction market features to its mobile app, and several crypto native platforms had launched competing offerings. The window for Truth Social to capture a meaningful share of the prediction market audience had narrowed considerably, and the cost of operating back end infrastructure for a feature that would compete with well funded competitors no longer justified the investment when a simpler marketing arrangement could provide the same user exposure at a fraction of the cost.

The API pivot also reframes Truth Social’s value proposition. Under the treasury model, Truth Social was a distribution channel for crypto products. Under the data licensing model, it is a source of sentiment signals that have value to financial firms. The platform’s concentrated user base, which skews toward politically engaged American adults, generates text and engagement data that reflects a demographic segment that is underrepresented in other social media sentiment feeds. High frequency traders and LLM developers are willing to pay for access to that signal precisely because it is different from what Twitter, Reddit, or Stocktwits provides. The commercial value of the platform may ultimately have less to do with how many users it retains than with how unique the data those users generate is.

What happens to Yorkville and the SPAC structure

Yorkville Acquisition Corp agreed to the termination alongside the other parties. The blank check vehicle was created specifically to take the CRO treasury venture public. Without that venture, the SPAC’s original purpose no longer exists.

However, one piece of the Yorkville structure survives. Yorkville America’s America First ETFs, branded as Truth Social Funds, will continue operating. These funds existed as a separate product line from the SPAC and were not dependent on the CRO Strategy venture for their investment thesis.

The survival of the ETF products while the SPAC terminates illustrates the fragmented nature of the original deal. What was presented as a unified strategic partnership between Trump Media, Crypto.com, and Yorkville was in practice a collection of loosely connected agreements. The treasury venture, the services agreement, the prediction market integration, and the ETF products could be separated because they were never truly integrated at the operational level.

For the broader SPAC market, the termination adds another data point to an already difficult environment. Blank check companies that target crypto related ventures have faced elevated scrutiny from the SEC, and the completion rate for crypto focused SPACs has declined steadily since 2024. The Yorkville termination does not set a legal precedent, but it reinforces the pattern of crypto SPAC deals that announce with fanfare and unwind quietly.

The economics of the SPAC structure also contributed to the deal’s fragility. Blank check companies carry a deadline for completing their acquisition or merger, typically 18 to 24 months from IPO. If the deal is not completed within that window, the SPAC must return capital to shareholders. This deadline pressure means SPAC targets are selected and structured under time constraints that do not always align with the pace at which crypto markets evolve. When the CRO Strategy venture was designed, the crypto treasury thesis was still generating investor enthusiasm. By the time the SPAC needed to close, the market had moved on.

Researchers found 821 accounts that made $8.2 million by manipulating bitcoin prices in the final seconds before Polymarket settled its short dated contracts. Polymarket has now replaced instant snapshots with time weighted averages, but the structural vulnerability they exposed is not unique to one platform.

On August 7, 2026, Polymarket announced it would replace the single price snapshot it had used to settle short dated crypto contracts with a time weighted average price, known as a TWAP. The change followed months of trader complaints, public warnings from onchain analysts, and a peer reviewed study from researchers at Stanford University and Singapore Management University that documented how a small number of accounts systematically profited by moving bitcoin prices on Binance in the final seconds before Polymarket’s settlement windows closed.

The exploit was not a hack. No smart contract was compromised. No private key was stolen. The vulnerability was a design choice: Polymarket settled its short dated crypto markets using a single price at a single moment. Anyone who could move that price for a few seconds could change the outcome of the contract. The researchers called the vulnerability “structural,” and their language was precise. “An asset price contract settles on a financial price,” they wrote, “and that price can be moved by trading the underlying market itself.”

The finding raises questions that extend well beyond one platform. Prediction markets handled $50.6 billion in volume in July 2026 alone. As these markets grow, the intersection between prediction contract settlement and spot market manipulation becomes a systemic concern, not a niche complaint from retail traders who lost money on five minute bets.

The timing of the fix also coincides with Polymarket’s reported effort to raise $1 billion at a $20 billion valuation. For a platform seeking institutional capital, the public documentation of a manipulation vulnerability that went unaddressed for months creates a due diligence problem. Institutional investors do not merely evaluate growth metrics. They evaluate integrity infrastructure. The TWAP transition can be read as both a genuine security improvement and a necessary precondition for closing a fundraising round with investors who take market structure risk seriously.

How the manipulation worked

The mechanics were straightforward. A trader would accumulate a position on Polymarket’s five minute bitcoin up or down contract. These contracts pay out based on whether bitcoin’s price is above or below a specific threshold at the moment of settlement. The trader would then wait until the final seconds before settlement and place a large order on Binance, the world’s largest spot exchange by volume, to push bitcoin’s price across the strike threshold.

The order did not need to be large enough to sustain a price move. It only needed to last long enough for the settlement snapshot. Once the contract resolved, the manipulator would close the Binance position, often at a small loss, and collect the Polymarket payout. The Binance loss was the cost of doing business. The Polymarket profit was the payoff.

The strategy worked because five minute contracts have thin time horizons. Moving bitcoin’s price by a fraction of a percent for five seconds is expensive but achievable for a well capitalized trader with access to Binance’s spot order book. The study found that the manipulation was concentrated in the final seconds, with unusually large orders appearing just before settlement and rapid price reversals immediately after.

The cost structure made the trade attractive. A trader might lose $5,000 to $20,000 pushing bitcoin’s price on Binance, but collect $50,000 or more from the Polymarket payout. The spot market loss was predictable and bounded. The prediction market gain depended only on whether the price crossed the threshold at the exact moment of settlement. As long as the Polymarket position was large enough relative to the Binance cost, the combined trade was profitable regardless of where bitcoin’s price ended up minutes later.

The researchers noted that the manipulators showed increasing sophistication over the study period. Early instances involved crude large orders that were visible in the order book for several seconds. Later instances used more fragmented order placement, splitting the price push across multiple smaller orders that arrived in rapid succession during the final two to three seconds. This fragmentation made the activity harder to detect in real time, though the statistical signature in the settlement data remained clear.

The paper did not prove that the Binance orders and the Polymarket positions were placed by the same individuals. It could not, because Polymarket operates as a DeFi native platform where traders use pseudonymous wallets. But the statistical patterns were consistent with coordinated activity, and the timing correlations were tight enough for the researchers to classify specific settlement windows as “likely manipulated.”

What the research actually found

The study, published on arXiv by researchers from Stanford and Singapore Management University, analyzed roughly two months of five minute bitcoin contracts on Polymarket. The researchers identified 821 accounts whose trading activity was concentrated in windows where the statistical signatures of manipulation were present.

Those accounts collectively made $8.2 million during the study period. The losses came almost entirely from retail traders. After excluding market makers, who are structurally neutral and profit from spreads rather than directional bets, the researchers found that 93 percent of the losses in likely manipulated windows were borne by retail participants.

One finding stood out. In manipulated windows, “a bet the market treated as near certain was overturned one time in three.” This means that contracts priced at 90 percent or higher probability of resolving one way were flipped by last second price movements. For retail traders relying on market pricing as a signal of likelihood, this created a situation where the odds they saw on screen bore little relationship to the odds they actually faced.

The scale of the losses was asymmetric. A retail trader placing a $100 bet on a five minute bitcoin contract that was priced at 90 percent certainty expected to win $10 in profit nine times out of ten. When manipulation flipped the outcome, that trader lost $100. Across hundreds of settlement windows and thousands of participants, these small individual losses aggregated into the $8.2 million figure the researchers documented. No single retail trader lost a fortune. The damage was distributed across a large number of small participants, each of whom had no way to know that the odds they saw on screen had been distorted.

The researchers described the vulnerability as inherent to any event contract that settles on a real time financial price. The specific platform, the specific asset, and the specific contract duration all affect how easy the manipulation is. But the underlying dynamic, that the settlement price can be influenced by trading the referenced asset, applies to any platform using instant price snapshots.

The manipulators exploited a timing asymmetry that exists whenever a financial contract settles on a single price observation. In traditional futures markets, settlement prices are typically calculated from a volume weighted average of trades over a defined window, precisely to prevent the kind of end of period manipulation that Polymarket experienced. The fact that Polymarket launched with a single snapshot mechanism suggests either that the platform’s designers did not anticipate this attack vector or that they accepted the risk as a tradeoff for simpler oracle design. Either way, the result was a settlement system that rewarded traders who could move prices by small amounts for brief periods, a capability that requires modest capital relative to the profits available from correctly positioned prediction market contracts.

Why it took months to fix

The capital efficiency of the attack is what made it particularly difficult to detect through conventional surveillance. Each individual trade was small enough to look like normal market activity. The manipulators did not need to sustain the price movement for more than a few seconds, and the prediction market positions they held to profit from the manipulation were on a separate platform from the spot exchange where they moved prices. This cross platform structure meant that no single exchange could see both sides of the trade. Binance saw small orders that briefly moved bitcoin’s price. Polymarket saw contracts settling at prices that happened to favor certain accounts. Only by correlating the timing of trades across both platforms could the researchers identify the pattern, and that correlation required access to data from both systems and the statistical tools to analyze it. The attack surface existed not in any single platform’s code but in the gap between two platforms that each operated correctly in isolation but whose interaction created an exploitable seam.

Polymarket knew about the problem before the study was published. Onchain analysts raised concerns publicly as early as May 2026. Variance Lover, a pseudonymous analyst, posted an extensive breakdown on May 21 documenting the manipulation mechanism and identifying specific settlement windows where the patterns were visible.

“By now, most people are aware that market manipulation has become a major problem on Polymarket’s five minute crypto markets,” Variance Lover wrote. “The mechanism is simple: accumulate a large position on Polymarket, then move the price on Binance during the settlement window to force the market to resolve in your favor.”

A contributor who goes by the handle Christine on X noted on May 11 that the manipulation was becoming more severe, citing “precise reversals in the last few seconds.” Josh Stevens, a Polymarket developer, responded publicly: “We are looking into this a bit deeper. Do not worry.”

The gap between acknowledgment and action spanned nearly three months. During that time, the manipulation continued. The delay is notable because the fix Polymarket ultimately deployed, replacing a snapshot with a TWAP, is not a novel solution. Time weighted averages have been standard in DeFi oracle design for years. Uniswap V2 introduced TWAP oracles in 2020 specifically to prevent single block price manipulation. The concept was available off the shelf.

Polymarket has not publicly explained why the fix took as long as it did. One possibility is that switching the settlement mechanism required changes to the smart contracts that resolve bets, which need auditing and testing. Another is that the platform was reluctant to change rules midstream for live markets with open positions. A third, less charitable interpretation is that the manipulation generated volume and fees that the platform was not in a hurry to curtail.

The delay had real costs. Variance Lover estimated that manipulation continued throughout June and into July, with increasingly aggressive activity as manipulators learned they could operate without consequences. Several retail traders posted screenshots showing positions that should have been winners based on prevailing market prices but resolved as losers because of last second price spikes. The trust damage was compounding. Each manipulated settlement that went unaddressed made the next complaint louder and the platform’s credibility thinner.

How Polymarket’s fix compares to Kalshi

The fix Polymarket deployed mirrors safeguards that Kalshi, its regulated rival, already had in place. Kalshi resolves its short dated crypto markets using a CF Benchmarks price index, which aggregates prices from multiple regulated exchanges. It then applies a 60 second moving average, making it significantly harder to move the settlement price with a brief burst of trading on a single venue.

Kalshi also operates under CFTC oversight, which gives it enforcement tools that Polymarket lacks. A Kalshi spokesperson told CoinDesk that the platform has conducted 150 to 250 material investigations per quarter and made 40 to 50 referrals to the CFTC so far in 2026. Those figures cover all Kalshi markets, not just short dated crypto contracts, but they illustrate the surveillance infrastructure that comes with operating within a regulatory framework.

The structural difference between the two platforms matters. Kalshi requires identity verification for all traders. Polymarket’s DeFi version does not. When manipulation is detected on Kalshi, the platform can identify the trader, freeze the account, and refer the case to a regulator. When manipulation is detected on Polymarket, the platform can identify a wallet address but cannot easily connect it to a person.

Polymarket does operate a US regulated version under CFTC oversight, but its integrity and surveillance infrastructure has been developed largely in response to external pressure rather than as a foundational design choice. The company is reportedly seeking a $1 billion raise at a $20 billion valuation, and the manipulation episode highlights the tension between growing rapidly as a DeFi protocol and building the compliance infrastructure that institutional capital typically requires.

What the exploit reveals about prediction market design

The comparison between Polymarket and Kalshi illuminates a broader tension in prediction market design between decentralization and market integrity. Kalshi operates as a CFTC regulated exchange with centralized order matching, surveillance systems, and the ability to cancel trades that result from manipulation. Polymarket operates on Polygon with smart contract settlement that is transparent but largely automated. The transparency means that manipulation is visible to anyone who examines the blockchain, which is how the Stanford researchers identified the 821 accounts. But visibility is not the same as prevention. A centralized exchange can intervene in real time when it detects suspicious activity. A decentralized protocol must design its settlement rules to be manipulation resistant from the start, because retroactive intervention contradicts the trustless execution model that gives blockchain based markets their appeal.

The Polymarket manipulation exposed a broader design tension in prediction markets that settle on financial prices. These markets sit at the intersection of two trading systems: the prediction platform where bets are placed and the spot market where the referenced price is determined. When the two systems are connected but not coordinated, the cheaper market becomes a tool for profiting from the more expensive one.

In this case, moving bitcoin’s price on Binance for a few seconds cost less than the Polymarket payout it generated. The arbitrage was negative in the spot market and positive in the prediction market, creating a combined trade that was profitable as long as the settlement mechanism allowed it.

This is not a problem unique to Polymarket or even to prediction markets. It is a variant of the same cross venue manipulation that regulators have spent decades combating in traditional finance. The SEC’s Regulation SHO, the CFTC’s anti manipulation rules, and the EU’s Market Abuse Regulation all address scenarios where trading in one market is used to influence outcomes in another. The difference is that those frameworks assume centralized, identity verified markets with shared surveillance feeds. The prediction market version plays out across pseudonymous DeFi platforms and centralized exchanges in different jurisdictions.

The TWAP fix addresses the most obvious attack vector by making brief price spikes less effective. But it does not eliminate the underlying vulnerability. A well capitalized manipulator who can sustain a price move for 30 or 60 seconds rather than five can still influence settlement under a TWAP system. The cost is higher, but the attack is not impossible. The question is whether the increased cost makes the manipulation unprofitable, and that depends on the depth of the liquidity in the referenced market and the size of the positions available on the prediction platform.

The economics of the manipulation also raise questions about market depth. The reason five second price pushes were possible on Binance is that bitcoin’s order book, despite being the deepest in crypto, still has moments of relative thinness. During periods of lower activity, particularly around the clock during Asian or European off hours, the cost of moving bitcoin’s price by a fraction of a percent drops significantly. The manipulators timed their activity to coincide with these low liquidity windows, compounding the settlement vulnerability with a liquidity vulnerability. A TWAP reduces one problem but does not address the other.

As regulatory frameworks for prediction markets continue to develop, the settlement mechanism question will become more prominent. Platforms that want to offer contracts on financial prices will need to either use regulated price feeds, implement robust TWAP mechanisms, or accept that their markets will remain vulnerable to the same class of manipulation that Polymarket just addressed.

What to watch

Manipulation patterns after the TWAP transition. The 30 second and 60 second TWAP windows raise the cost of manipulation but do not eliminate it. Watch for evidence of sustained price pushes that last longer than the averaging window, which would indicate that manipulators are adapting to the new settlement rules.

Polymarket volume changes on short dated crypto markets. If volume declines significantly after the transition, it may indicate that a portion of the trading activity was driven by manipulators rather than genuine prediction market demand. A volume drop would validate the concern that the platform’s growth was partly artificial.

Regulatory response from the CFTC. The study provides a public, peer reviewed record of manipulation on a platform that overlaps with CFTC jurisdiction. Watch for formal inquiries, enforcement actions, or rulemaking proposals that address cross venue manipulation between prediction markets and spot exchanges.

Chainlink Data Streams performance. The TWAP mechanism depends on Chainlink for price data delivery. Any latency issues, outages, or oracle manipulation attempts would expose a new attack surface. The reliability of the data feed is now a critical dependency for Polymarket’s market integrity.

Kalshi and Polymarket competition for institutional capital. The manipulation episode and subsequent fix narrow the structural gap between the two platforms. Watch for whether Polymarket’s $1 billion fundraising effort is affected by the integrity concerns or whether the fix is treated as sufficient by prospective investors.

Copycat manipulation on other DeFi prediction platforms. The settlement vulnerability that Polymarket’s TWAP fix addresses exists on any platform that uses instant price snapshots. Smaller prediction market protocols with less liquidity and less sophisticated monitoring are potentially even more vulnerable. Watch for reports of similar manipulation patterns on competing platforms that have not yet adopted TWAP settlement.

Academic follow up research on TWAP effectiveness. The Stanford and Singapore Management University researchers documented the pre-fix manipulation in detail. A follow up study measuring whether manipulation persists or adapts under the TWAP regime would provide the first empirical test of whether the fix works in practice, not just in theory. The research community’s engagement with this question will shape regulatory confidence in TWAP as a sufficient safeguard.

Frequently asked questions