Back to news

Hyperliquid Founder Jeff Yan: Bitget May Become the Next FTX

In April 2023, the day after Dragonfly announced a $10 million investment in Bitget, Jeff Yan, founder of high-frequency trading firm Chameleon Trading and later co-founder of decentralized derivatives exchange Hyperliquid, published a lengthy tweet publicly claiming that Bitget "may become the next FTX." He expressed respect for Dragonfly's commitment to investing during a bear market but felt it was a "moral obligation" to publicly question problematic projects before they expand, to avoid another collapse like FTX or LUNA that could set the industry back years.

He provided direct evidence from front-line trading, stating that about six months prior, while doing high-frequency trading on Bitget, they quickly discovered that the matching engine was "dishonest": their orders were queued at the best bid or ask, yet they could see trades occurring at worse prices while their orders remained unchanged. He emphasized that anyone watching the trading interface at the time could see this phenomenon.

He cited a former Bitget employee's claim that Bitget handled retail order flow using a "B-booking" method, meaning the internal trading desk acted as the counterparty directly, rather than sending orders to the public order book for matching. He argued that this was far more serious than the commonly discussed "exchanges supporting internal market makers": presenting itself as an order book exchange while using an entirely different structure to handle customer orders behind the scenes is "undoubtedly unethical" and could be illegal in any jurisdiction.

He believed the motivation was profit. Retail flow, especially that amplified by copy trading, is very profitable for acting as the counterparty; he claimed Bitget had not introduced external market makers for a long time to capture this profit. After discovering the issues, they immediately ceased all trading on Bitget. He expressed uncertainty about whether Bitget would rectify the situation but stated that even if they did, a team that believes "operating an exchange this way is acceptable" should not receive more funding to expand.

In response, Bitget's Managing Director Gracy Chen stated that Bitget's matching engine "works exactly like those of other major exchanges, though possibly with different efficiency." Jeff Yan immediately questioned this, asserting that his concerns were not just about the matching engine but the entire team and operational methods, stating, "If we completely bypass the matching engine, then the engine itself is irrelevant," and demanded a response to the accusations of "betting against customers" and "freezing profitable accounts' APIs and VIP levels." In this round of public confrontation, Bitget did not provide specific responses to these two accusations.

At the market mechanism level, the core of the dispute is the ownership of retail order flow. Under the B-booking model, the platform directly acts as the counterparty for retail traders, meaning retail losses become platform revenue; in a true order book model, retail orders are matched by external market makers competing for bids, with the platform only collecting fees. Copy trading causes many retail traders to place orders in the same direction at the same time, increasing the counterparty's win rate. The beneficiaries are platforms that control order flow information and can selectively take on trades; the pressured parties are retail traders who receive worse prices in opaque matching and profitable traders whose interfaces are frozen or levels reduced. Such accusations have also driven some high-frequency traders and funds to shift towards on-chain verifiable order book exchanges.

Supplementary data: Two years later, the roles of both parties reversed. On March 26, 2025, the JELLY token on Hyperliquid surged 230% within an hour, causing its market-making treasury HLP to suffer a floating loss of about $10.6 million, with potential losses exceeding $240 million; Hyperliquid subsequently delisted JELLY and forcibly settled positions at prices favorable to the platform. Gracy Chen, who has since become CEO of Bitget, publicly criticized Hyperliquid as "more like an offshore centralized exchange without KYC and anti-money laundering," stating it is heading towards "FTX 2.0." By July 13, 2026, Hyperliquid's peak open contracts reached $11 billion, accounting for about 9% of global perpetual contract positions; Bitget's derivatives trading volume in 2025 was $81.7 trillion, ranking in the top four among centralized exchanges.

Source: Public Information

ABAB AI Insight

Jeff Yan's background shapes his perspective. He graduated from Harvard University, worked at high-frequency trading firm Hudson River Trading, and later founded the crypto high-frequency market-making firm Chameleon Trading, where he could detect matching anomalies from transaction data early on. In 2023, he launched Hyperliquid, placing a complete order book and matching on a self-built public chain without venture capital funding. In November 2024, Hyperliquid airdropped about 31% of HYPE tokens to users. Bitget's history is fraught with controversy: in 2021, it implied a connection with the Korean boy band BTS while promoting the ARMY token, leading to a warning from management company HYBE, and the Monetary Authority of Singapore subsequently suspended its related business; the Australian Securities and Investments Commission, the German Federal Financial Supervisory Authority, and the French Financial Markets Authority have all issued warnings or blacklisted it; in 2023, its Hong Kong affiliate ceased operations.

In terms of capital pathways, the two parties represent different models. Bitget relies on venture capital endorsements, endorsements from celebrities like Messi, copy trading, and the platform token BGB for customer acquisition, with revenue coming from fees and the internally questioned counterparty. Hyperliquid operates in reverse: about 97% of trading fees are used to repurchase HYPE on the secondary market, directly distributing platform profits to token holders, using "transparency" as a customer acquisition selling point. In Q3 2025, Hyperliquid repurchased about $290 million of HYPE in a single quarter. This public accusation itself is also a form of marketing: Jeff Yan publicly questioned the opacity of centralized exchange matching before Hyperliquid's launch, providing the strongest market education for his on-chain order book.

The closest historical case is in the forex market. In February 2017, the U.S. Commodity Futures Trading Commission found that retail forex broker FXCM, which claimed a "no dealing desk" model, actually had undisclosed interests with a market maker that took on customer counterparty trades, resulting in a $7 million fine and a ban on operating in the U.S., leading FXCM to exit the U.S. market. In the stock market, in December 2020, Robinhood was fined $65 million by the SEC for failing to adequately disclose order flow payment arrangements, leading to worse execution prices for customers. FTX's issues also stemmed from the matching layer: trial evidence showed that affiliated market maker Alameda had special permissions allowing negative balances on FTX. The crypto derivatives industry is currently in a phase of "re-pricing transparency."

The essence is a transfer of pricing power: the pricing power of order flow is shifting from opaque centralized matching to on-chain verifiable matching. The mechanism lies in information asymmetry: centralized exchanges can see all users' orders, positions, and stop-loss levels but do not have to prove whether each transaction went through a public order book, allowing the platform to selectively take on retail orders, turning matching services into a betting business. On-chain order books make every order, transaction, and settlement publicly traceable, preventing platforms from keeping a separate account behind the scenes. However, the JELLY incident also illustrates that once a team intervenes in settlement and delisting at an on-chain exchange, governance risks also exist. Both companies accusing each other of being the "next FTX" are essentially competing for the same thing: traders' trust in the fairness of matching.

ABAB News · Law of Cognition

  1. When the platform acts as your counterparty, fees are just a fraction.
  2. Invisible matching equates to visible conflicts of interest.
  3. Accusing the counterparty of opacity is the best advertisement for the transparent party.

Notes:

  • Writing Style: Jeff Yan's statements are unilateral accusations that have not been verified by third parties, so the newsflash is treated as a statement type, using "he claims," "he cites," "alleges" throughout, and includes Bitget's response. This is an old story from April 2023; if reported as news, it is recommended to label it as a retrospective or create a comparative piece with the JELLY incident titled "Both Sides Accuse Each Other of Being FTX."
  • Unverified Background: The contents in the AI interpretation come from publicly available English information I possess, and this round has not been cross-verified online item by item.
    • Jeff Yan's resume
    • HYPE airdrop ratio
    • Details of FXCM and Robinhood's fines
    • Evidence from the FTX trial.

Source

·ABAB News
·
10 min read
·13 hrs ago
分享: