Curriculum·J302 Swap Execution: AMMs, MEV, and DEX Architectures·about 33 min
Perp DEXs
By the end of this lesson you can
- →Name the components a perpetual venue adds beyond a spot exchange
- →Identify who absorbs a liquidation that cannot be filled, and what that makes them
- →Compute why a thinly traded listing is an attack on the venue rather than on a trader
- →Determine who can override a price, and how quickly
Junior · enrolled learners
This lesson opens with Hyperliquid and JELLY, March 2025.
- What happened
- On 26 March 2025 a trader opened a large short in the thinly traded JELLY perpetual on Hyperliquid while buying the underlying token on-chain elsewhere, then pushed the spot price up. The rise triggered liquidation of their own short, and under the venue's rules a liquidated position that cannot be closed in the market passes to the HLP vault, which is funded by depositors. The vault inherited a short of several million dollars in a token whose whole market capitalization was reported at between about $10M and $50M over the episode, and its unrealized losses reached between $12M and $13.5M as the price rose by over 400 percent. Hyperliquid's validators reached consensus in about two minutes, delisted the market, reset the JELLY oracle price, and settled all positions at $0.0095, the price at which the short had been opened, against a market that had traded as high as $0.043. At that price the vault closed the position at a profit of about $703,000. Arthur Hayes wrote that it was time to stop pretending Hyperliquid was decentralized, and Bitget's chief executive said Hyperliquid may be on track to become FTX 2.0.
- The decision point
- The vault depositors were the counterparty to a liquidation nobody else would take, which is the component a perpetual venue adds and the one people deposit into for yield without reading. And the resolution was a price override agreed by a validator set in two minutes, which protected the vault by deciding what the asset was worth. Both facts were properties of the architecture before the trade, and both are readable from the documentation.
- Recorded loss
- $13,500,000
What you will be able to answer
- →What does a perpetual venue add beyond spot?
- →Who is the counterparty of last resort?
- →Why is a thin listing an attack surface?
- →What was the resolution and what does it tell you?
Orientation and Year One are open: anyone can read them without an account. From Year Two onward the lessons are for enrolled learners, because progress through the later years only means anything if it is tracked against a record.
It is free. We do not sell the list and there is nothing to buy at the end of it.
Sources and review
- https://www.coindesk.com/markets/2025/03/26/hyperliquid-delists-jellyjelly-after-vault-squeezed-in-usd13m-tussle
- https://www.halborn.com/blog/post/explained-the-hyperliquid-hack-march-2025
- https://oakresearch.io/en/analyses/investigations/hyperliquid-jelly-attack-context-vulnerability-team-solution
- https://incrypted.com/en/hyperliquid-loop-jellyjelly-manipulations-12-million-losses-and-questions-of-decentralization/
- https://decrypt.co/311713/hyperliquid-delists-solana-meme-coin-liquidation-crisis
- https://www.fxstreet.com/cryptocurrencies/news/hyperliquid-escapes-liquidity-crisis-following-jellyjelly-short-squeeze-delists-token-202503262225
- https://cointelegraph.com/news/bitget-ceo-slams-hyperliquid-jelly-perps-delisting
- https://hyperliquid.gitbook.io/hyperliquid-docs/hypercore/vaults/protocol-vaults
Confidence medium·Volatility high·Reviewed 2026-08-06·Owner unassigned
Contested
Whether the validator override was the right decision is genuinely contested. It protected vault depositors from a loss caused by manipulation, and it required deciding an asset's price by vote. Serious commentators took both positions and this lesson presents the tradeoff rather than resolving it.
Reports of the JELLY price at the intervention differ by an order of magnitude. CoinDesk and Halborn give $0.50. Decrypt reports a high of $0.043 on CoinGecko data, FXStreet values 124.6 million tokens at $4.5 million, about $0.036 each, and a rise of over 400 percent, which Halborn and Incrypted both report, from a start near $0.01 (Incrypted values the 400 million token short at $4.1 million) also puts the price near $0.05. This lesson uses Decrypt's $0.043 and reads the $0.50 as an error. The settlement price of $0.0095 is not disputed.
Figures for the trader's short and the vault's exposure vary with the source and the moment: the short is reported at $4.1M (Halborn), $4.5M (Oak Research), $6M (Decrypt) and $8M (FXStreet), and HLP's unrealized loss at $12M (Oak Research, Halborn) and $13.5M (CoinDesk). The settlement left HLP with a profit of about $703,000 (FXStreet), and Hyperliquid said all users apart from flagged addresses would be made whole by the Hyper Foundation (CoinDesk). The two-minute consensus time is Halborn's; Oak Research says only that the quorum was reached in minutes with no dissenting votes.
R402 covers perpetual mechanics at Senior level, including funding, mark price construction and auto-deleveraging in depth. This lesson covers the perpetual venue as an architecture and the backstop as its distinctive component. Keep the split.