Exchange Updates

Riset terverifikasi sumber: Hyperliquid / OKX

7 mnt

Isi artikel ditampilkan dalam bahasa Inggris asli yang telah diverifikasi. Terjemahan ditambahkan setelah tinjauan sumber.

The short answer

Papertrade, a new perpetual futures exchange on Hyperliquid's HyperEVM that offers up to 1,000x leverage, priced every trade at the midpoint of Hyperliquid's best bid and ask. In its first eight hours after the 10 October 2026 launch, one wallet made about $1.28 million, and on-chain analysis cited by The Block shows it buying and selling thousands of ETH on Hyperliquid to nudge that midpoint while holding huge positions on Papertrade. Papertrade says it has tightened open-interest limits and that the weakness was loose launch limits, not its design; its own documentation, however, lists this exact manipulation path as a residual risk. Hyperliquid itself was not hacked.

Tampilkan lebih sedikit

What Papertrade is and why its price feed matters

Papertrade went live on HyperEVM, the smart-contract layer connected to Hyperliquid, on 10 October 2026, offering synthetic Bitcoin and ETH contracts with leverage up to 1,000x, according to Crypto Briefing. The design is unusual in three ways, all described in Papertrade's own documentation:

  • No order book. Every trade is a synthetic swap between the user and a protocol pool. When you open or close, the contract reads Hyperliquid's best bid and best offer (the "BBO") through a precompile and uses the midpoint as your price. No perpetual contract is traded on Hyperliquid itself.
  • A pool that starts at zero. The liquidity pool has no seed capital and grows only from traders' losses. If it cannot pay a winning close in full, the unpaid profit goes into a first-in, first-out queue that is paid as later traders lose.
  • A token for losing. Losing trades and liquidations mint PAPER. While the tracked pool is below $2 million, the rate is 100 PAPER per dollar lost; PAPER stakers share protocol revenue, which comes from trader losses.

The price feed is the weak point. At 1,000x leverage, a 0.1% move equals your entire margin, and Papertrade's liquidation page says a 1,000x position is liquidated after a move of about 0.052%. If someone can shift the reference price by a few hundredths of a percent at the right moment, the payoff on a nine-figure position is large, while the cost of pushing a deep Hyperliquid order book for a minute is only slippage and fees.

What the trader did, step by step

The allegation surfaced on launch day. An X user, @Dr_bobo54, posted on 10 October that two wallets were moving ETH 10 to 20 basis points on Hyperliquid with trades of roughly $20 million each while holding nine-figure longs on Papertrade, as BlockTempo reported with the post embedded.

The Block then analysed one wallet's records; its findings were carried by bloomingbit:

Tabel artikel: gulir secara horizontal atau fokuskan area tabel dan gunakan tombol panah.

MeasureFigure
ETH bought on Hyperliquid within about one minute on 10 October~6,060 ETH, lifting the price about 0.12%
ETH then sold~5,210 ETH, pushing the price back down
ETH traded on Hyperliquid that day~$231 million
Volume on Papertrade~$3.18 billion
Profit in the first eight hours~$1.28 million
Deposited / withdrawn~$1.2 million / ~$2.48 million

Crypto Briefing added that the wallet made 305 trades in those first eight hours and that its gains peaked at about $1.73 million. The run then reversed. BlockTempo, citing on-chain monitoring, reported that the wallet opened 30 short positions at about $2,507.35 with roughly $297 million of notional, sold about 6,600 ETH on Hyperliquid to push the price to $2,504.40, failed to close, and was liquidated when ETH bounced to $2,509.95. The loss was about $450,000; bloomingbit puts that liquidation on an ETH short opened at 660x.

Papertrade's response

After the reports spread, Papertrade said it was tightening open-interest limits and lowering caps both on-chain and in its trade relay, according to the bloomingbit account of The Block's report. The exchange said the attempt did not come from a flaw in the system's design but from position limits that had been set loosely to handle early demand. It also said the trader had returned most of the profit to the pool through another account; The Block said it could not verify that, and public records still showed net withdrawals of about $1.28 million. The pool balance fell from about $5 million to about $2.4 million during the episode before recovering.

That account sits uneasily with the documentation. Papertrade's risk page describes "BBO jitter" in plain terms: an attacker can move the best bid or ask, trade Papertrade at the distorted midpoint, and profit when the quote returns. Its pricing page says the oracle rejects only a zero or crossed quote and has no maximum-deviation, time-weighted or lookback check. The listed defences are a 0.2 basis point deadband on winning closes, a haircut that hits small wins hardest, per-side open-interest caps and a team-run keeper that can retire a market. The documentation states these reduce payoffs but do not make manipulation unprofitable.

Why this was possible on Papertrade but not the same risk on Hyperliquid

Hyperliquid's own perpetuals do not settle on a raw midpoint. Its documentation says the oracle price used for funding is a weighted median of centralised-exchange prices updated about every three seconds, and the mark price used for liquidations is a median of three inputs, including perp prices on Binance, OKX, Bybit, Gate and MEXC. Moving Hyperliquid's book for a minute does not move those external venues. Papertrade chose the raw BBO because it is free, instant and readable by any HyperEVM contract; analysts quoted by Crypto Briefing and BlockTempo suggested switching to Hyperliquid's oracle as the fix.

How the launch went for everyone else

The manipulation story sits inside a brutal first day. Crypto Briefing reported pre-deposits above $137 million from more than 11,000 addresses and $3 billion of open interest in BTC contracts on launch day, with net realised trader losses of about $18.2 million to $18.6 million and staker rewards of about $10.3 million to $12.6 million. BlockTempo, using protocol statistics and the TradingBeats tracker, put net losses near $30.24 million, counted 14,158 liquidated positions, and found that 73.3% of active addresses used 1,000x, that 1,000x positions accounted for 89.4% of liquidations, and that the median time from opening to liquidation was 7 minutes 35 seconds. The figures differ by source and method; the direction does not.

What to watch next

  • An oracle change. A move to a multi-source price, or a deviation check against Hyperliquid's oracle, would close the path the documentation describes. Tighter caps only shrink it.
  • The payout queue. If winners stack up faster than losers refill the pool, profits wait in the queue. Watch whether it clears within days after a one-way move.
  • Contracts and audits. BlockTempo reported that, as of MONI Research's pre-launch review, verified contract source and an audit report had not been published. Check whether they appear.
  • Copycats. Any protocol that settles against a single venue's live quote carries the same exposure.

How to act on this

If you trade perpetuals, the practical lesson is to check how a venue sets the price it liquidates you at, not only the leverage it offers. A mark price built from several exchanges is far harder to push than one order book's midpoint, and 1,000x leaves no room for even a routine wick. Keep leverage low enough that a 1% move does not end the position, use isolated margin, and track crowding on our funding-rate board and liquidation tracker. If you want on-chain perps, Hyperliquid's own markets use the multi-exchange mark price described above, and you can sign up with code NEXU888 (Hyperliquid profile). On a centralised venue, OKX code nexu888 returns 40% of trading fees, settled hourly (OKX profile).

Quick answers

Was Hyperliquid hacked? No. Papertrade is a separate app on HyperEVM that reads Hyperliquid's quotes. The trader bought and sold ETH on Hyperliquid normally; the loss risk fell on Papertrade's pool.

How much did the trader make? About $1.28 million in the first eight hours, according to The Block's analysis, before losing about $450,000 on a liquidated ETH short.

Is this legal or a hack? No code was exploited; the trades used the price feed as documented. Whether it amounts to market manipulation under any law is not established, and no authority has said so.

Did Papertrade fix it? It tightened open-interest limits. Its documentation still describes the BBO midpoint as the price source, with no deviation check.

Can I lose more than my margin on Papertrade? The documentation says liquidation forfeits the full isolated margin; your other balance is separate.

What we could not verify

We could not read The Block's article directly and rely on bloomingbit's detailed account of it, plus Crypto Briefing and BlockTempo. We did not verify the wallet transactions ourselves, Papertrade's claim that profits were returned, or the identity of the traders. Launch statistics differ between sources.

Limits of this article

This is reporting on a third-party protocol, not a recommendation to use Papertrade or any 1,000x product. Figures were read on 11 and 12 October 2026 and may change as more data appears. Nothing here is investment advice. Partner links may earn NexuAlpha a commission paid by the exchange, at no extra cost to you.

Sumber

07
  1. PapertradePapertrade Docs (Trades and pricing, Pricing: BBO precompile, Risk, Liquidations, The LP and the payout queue, Asymmetric impact PnL)
  2. bloomingbit (citing The Block)Suspected Price Manipulation on Hyperliquid-Based Exchange Papertrade Nets Trader $1.28 Million
  3. The BlockTrader nets $1.28 million on new 1,000x exchange Papertrade by moving Hyperliquid's ETH prices
  4. Crypto BriefingTrader nets over $1M on Papertrade's launch, then loses $450K
  5. BlockTempo贏了要排隊,輸了能領幣:一次看懂 Papertrade「爆倉即挖礦」的 1000 倍莊家生意
  6. Hyperliquid DocsRobust price indices
  7. NexuAlphaPartner feed: invite codes, rebate shares and settlement cadence
Pilih jalur riset berikutnya