A $598M short book on one venue, and 24% of it is bearish the venue's own token

A market-neutral book does not care much about direction, which is why the $80.8 million unrealized loss is the least interesting number in the file ^[6]. Set against realized profits above $300 million, most of it harvested from funding payments rather than price calls, the drawdown reads as a carry cost ^[7]. Net of it, the strategy is still ahead by more than $219 million ^[16].

The number that does work is the HYPE line. A $141.6 million short on HYPE is 23.7% of the $598 million book, and HYPE is the native token of the venue clearing every one of these trades ^[2][14][1]. Abraxas has paired that short with spot HYPE, so the position earns the funding differential rather than needing HYPE to fall ^[9]. Reasonable as a trade. Awkward as a structure: unwinding it means buying back nine figures of the same token on the same platform, where crypto.news describes this firm as frequently the largest single whale ^[5].

The ETH leg shows how tightly it is run. $173.17 million pulled off Binance over four days, against a $193.9 million ETH short, covers roughly 89% of the notional in spot and leaves about $21 million genuinely exposed to price ^[3][2][13]. That is a funding trade with a rounding error attached, not a view on Ethereum.

Add the four named positions and you reach $576.7 million, or 96.4% of the reported total ^[11]. There is very little room inside "predominantly short" for an offsetting long book, which means the gross figure and the directional figure are nearly the same thing. At the firm level, roughly 15% of more than $4 billion under management is sitting as notional on a single decentralized perpetuals venue, in two wallets ^[12][1].

Then there is the discrepancy in the reporting itself. The same story is headlined $783 million of short positions while its body puts the Hyperliquid book at roughly $598 million, a gap of about $185 million ^[10][15]. Anyone trying to size the overhang above this market is picking between two numbers from one publisher.

History says the book moves in size. Abraxas took shorts from $760 million in November 2025 down to about $270 million, a 64% cut, then rebuilt ^[8][17]. It has also added roughly $19.5 million of gross shorts inside a two-hour window ^[4]. Since mid-2025 the range has run from $500 million to more than $900 million, with a peak above $900 million ^[5]. A participant that can add $19.5 million in two hours and shed $490 million over a month is not a stable feature of the order book; it is a variable other traders on the venue have to model.

That is the actual transferable risk here. Abraxas is roughly hedged and, by its own P&L, well paid for the trouble. The people who are not hedged are everyone else quoting HYPE, ETH, BTC and SOL perps on the same platform, whose fills depend on depth being there on the day the largest single account decides to reverse ^[5][2].