Hyperliquid Manual Borrows Hit $269M on Day One and Opened a Liquidation Channel Open Interest Cannot See

Hyperliquid switched on manual borrows on September 18 and users pulled $269 million in assets on the first day, according to the protocol's own announcement. HYPE traded up more than 13% that session to $90.45 with 24 hour volume higher by 75%, CoinGape reported. The number that matters for orderflow is not the price move. It is that Hyperliquid now has a liquidation channel that does not appear anywhere in perp open interest.

What manual borrows do on HyperCore

Manual borrows went live on September 18 and let users supply HYPE or BTC as collateral to borrow quote assets, specifically USDC and USDT, per Hyperliquid's documentation. Portfolio margin and manual borrows run on the same underlying HyperCore infrastructure, which means the collateral engine is not a bolted on side product. It sits in the same place as the margin system that already governs perp positions.

The feature is available to Manual, Standard and Unified Account users. For portfolio margin accounts the borrowing is automated and the manual action is disabled, the protocol said in its announcement.

Borrowed quote assets pay interest and supplied quote assets earn it, with rates set by utilization. Hyperliquid retains 10% of borrower interest as a buffer against future liquidations, since the interest paid by borrowers is distributed proportionally across a larger set of suppliers.

The LTV numbers set the liquidation math

HYPE carries a loan to value of 65% and BTC carries 50%, per the launch parameters reported by CoinGape. The partial liquidation threshold sits at 82.5% for HYPE and 75% for BTC.

Run the arithmetic. A trader who supplies $1,000 of HYPE and borrows the full $650 reaches the 82.5% partial liquidation band once the collateral is worth $788. That is a 21% decline in HYPE from the moment the borrow was opened.

The BTC side is looser. A $1,000 BTC deposit borrows $500, and the 75% threshold arrives when the collateral is worth $667, a 33% decline. Same protocol, very different room to move.

Context for that 21%: on launch day alone HYPE traded a 24 hour range of $79.35 to $90.92, a 12.7% swing from high to low. A borrower who opened at the maximum LTV and then sat through nothing worse than one ordinary session would have burned more than half the buffer.

Why collateral liquidations do not show up in open interest

A perp liquidation closes a derivative position. Open interest drops, the liquidation map registers a hit, and every dashboard tracking Hyperliquid sees it happen.

A collateral liquidation is a different event entirely. The engine sells spot HYPE or spot BTC to repay a stablecoin debt. Perp open interest does not move. The liquidation heatmap shows nothing. The only print lands in the spot book.

That gives you a specific pattern to watch: spot cumulative volume delta turning sharply negative while perp open interest stays flat. CVD tracks the running difference between aggressive buy and aggressive sell volume, so forced collateral sales show up as a run of market sells with no matching unwind in derivative positioning. When the two series separate, the selling is coming from somewhere other than perp traders closing out.

There is a reflexive edge to this on the HYPE side specifically. HYPE is both the collateral asset and the asset whose price decides whether the collateral holds. Anyone who supplied HYPE, borrowed stablecoins and bought more HYPE has built a position where a price decline forces sales that push the price lower again.

Putting $269 million in proportion

Day one borrows of $269 million are not a run rate. A launch pulls in every trader who was waiting for the feature, and utilization based rates mean the cost of carrying the borrow climbs as the book fills. The number to watch is not day one. It is whether total borrows hold above that level through a week where HYPE does not go up.

The direction of travel is clear enough. Hyperliquid is adding a credit layer on top of a venue that already settles a large share of global perp open interest. The September 16 announcement that Kraken parent Payward intends to deploy CFTC regulated perpetual futures on Hyperliquid HIP-3 markets, published by BusinessWire, points the same way. More capital on the venue, more collateral layers, more places for a forced sale to originate.

Broader conditions were supportive on launch day rather than stressed. BTC reclaimed $78,000 and total BTC futures open interest rose about 4% to nearly $53.5 billion in 24 hours, CoinGape noted citing CoinGlass, with CME open interest up more than 5%. Collateral systems are easy in that tape. They are tested in the other one.

What to track on the tape

Three things are worth watching now that borrows are live.

First, spot depth on HYPE around the levels where a cohort of borrowers would reach the 82.5% band. Order book imbalance at those levels tells you whether there is real resting size to absorb forced sales or a vacuum underneath.

Second, the divergence described above: spot CVD against perp open interest. A sustained gap between the two is the cleanest available signal that collateral, rather than positioning, is driving the flow.

Third, the BTC leg. With a 50% LTV and a 75% threshold, BTC borrowers carry a third of downside room before trimming starts. The HYPE leg is the fragile one, and it is the leg where the collateral and the venue token are the same asset.

Buildix tracks this on the HYPE pair page: spot CVD, order book imbalance, volume profile and whale wallet activity in one view, with Hyperliquid native coverage rather than a centralized exchange feed stitched onto a DEX. The wallet tracker covers the addresses large enough to matter when a collateral position unwinds.

Borrowing against a venue token is not new in crypto, and how it ends depends entirely on how much of the borrowed size sits near the top of the LTV band rather than comfortably below it. That distribution is not published anywhere. The order book is, and it will show the stress before any dashboard does.