Same five questions as every week, so the series reads as a sequence. This week: a leverage crossover 21 months in the making, funding that refused to confirm the squeeze, and a liquidation record that flipped sides twice in 48 hours.
With the altcoin complex. Aggregate altcoin futures open interest overtook bitcoin's during the week to 6 September - the first crossover in 21 months - while bitcoin's own book contracted to $53.44bn, -1.64% on the week. The total aggregate held near $136bn, so this was relocation, not expansion. Assets outside bitcoin and ether had lost nearly 23% in the first half of 2026; the crossover says the leverage that returns to them returns faster than it returns to the majors. Hyperliquid's own numbers rhyme: 30-day volume of $212.09bn, up 29.59%, against top-40 open interest of $10.15bn, up from $9.69bn a week earlier - turnover leading positioning.
The single-name detail sharpens the picture. ZEC alone carried roughly $2.3-2.4bn of aggregate futures open interest at the week's close, with $542m of it on Hyperliquid against $485m a week earlier, a +11.6% rise. A name that gained +22.21% on the week while its funding stayed at the venue's average is the crossover in miniature: positioning grew, price ran, and the cost of holding both stayed unremarkable. Across the venue's tracked cohort of 25 markets, 24-hour turnover reached $27.46bn, 41.47% on the day of the snapshot - the flow, again, is doing the work the book has not yet done.
Less than the price action implied. Aggregate bitcoin perpetual funding printed 4.39% annualised for the week - positive but subdued through a squeeze that ran bitcoin through $82,000. The 8-hour reading on 4 September was 0.0079% (8.65% annualised). On Hyperliquid, bitcoin funded at 0.0012% per hour (10.5% annualised), and the OI-weighted cohort reading was 10.88% annualised on $9.87bn of open interest. The dispersion within the cohort was minimal: 23 of 25 tracked markets funded between 0.0012% and 0.0013% per hour, with LIT the notable negative at -0.0010% (-8.8% annualised, shorts paying) and FARTCOIN the one elevated positive at 0.0051% (roughly 45% annualised). Basis, in the sense of a term structure, remains absent from the data this cohort exposes.
The carry arithmetic makes the point concrete. At the snapshot rate, holding $100,000 of bitcoin perp exposure on Hyperliquid costs a fraction of one percent of notional over a month - cheaper than the week's realised price movement by an order of magnitude. Exposure was not expensive; it was uncertain. That distinction is what a funding print at the venue's own average conveys: the market financed repositioning gladly and conviction reluctantly.
| Holding period | Funding cost | Share of notional |
|---|---|---|
| 1 day | $28.80 | 0.03% |
| 1 week | $201.60 | 0.20% |
| 1 month | $864.00 | 0.86% |
| 1 quarter | $2,628.00 | 2.63% |
Hyperliquid held 36.6% of reported 24-hour perp DEX volume across 87 tracked venues on 4 September, per PerpFinder citing DefiLlama. Aster followed at 11.4%, ApeX Omni at 7.1%, edgeX at 6.5% and Lighter at 6.2%; no other venue reached 6%. The concentration is unchanged in shape from prior issues, but the volume base under it grew: 30-day fees at the venue printed $70.0m. The regulated side of the market advanced on schedule rather than in volume: the Bitnomial pathway reported on 31 August carries a 10 to 12 month approval estimate.
| Venue | Share of 24h volume |
|---|---|
| Hyperliquid | 36.6% |
| Aster | 11.4% |
| ApeX Omni | 7.1% |
| edgeX | 6.5% |
| Lighter | 6.2% |
| Variational | 5.9% |
| GMTrade | 5.6% |
| Grvt | 1.9% |
| StandX | 1.8% |
| Pacifica | 1.6% |
Two features of the distribution are worth carrying forward. The leader's share is roughly three times the second venue's, but the middle of the table is thickening - six venues now sit between 5% and 12%, which was not the shape two quarters ago. And the volume being shared is larger: the category's 24-hour turnover on the snapshot day reached $27.46bn on the tracked cohort alone, 41.47% on the day. Concentration in share is coexisting with competition in kind.
The week's liquidation record reads as three regimes. The 24 hours to 3 September cleared $286m, 57% longs - the strike-driven risk-off day. The 24 hours to 4 September cleared $557m at 99,000 traders, with shorts taking 84% of the flow as the Waller squeeze ran; the largest single position was $5.26m. Around the payrolls print, roughly $278m liquidated, predominantly longs. The ZEC window is the cleanest single-name case: $36.6m liquidated in 24 hours, $34.5m of it shorts, against a price that settled the week +22.21% higher at $1,014.58. Forced short covering, then a long flush on the data surprise, inside 48 hours.
| Window | Total | Long / short | Traders | Reported by |
|---|---|---|---|---|
| 24h to 3 September | $286.0m | $164m long / $122m short | 105,000 traders | Gate news, 3 September 2026 |
| 24h to 4 September | $557.0m | $88m long / $469m short | 99,000 traders | TASS |
| Around the payrolls print, 4 September | $278.0m | predominantly longs | - | Zerocap |
The 3 September window carries its own detail: $286m cleared, 57% of it longs, across 105,000 traders, with the largest single position at $4.74m. Set beside the 4 September squeeze and the payrolls flush, the pattern is a market that liquidated its weak longs on the risk-off day, its weak shorts on the squeeze, and its late longs on the surprise - a full wash-and-rebuild inside three sessions, with the aggregate book barely changed in size.
The desk's summary of the week: risk moved, size did not. The aggregate book held near $136bn, funding held near the venue's average, and what changed was where the leverage sits. Weeks like this one resolve on scheduled data rather than on positioning - which is why the two events above, not the crossover itself, are the near-term risk events for anyone holding alt-side perps into the middle of the month.
Market structure (where the risk sits), funding and basis (what exposure costs), venues (who holds the flow), liquidations and risk (who was forced and when), and outlook (what would change the desk's reading). The structure is fixed so the series is comparable week to week.
Aggregate futures open interest on assets excluding bitcoin crossed above bitcoin's for the first time in 21 months, during the week to 6 September 2026. It is a leverage-location signal: it says where risk is positioned, not where prices will go. The total aggregate held near $136bn, so the event was relocation rather than expansion.
Because it indicates the move was driven by short covering rather than new leveraged longs chasing. Aggregate bitcoin funding printed 4.39% annualised for the week - positive but near its own average - and Hyperliquid's OI-weighted cohort reading was close to the venue norm. Fresh leveraged demand would show up as a funding lift-off.
US CPI for August on 11 September 2026 - the condition Governor Waller attached to waiting one meeting - the September meeting, priced at 60.4% odds of a hike after the payrolls print.
Hyperliquid held 36.6% of reported 24-hour volume across 87 tracked venues on 4 September 2026, per PerpFinder citing DefiLlama. Aster followed at 11.4%; no other venue reached 8%.
Bitcoin's dollar open interest fell 1.64% on the week, but its coin-denominated book grew 3.2%, to 709,000 BTC on 4 September from 687,000. The dollar value fell on price, not on coins leaving the market. Reading both measures separates exit from mark-down - which is exactly the distinction the crossover debate turns on.
At the 4 September snapshot rate, holding $100,000 of bitcoin perp exposure on Hyperliquid costs a few tens of dollars a day and under a thousand dollars over a month - a fraction of the week's realised price movement. The full table of holding periods is printed in section 2. Carry was not the week's risk; data surprise was.