Funding told the truth and the OI ledger confirmed it: there was no crowded long book going into the event, because it had already been paid to leave.
Perp funding across major venues sat at approximately +0.005% per 8 hours on Binance and OKX and +0.01% on Bybit through the week - at or just below the 0.01% neutral anchor, about 5.47% annualised (Velo via Zerocap). The number is easily misread. Binance's funding formula anchors at the neutral rate whenever the premium index sits inside the +/-0.05% clamp, so a 0.005% print implies the perpetual trading at a slight discount to spot. This is a flat basis, not leveraged long demand - and in an event week, flat basis plus falling open interest is the signature of a market that de-risked rather than positioned.
The rates market supplies the reason. Hike odds for the 16 September decision moved from 64% to 70% around the PPI print and ended the week in the 85%-90% band, from 58% at the start of it (Zerocap, futures and prediction markets) - a first hike since July 2023 now the base case rather than the tail. Goldman Sachs, JPMorgan, Citi, MUFG and TD have all shifted to call a 25bp hike and Barclays calls consecutive hikes in September and December published hike calls into the meeting. The options book priced the binary rather than a tail: DVOL near near 37%, the 18 September expiry at about 35% ATM, and risk reversals at +1.14 vol to -0.37 on the weekly tenor and +1.03 vol to -0.75 on the monthly - skewed, not panicked. A funding print at the neutral anchor inside that rates context is a book that left, not a book that doubled down.
The only OI reading published this week with a clean weekly delta: Binance BTC futures OI fell roughly $529m over seven days to about $8.0bn (Velo via Zerocap). Aggregate cross-venue OI was not republished by a whitelisted source this week; the desk does not carry last week's aggregate forward, because a stale figure presented as current is worse than no figure. On the on-chain book, Hyperliquid's own API printed $10.60bn across 234 markets on 10 September, with $6.85bn of 24-hour volume.
| Venue / basis | Open interest | Note |
|---|---|---|
| Hyperliquid (DeFiLlama basis) | $14.67bn | 76% of the $19.20bn tracked cohort |
| Aster | $2.50bn | second-largest perp DEX by open interest |
| Lighter | $1.10bn | third-largest perp DEX by open interest |
| Hyperliquid (exchange API basis) | $10.60bn | the venue's own measure, 10 September - a different denominator, printed side by side |
The dual print is worth a paragraph of method. The exchange API figure counts every market the venue lists; the aggregator cohort counts only the markets its coverage reaches, so the same venue lands at $14.67bn on one basis and $10.60bn on the other. Neither is wrong - they answer different questions - but only the narrower basis has a clean week-on-week delta this week, which is why the bleed argument above rests on Binance rather than on the DEX cohort.
The table below prices the cost of holding a leveraged long at the funding level the market actually printed - +0.005% per 8 hours, annualised to 5.47% - on a $100,000 notional position. It is the cheapest leveraged long of the recovery so far: flat basis means the carry, not the crowd, is the story.
| Holding period | Funding cost | Share of notional |
|---|---|---|
| 1 day | $15.00 | 0.01% |
| 1 week | $105.00 | 0.10% |
| 1 month | $450.00 | 0.45% |
| 1 quarter | $1,368.75 | 1.37% |
Ether funding flipped to -0.0056% per 8 hours on the 10 September flush (OKX), the week's only negative print on a major market - shorts paying to stay short as price broke $77,000. The timing is the information: the flip lasted one session and reversed into the squeeze that took $250m+ of shorts out within an hour on 11 September. A funding sign flip that survives less than a day is positioning noise; the same flip held for a week would be a regime change. This one did not hold.
The same session produced the week's cleanest piece of OI microdata. The largest bitcoin long on Hyperliquid - $233m - went from profit to a $-3.39m unrealised loss on the 10 September flush (OKX), and stayed open through the 11 September squeeze. A position of that size riding through both sides of a two-regime week is the opposite of the liquidation-cascade narrative: the whale did not get forced, the crowd around it did. It is also a reminder that headline liquidation totals measure the forced margin of the book, not its centre.
The August monthly figures are retained from the monthly report as context and were not re-verified this week: $106m in Hyperliquid fees (+23% month on month) on $400.00bn of volume, an annualised revenue run rate of $1.251bn. September's trailing-7-day ledger from DefiLlama - $13.5m in fees, $10.42m in revenue - is this week's fresh read, and the weekly cadence will build the September comparison as the month closes.
Funding is a flow against a stock, and the stock kept growing through the de-risking week. The Big-3 stablecoin base rose +$742.7m over the seven days to 9 September, from $261.79bn to $262.53bn (DeFiLlama via NeoDrop): USDC added +$661.6m, USDT +$80.6m, and TRON's USDT float grew +$504.2m to $92.27bn - settlement collateral rather than margin, but the rail the trading float ultimately clears against. Hyperliquid's own USDT margin base stood at $85.4m. Read together with the falling open interest, the week's shape is unusual: more buying power entering the system while leveraged exposure leaves it. That combination - dry powder accumulating against a de-risked book - is the mechanical setup both a squeeze and a flush feed on, which is exactly what the following week delivered in miniature on 11 September.
The FOMC decision lands 14:00 Eastern on 16 September, with the Summary of Economic Projections. For the funding ledger, the readings that matter are: whether the post-decision basis lifts off the neutral anchor (a sign of re-leveraging), whether Binance BTC OI stabilises near about $8.0bn, and whether ether funding stays positive after its one-session negative print. Each is a condition, not a prediction.
On Binance and OKX it printed as a slight perp discount to spot: the funding formula anchors at the 0.01% neutral rate while the premium index sits inside its clamp. Annualised it reads about 5.47% - but the correct reading is 'flat basis', not 'cheap leverage on a bullish consensus'.
At the week's Binance/OKX print of +0.005% per 8 hours (5.47% annualised), a $100,000 long costs about $15.00 a day, $105.00 a week, $450.00 a month and roughly $1,368.75 a quarter in funding alone. The full table is in the article.
Because no whitelisted source republished the cross-venue aggregate with a weekly delta this week, and the desk does not carry a stale figure forward as current. The clean reading that does exist - Binance BTC OI down $529m over seven days to about $8.0bn - is printed with its own scope.
No. It lasted one session: -0.0056% per 8 hours on the 10 September flush (OKX), reversing into the 11 September squeeze that liquidated $250m+ of shorts within an hour. A one-session flip is positioning noise; the desk treats a week-long negative print as the regime signal.
Because funding is the price of leveraged exposure, and the rates market sets the appetite for taking it. Hike odds moving from 64% to 70% around the PPI print - and into a 85%-90% band by Friday, with Goldman Sachs, JPMorgan, Citi, MUFG and TD have all shifted to call a 25bp hike and Barclays calls consecutive hikes in September and December publishing hike calls - is the reason the OI ledger bled $529m into the meeting. A funding page that ignored the rates path would print the symptom without the cause.
No - it is the week's most useful tension. The stablecoin base ($262.53bn, up +$742.7m) measures buying power entering the system; open interest measures leveraged exposure staying in it. Both can grow, both can shrink, and when they diverge the divergence is the signal: dry powder accumulating against a de-risked book is the mechanical setup for outsized moves in either direction, which is what 11 September then demonstrated.