Bitcoin 2026-08-27 daily brief — AI stance: bearish

Benign, and that is the surprising part. Open-interest-weighted funding across venues annualizes to 4.42% — under half the roughly 11% annualized that the exchange-default rate represents as the neutral baseline. Traders are paying almost nothing to carry leveraged longs after a 27.3% run to the 90-day high, which is not what a crowded, stretched market looks like. The single major-venue read at 4.54% annualized is 0.12 percentage points off the all-venue aggregate, which is noise; no venue is carrying concentrated crowding. Even at the squeeze peak on August 22 the cross-venue rate reached only about 12.5% annualized, barely above neutral. The honest conclusion is that perpetuals are simply not where the risk sits in this move — and anyone reading clean funding as an all-clear is looking at the wrong instrument, for the reason set out below.

Market-wide futures open interest is $56.48B against $47.81B on August 15 — but that is a price illusion. Deflated by spot, it is roughly 758,000 BTC-equivalent falling to about 703,000, down 7.2%, while price rose 27.3%. Open interest declining into a rally is closing, not accumulating. Options tell the opposite story on the same apples-to-apples basis: open interest of $42.97B against $26.26B on August 15 works out to roughly 416,000 rising to 535,000 BTC-equivalent, up 28.5% in coin terms — a genuine build, not a price effect. So leverage did not leave the market, it changed venue, out of perpetuals and into calls, which is exactly why funding screens so clean. That combination has a different failure mode from a funding-driven blow-off: no cascade risk, but a position that bleeds on time and vega if price merely stalls. The single-venue book at $1.42B, down slightly in dollars from $1.45B on August 15 and therefore down sharply in coin terms, corroborates the same de-grossing at venue level.

The squeeze is finished as a mechanical force. Short liquidations ran $811.1M on August 19, $261.6M on August 20 and $470.1M on August 21 — the tape puts roughly $3B liquidated across a 45-hour window on the 19th and 20th, about 92% of it shorts — then decayed to $49.9M by August 27, a 94% collapse from peak. The character changed mid-window too: August 25 saw $87.8M of LONGS liquidated against $175.1M of shorts, and August 26 was essentially balanced at $21.0M versus $20.7M. Today's 2.04-to-1 short-to-long ratio on total activity in the 57.8th percentile of the past year is ordinary two-way flow, not a squeeze. The trapped shorts that powered this advance have been paid out and are no longer available as fuel.

The US premium is +0.31 basis points — effectively flat — and the fourteen-day path matters far more than the snapshot. It sat at -10.85 bps on August 15, right at the bottom-decile extreme, then improved almost monotonically: -10.66, -10.07, -8.50, -6.81, -5.01, -2.21, -1.59, then oscillating around zero to today's +0.31. So US spot went from actively lagging offshore to merely level with it. Critically, it never pushed into the positive-extreme zone above +10 bps that would mark aggressive US-side accumulation. That fits the ETF picture precisely: the American bid over the last two weeks was passive absorption of available supply rather than assertive buying — and with net flow now negative at -439.6 BTC, even the absorption has stopped.

CONFLICT

Spot ETF flow is the whole story, and it just turned. The eight-session inflow streak totalling about $2.8B — August's roughly $3B is the strongest month of 2026, near double April — broke on August 27 with a net outflow of 439.6 BTC. Absorption against daily miner issuance decayed monotonically across the window: 21.0x on August 20, 9.7x on August 24, 9.0x on August 25, 6.2x on August 26, then -0.96x. The 30-day flow z-score went from +2.47 on August 20 to -0.70 today, and excess absorption is -896 BTC. The 7-day average is still robust at 4,360 BTC in the 88th percentile, so this is a fresh break rather than an established trend. Elsewhere the micro is thin: miner economics are merely average with Puell at 0.985 in the 42nd percentile and hash ribbons at 0.9876 in the 9.5th, and fee revenue is no help — transaction count is at the 98.8th percentile while fees per transaction sit at the 11th, so blockspace is busy and near-worthless and the subsidy is carrying miners alone. On the constructive side, the SEC sent crypto custody rule changes to the White House on August 26, which is structurally positive but slow-acting. Bitcoin Core 32.0 rc1 targeted for September 10 is not a price event.

hawkish. Fed funds sit at 3.63% against a 10-year at 4.66%, with M2 growing 5.41% year over year. The decisive fact is that the market prices roughly one-in-three odds of a HIKE at the September 15-16 FOMC — the live debate is up, not down — and those odds are already the reduced number, cut to about 30% by a July payroll report showing a 23,000-job contraction. Kevin Warsh delivers his first Jackson Hole keynote as Chair on August 28 at 10:00 ET, on a symposium theme of financial innovation and payments policy; no policy statement is issued there, so the risk is tonal rather than mechanical, though a Fed chair speaking on payments has a non-zero chance of touching digital assets directly. Against that, Fear and Greed reads 71 (Greed) — sentiment is leaning hard into risk while a hawkish policy tail sits unpriced.

The 10-year is 4.66%, up 3 basis points from 4.63% on August 15, having peaked at 4.74% on August 24 before the Treasury's announced expansion of long-end buyback operations — at least doubling maximum sizes to $4B per operation in the 10-to-20-year and 20-to-30-year sectors — pulled yields back. Fed funds at 3.63% against that 4.66% 10-year gives a roughly 103 basis point positive slope. There is no credit-spread feed in this dataset, so I have no read on credit at all. That is a genuine blind spot rather than a benign omission, with ISM Manufacturing and JOLTS on September 1, ADP September 2, ISM Services September 3, payrolls September 4 and CPI September 11 all inside or adjacent to the window.

DXY 99.16, effectively unchanged over thirteen sessions — 99.50 on August 16, a 98.74 trough on August 20, back to 99.16 now. Range-bound, and the important inference is what did not happen: bitcoin put in +27.3% while the dollar moved under 1%. This was not a debasement or dollar-weakness trade. There is therefore no dollar tailwind to lean on for continuation, and equally no dollar headwind to fear. A neutral input that neither side of the argument gets to claim.

Risk-on and untroubled. The S&P sits at 7,731, up 0.7% from 7,676 on August 26 and within 0.2% of its 14-day high of 7,745 set on August 17. VIX at 14.51 is the lowest print in the entire 14-day window, whose range was 14.51 to 16.01. Gold at 4,637.70. Nothing in equity vol is signalling stress, which is precisely why the bitcoin-specific exhaustion signals have to carry the bearish case on their own.

This is a stall-and-retrace thesis, not a crash call, and the reason is specific: open-interest-weighted funding at 4.42% annualized and coin-denominated futures open interest down 7.2% mean there is no leveraged-long overhang to cascade. The modal path is a grind back into the five-session shelf that held from August 22 to August 26, between 76,998 and 78,879 — a 1.8% to 4.1% give-back that barely qualifies as a drawdown. The more meaningful zone is the air pocket price jumped through on August 19-20, from 72,661 down to 69,221, crossed in two sessions with essentially no time spent inside it; the top of that gap is about 9.5% below spot and the bottom about 13.8%. I weight the shallow outcome well above the gap-fill, and the gap-fill well above the tail. The genuine tail is a hawkish Jackson Hole tone on August 28 or a payrolls print on September 4 that revives the roughly one-in-three September hike odds — that is the path to the 50-day at 66,410, or -17.3%. Note the shape as much as the size: because the crowding now sits in calls rather than perpetuals, the likely failure mode is a slow bleed on stalled price, not a fast liquidation flush.

moderate. The 30-day implied-vol index closes at 41.52, in the 44.4th percentile of the past year — squarely mid-range, not elevated. Against that, 30-day realized is 42.96% and 30-day ATM implied is 38.98%, which is the 72nd percentile of the last 90 days but still roughly 4 points below realized, so options are cheap relative to what the market has actually been delivering. The real dislocation is in the term structure rather than the level: 7-day implied sits 2.1 vol points above 30-day, and 7-day realized at 57.68% is 15 points above 30-day realized at 42.96%. Moderate overall, with the risk concentrated in the next five sessions.