Bitcoin 2026-08-28 daily brief — AI stance: bearish

Weighted by open interest across venues, funding annualizes to roughly 10.5% — that is at or just below the level implied by the exchange-default eight-hour rate, which is the neutral baseline. Longs are paying an ordinary carry, not a stretched one, and nothing here supports a crowded-long narrative. The single major venue in the local feed runs cooler still, near 8% annualized, about 2.5 percentage points below the all-venue aggregate; that says the modest crowding which does exist sits away from that particular venue, and it is a venue-level observation rather than a spread between two market-wide measures. The important inference is what this combination means for the 08-28 flush: 102.8 million dollars of longs were liquidated without funding ever having run hot. That makes it a positioning event rather than a leverage unwind — and it means there is no accumulated funding excess whose reset would mark a low. The usual signature of an exhausted downside move is simply not available here.

All-venue futures open interest stands at 54.54 billion dollars, up 15.0% from 47.43 billion on 08-16 but already 4.8% below its 08-24 peak of 57.30 billion — leverage was built through the advance and has begun to bleed off. The venue-level futures book in the local feed tells a consistent story on its own smaller scale, drifting from 1.49 billion on 08-17 to 1.42 billion. Options are where the aggression actually went: open interest of 45.09 billion dollars is up 74.7% from 25.81 billion twelve sessions ago, with 6.56 billion of volume on 08-28 and roughly 6.4 billion of Deribit expiry clearing the same Friday. A market that nearly doubled its options exposure while paying only baseline funding is expressing direction through convexity rather than leverage — and with one-week 25-delta skew at -0.045, in the 3rd percentile of its thirty-day range, and the thirty-day at -0.022 in the 16.7th, that convexity is decidedly on the call side at both tenors. Worth flagging a contradiction in the volatility data itself: the regime label attached to it reads fear-priced with put premium, but both skew values are negative, which under the stated sign convention means calls bid. The numbers are what should be read, and they say the downside is under-hedged going into payrolls.

The composition inverted, and the timing is the tell. On 08-28 longs lost 102.8 million dollars against 33.8 million for shorts, a 3.0:1 ratio, and the day's total sits at the 83rd percentile of the past year — a large day, not a rounding error. Through the advance the pattern was the exact opposite and far larger: shorts lost 811 million dollars on 08-19, 262 million on 08-20 and 470 million on 08-21, which is what a short squeeze looks like mechanically. Those three sessions produced +6.92%, +4.85% and +7.66% moves and account for most of the twelve-day gain. The advance was substantially a forced-covering event layered on top of genuine ETF demand; with the shorts now cleared out, that fuel is spent, and the first session without the demand leg immediately produced long-side pain instead.

The US spot premium is +3.21 basis points, comfortably inside the neutral band given that roughly plus or minus 10 basis points marks the practical extremes. The path is what matters and it is unambiguously constructive: from -10.66 basis points on 08-16, through -6.81 on 08-19 and -2.21 on 08-21, to positive readings on 08-24, 08-27 and 08-28. Over the fortnight US-side buying went from a persistent discount to a modest lead. The single most important observation is that it stayed positive on 08-28 — the day price fell 3.24%, the day long liquidations hit 102.8 million, and the day the ETF print went negative. The domestic bid did not leave when the flow number did. This is the cleanest piece of evidence against the bearish view in the entire dataset, and it is the main reason the call is medium confidence rather than high.

ALIGNED — the rates and dollar legs push the same direction as the flow and derivatives read, with one dissent that must be named. Aligned: hawkish repricing toward a September hike, DXY at a fortnight high of 99.677, the 10-year at 4.67%, and the ETF bid turning negative are all the same trade, because the ETF channel is precisely how dollar-funded allocation reaches BTC. The dissent is the equity tape — the S&P within 0.43% of its high with VIX at 14.43, and gold at an all-time 4,504 against 5.41% money supply growth, together say risk appetite and the debasement bid are both intact. That dissent is why confidence below is medium and not high: the macro is aligned on the transmission channel that matters most right now, but not on every channel.

The ETF flow channel is the entire story of the last twelve sessions, and it just turned. Nine consecutive inflow sessions from 08-17 totalling roughly 3.04 billion dollars (BlackRock's product alone accounting for about 2.02 billion, or 72%) coincided exactly with a +27.8% advance from 62,837 to 80,297. On 08-28 that flipped: 2,167 BTC of net outflow, a 30-day flow z-score of -1.30, and net absorption running 2,580 BTC BELOW miner issuance for the first time in the run. Today's news digest still describes the streak as intact through 08-27 — the local tape is one session fresher and it shows the break. Miners cannot fill the gap: the Puell Multiple at 0.865 sits in the 34th percentile, meaning miner revenue is below trend, and hash ribbons at 0.9885 have only just begun turning up from 0.9859 on 08-25. On-chain throughput is high without being valuable — transaction count at the 96th percentile against fees of 603 sats per transaction in the 12th percentile, which is a network running full with no congestion premium and therefore no fee-driven demand signal. The short-term-holder cohort is the pressure point: at an MVRV of 1.112 their aggregate cost basis is roughly 69,900, meaning the marginal buyer of this advance is sitting on an 11% gain and has a reason to take it.

Hawkish, and hawkish in a way that has moved in the last 48 hours. Fed funds is at 3.63% with the 10-year at 4.67%, and Warsh's Jackson Hole debut on 08-27 stressed commitment to bringing inflation down without signalling a move — the market response was to raise the implied probability of a September HIKE, not a cut, which is a live repricing of the direction of policy rather than its pace. Money supply growth at 5.41% year-over-year is the one accommodative element in the picture, and gold at an all-time 4,504 says that channel is being expressed elsewhere. Sentiment is running well ahead of the policy tape: the Fear and Greed reading is 73, firmly in Greed, on a day price fell 3.24%. Three labour prints land before the 09-15/16 FOMC — JOLTS on 09-01, ADP on 09-02, and August payrolls on 09-04 — with CPI on 09-11. The payrolls print is the highest-impact release in the two-week window and it sits inside this brief's horizon.

The 10-year is 4.67%, up from 4.63% on 08-16 with a local high of 4.74% on 08-25, and it firmed into the Jackson Hole repricing. At 4.67% against a 3.63% policy rate, the long end sits roughly 104 basis points above fed funds — a curve shape that is pricing neither imminent easing nor recession. Direction matters more than level here: yields rising while the dollar makes local highs is a tightening-of-financial-conditions impulse, and it is arriving in the same week as three labour prints. On credit: there is no credit-spread feed in this stack. No high-yield or investment-grade spread read is available and none should be inferred — the absence is a genuine gap in the macro picture, not something to fill with a guess.

Firming, and the local tape contradicts the headline narrative. DXY closed 99.677 on 08-28, up 0.52% in a single session from 99.159 and 0.95% off the 08-20 trough of 98.735 — that is the highest reading in the fourteen-day series and it printed on the day of the hawkish repricing. One of the macro wires in today's digest describes a weakening dollar with gold near 4,602; the local close is 99.677 and gold is 4,504, so that piece is describing a tape that has since turned. A dollar making local highs on rate-differential repricing is a direct headwind for an asset trading 16.55% above its own 50-day mean, and the mechanism is not subtle: the marginal buyer of the last twelve sessions was a US dollar-funded ETF allocator.

Risk-on and complacent, and this is the leg of the macro tape that argues AGAINST the directional view below. The S&P closed 7,711.76, only 0.43% below its 08-17 high of 7,745.06, and it headed for a winning week despite the hawkish Jackson Hole keynote. VIX at 14.43 is the lowest reading in the fourteen-day series. Equities have absorbed the September-hike repricing without flinching. If that holds through payrolls, any BTC drawdown is a positioning shakeout rather than a risk-regime break — the correlation channel that transmits equity stress to crypto is currently closed.

The nearest support is shallow and close: the 08-22 and 08-23 closes at 76,998 and 77,589 sit roughly 1% below spot and will be tested first. Below that the advance left an air pocket, and this is the material point — the 08-20 close was 72,661 and the 08-19 close 69,221, meaning almost no traded structure exists between roughly 72,700 and 77,000 because the market covered that ground in two sessions on 811 million and 262 million dollars of short liquidations. Price that ascends on forced covering tends to retrace through the same range quickly, because there is no accumulated position to defend it. A payrolls-driven risk-off would more likely fall into that gap than stop at the 1% shelf. Below 72,700 there is a genuine confluence: the short-term-holder aggregate cost basis near 69,900 and the 200-day at 69,262, roughly 10-11% below spot, form the zone where a pullback would stop being a shakeout and start being a trend problem — that is also where the cohort holding an 11% gain moves to breakeven and stops selling. The 50-day at 66,701, 14.2% below, is where the mechanical trend gate flips. The modal path over 5-7 days is a 3-6% pullback into 73,000-75,000; the tail case is a weak payrolls print into a market that is call-heavy and put-light, which reaches the 72,700 shelf quickly. The condition that converts modal into tail is specific and observable: two or more consecutive negative ETF sessions. One outflow is noise; a second confirms the channel has closed.

Moderate, with a specific dislocation worth trading around. The Deribit thirty-day implied index is 38.34, at only the 25.8th percentile of the past year — implied volatility is cheap by its own annual distribution. But thirty-day realized is 45.17% and ninety-day realized is 42.07%, so implied sits roughly seven vol points BELOW what the asset has actually been delivering. Meanwhile thirty-day ATM implied at 39.0% is in the 72nd percentile of the trailing ninety days and has FALLEN about one vol point over the last five sessions. Reading these together: volatility has repriced upward within the quarter but remains inexpensive against both the year's range and recent realized, and it softened into the rejection of the high. Seven-day realized at 35.42% is the calmest window in the set, which is the classic pre-event reading. Options are not overpaying for the payrolls week — if anything, protection is cheap relative to what the tape has recently produced.