Bitcoin 2026-08-17 daily brief — AI stance: neutral

Carry has gone to essentially zero. Weighted across venues by open interest, funding annualizes to 1.07%, against a neutral baseline of roughly 11%/yr for the exchange-default rate — so longs are paying about a tenth of the standard cost of carry. The single-venue average annualizes to 0.98%, and the 0.09pp gap between the two means there is no venue-divergence story worth telling; the whole market is at the same near-zero level. The path matters more than the level: funding ran between roughly 6%/yr and 10%/yr annualized through August 8-16 — at or just below neutral the entire time — then collapsed to ~1%/yr on the day price rose 2.21%. At no point in the last fortnight has leverage been crowded long, and a rally that happens while carry falls toward zero is being paid for in spot and short-covering, not by leveraged buyers chasing. The practical consequence: there is no long overhang whose unwind could accelerate a decline.

Flat and unlevered. Market-wide futures open interest is $49.18B, up 3.7% from $47.43B the prior day but effectively unchanged from $49.50B on August 5 after dipping to $46.46B on August 10 — a round trip, not a build. Options open interest of $25.75B is down from $27.20B on August 7, and options volume of $878M is the second-lowest of the fortnight against a $2.9B high, so the derivatives complex is disengaging rather than positioning. The term structure is in contango with the 30-to-90-day slope at +0.061, which is the normal shape and consistent with no near-term event being priced. One venue-specific note for reference: single-venue open interest of $1.49B also rose about 2.5% on the day, in line with the aggregate.

Two-sided flush, completed. Longs were liquidated for $35.1M on August 10, $29.9M on August 13 and $21.1M on August 14 as price fell from 65,030 to 62,830; then shorts were liquidated for $45.0M on August 17 against just $3.5M of longs — a 12.9:1 ratio — on the 2.21% advance. Yesterday's move was therefore a short squeeze, not fresh demand. But total liquidation activity is only at the 37th percentile of the past year, so neither flush was a capitulation. With both sides of the book cleared inside 14 days, positioning no longer supplies a directional edge to either camp.

The clearest bearish input in the dataset, and the most persistent. The Coinbase-versus-offshore spot spread is -10.07 bps, and the 14-day series has run between -8.7 and -10.9 bps every single session from August 5 through August 17 — at or through the practical extreme threshold of ±10 bps for two straight weeks, with no day of US-side leadership. That is not a snapshot to discount; it is a fortnight of offshore bids clearing the market while US buyers stand aside, and it corroborates the ETF trend of -336.7 BTC/day from an entirely independent measure. It also sits awkwardly against the +2,137 BTC ETF print stamped August 17, when the premium was -10.07 bps — one of those two readings will look different once the pending August 18 Farside update lands, and I am flagging the inconsistency rather than picking a winner. Until the premium moves back toward zero, rallies here should be read as offshore and leverage-driven rather than institution-led.

CONFLICT. Macro risk appetite is on — S&P near records at 7,745, VIX 15.19, dollar flat at 99.60 — while the BTC-specific bid is off: US premium between -8.7 and -10.9 bps for 14 straight sessions, ETF trend at -336.7 BTC/day, and the largest corporate holder in cash-accumulation mode. I side with the BTC-specific read on a 5-7 day view, because with VIX at 15.19 there is no macro impulse available to transmit either way. But that side argues for drift and range rather than a trend, because the same tape offers no downside fuel: valuation metrics are lower-quartile (MVRV-Z 23rd percentile, reserve risk 2.4th) and carry is effectively zero. One tell on where the monetary-hedge flow is going instead: gold at $4,459.50 near records with a flat dollar and M2 up 5.53%, while BTC sits 17% below its 90-day high. That trade is being expressed in gold, not bitcoin.

Four strands, all pointing the same way. (1) Spot ETF demand is negative on trend: the 7-day average is -336.7 BTC/day and the week of August 10 was the largest six-week outflow at -$389.7M, which cross-checks against the daily series (-2,261, +123, -964, -2,065, -894 BTC = -6,061 BTC over August 10-14, roughly $385M at $63-64k). August 17 printed +2,137 BTC (64th percentile, 4.33x miner issuance, excess absorption +1,644 BTC) — one day against five, and it sits in tension with a Coinbase premium of -10.07 bps that same session; the August 18 Farside update is still pending in the calendar, so same-day as-of alignment between the flow print and the premium is not established. I am not resolving that tension, only naming it. (2) The corporate bid has gone defensive: Strategy sold 3,458,866 MSTR shares for about $333.7M between August 10-16 and bought zero bitcoin, lifting its USD reserve to $4.8B. (3) Miner economics are squeezed from both ends — Hash Ribbons at 0.9882 (9.8th percentile, 30-day hashrate under 60-day), Puell 0.8395 (31.9th percentile), and fees at 366 sats/tx (10th percentile) despite transaction count at 756,022 (99.65th percentile). Record throughput with bottom-decile fee revenue means no cushion, and HIVE's $350M five-year GPU-cloud deal lifting contracted ARR to ~$180M is the sector's answer: revenue away from hashing. (4) Policy catalysts have been pushed out — the SEC delayed its tokenization 'innovation exemption' and pulled the August 14 meeting on its ~400-page Regulation Crypto proposal, and the CLARITY Act stalled in the Senate with Congress in recess. The only near-term venue is the Jackson Hole symposium, August 27-29, whose 2026 theme is 'Financial Innovation: Implications for Payments and Policy' — outside this horizon.

Hawkish, and hawkish in the way that matters most for a long-duration asset: the live debate is a September hike versus a hold, not a cut. Fed funds sits at 3.63% against core PCE of 3.3% year-over-year (June), so the real policy rate is only about +0.33pp — not actually restrictive, which is precisely why the hiking option stays open. Chair Warsh has withdrawn forward guidance, leaving the September meeting genuinely two-sided, and FOMC minutes land 2:00 PM ET on August 19, inside this horizon. Offsetting that: M2 is growing 5.53% year-over-year and the growth data is deteriorating — University of Michigan preliminary August sentiment fell to 51 with softer retail spending, while July ISM manufacturing was 55.6. That mix is stagflationary rather than easing. On sentiment: the Fear & Greed reading of 25 ('Extreme Fear') is dated 2026-07-18 and is a month stale — do not treat it as live. The current gauge is the implied-vol complex, and at a 3.3rd-percentile Deribit 30-day index the tape is complacent, not fearful.

The 10-year is 4.68%, up from 4.63% across August 14-16, and the 30-year reached its highest level since June 2007 on August 17 as oil advanced. That is long-end term-premium expansion, which historically bites hardest on assets with no cash flows. There is no credit-spread feed in this context, so I will not characterize credit conditions — the rates read is a Treasury-curve read only.

DXY 99.598, and flat — the entire 14-day range is 99.50 to 100.02, under 0.6%. There is no dollar impulse in either direction, and that is the informative part: BTC fell from 66,257 to 62,780 over the last 30 days without any dollar tailwind against it, so the weakness is idiosyncratic rather than macro-imposed. A neutral dollar removes the usual excuse for crypto underperformance.

Risk-on, and BTC is not participating. The S&P 500 at 7,745 is within 0.7% of its 14-day high of 7,798.99 and near record levels, with VIX at 15.19. BTC is 17.19% below its own 90-day high over the same window. That divergence is the single most important cross-asset fact here: BTC is not currently trading as high-beta equity — if it were, it would be near highs — it is trading on a broken internal bid.

Downside is layered, not a single number. The first line is the 50-day at 63,685, only 0.87% away, and a close below it puts 62,780 (the 30-day low, and the 62,830 close of August 14) in play — a 2.3% move that is well inside a normal week at 20.12% realized vol, so treat it as more likely than not to be touched at some point. Below that the structure thins out considerably: there is no meaningful shelf between 62,780 and the 60-day low of 58,519, a further 6.8% down, which is where the genuine risk sits — an 8.9% total decline from here. What argues against reaching it in this window is the absence of forced sellers: funding at 1.07% annualized means no leveraged longs to cascade, total liquidations sit at only the 37th percentile of the year, and 65.80% of UTXOs remain in profit. What argues for it is that long-term holders are already realizing losses (long-term-holder SOPR 0.807, 17.8th percentile) and the US bid has been absent for 14 sessions — distribution into a thin book does not need leverage to grind lower. On the upside, 66,257 caps the range 3.05% away and the 200-day at 69,158 is 7.11% away; reclaiming the latter is not a 5-7 day proposition at this volatility.

low

Deribit 30-day implied index 34.72 at the 3.3rd percentile of a year and 30-day ATM IV at the 4.4th percentile of 90 days, against 7-day realized 20.12% and 30-day realized 22.65% — low on both implied and realized. The ~12-point gap between implied 34.72 and 30-day realized 22.65 is a normal variance premium, not a stress signal.