Bitcoin 2026-08-03 daily brief — AI stance: bearish

Leverage demand is apathetic, not crowded. Weighted by open interest across venues, perpetual funding annualizes to 5.25% — under half the roughly 11% a year that the exchange-default 0.01% per 8 hours implies, which is the neutral line. The single reference venue is even quieter at 2.42% annualized, and the 2.83 percentage point gap between them says what little long demand exists is concentrated on the larger-open-interest venues while the reference venue is close to flat. Funding also cooled sharply into Monday: the reference venue fell from 0.0065% to 0.0022% per 8 hours and the open-interest-weighted measure from 0.0086% to 0.0048%, i.e. traders paid up to be long through late July and stopped ahead of payrolls. The framework bias is NEUTRAL and I agree with it. The tactical implication cuts against the usual bearish playbook: with no crowded long book, a break lower would have to be spot-driven rather than a liquidation cascade, which removes the sharp flush-and-reclaim that normally marks a local low. Expect a grind rather than a wick.

Light, unlevered and disengaged. Market-wide futures open interest stands at $48.91bn, rebuilding modestly off $47.75bn on August 1 but still below the $49.78bn of July 22 — so this is a local low being worked off, not a fresh positioning build. CME open interest is back near 2023 levels according to the desk commentary, which points at basis-trade and institutional disengagement rather than repositioning. Options open interest is $25.10bn after the July 31 monthly expiry rolled roughly $10.7bn off the book ($35.99bn on July 31 to $25.26bn on August 1), and options volume of $1.28bn is thin against the $4.83bn traded on July 22. Traders describing the drop from $65,000 as thin volume rather than panic selling matches this exactly. The structural read: forced-liquidation crash risk is low, and so is squeeze fuel. Whichever way this range resolves, spot flow has to do the work.

Nothing forced is happening. $27.7M of longs against $26.8M of shorts on August 3, a 0.97 ratio at the 43.6th percentile of the past year — two-sided and unremarkable. The contrast is instructive: July 27 saw $71.7M of long liquidations and July 31 another $71.3M, and those are precisely the two sessions that produced -2.5% and -3.0%. Those flushes have already happened and the leverage that fuelled them is gone, which is why funding now sits below the neutral baseline. The corollary is that the next leg down, if it comes, gets no help from cascading stops — and neither does the next leg up.

The clearest single bearish datapoint in the set. The US-versus-offshore spot spread sits at -11.09 basis points, past the ±10 bps practical extreme threshold, and it has been negative in every one of the last thirteen sessions — -6.0 bps on July 22 widening to -13.8 bps on July 30 and back to -11.1 bps now. That is not a snapshot, it is a fortnight-long regime: offshore venues are setting the price and the US channel trades at a persistent discount. It is also the second independent read on the same absence, because the same US channel is showing a 7-day ETF flow average of -557.7 BTC per day and a 30-day average in the bottom decile of its history. Two different instruments, same conclusion: the US institutional bid that drove 2024-25 is not in the market this week.

CONFLICT — and it is the central fact of this brief. The macro tape is as friendly as it has been in months: equities at 7,600.5, VIX 15.86, Brent below $90 after the Iran de-escalation, the dollar having touched 99.71. Bitcoin's own channels point the other way: 22.73% below its 90-day high, ETF flows in the bottom decile on a 30-day basis, the US spot premium at -11.09 basis points for thirteen straight sessions, and its largest corporate accumulator selling. When those disagree I side with the asset-specific channels on a one-week view, because they describe who is actually transacting in BTC right now, while the macro tape describes a bid that is demonstrably going somewhere else.

Four asset-specific negatives and one genuine positive. First, the US spot channel is a net supplier: the 7-day ETF flow average is -557.7 BTC per day at the 25th percentile of its history and the 30-day average sits in the bottom decile at the 8.8th; funds recorded $61.53M of net outflows in the week ending July 31, breaking a three-week inflow streak, with IBIT shedding $122.7M and Fidelity $54.8M, and 54% of all 2026 sessions have been net-outflow. Monday's +923.7 BTC single day, absorbing roughly twice daily miner issuance, is one session against that trend. Second, the largest corporate holder sold 1,638 BTC and issued 3 million shares into a cash reserve, extending a purchase pause running since June 22, with resumption gated on a preferred security recovering from about $90.60 toward $100 par — the structural bid that defined 2024-25 is now a supply source. Third, miner economics are stressed: Puell Multiple 0.771 at the 26th percentile, Hash Ribbons at 0.988 with the 30-day hash average below the 60-day, and fees at 553 sats per transaction in the 11th percentile despite transaction count at 703,730 — the 99.2nd percentile, near record throughput generating almost no fee revenue. Russia's mining ban takes effect August 15 across Moscow, the Moscow Region and parts of Kursk, hitting 65 data centres and 734 MW in a jurisdiction holding roughly 17.2% of global hash rate. Fourth, the regulatory catalyst is off the table inside this horizon: the CLARITY Act has no floor vote, no cloture motion and no allocated time before the August 7 recess deadline, pushing any action to mid-September, with Polymarket odds of 2026 enactment down near 28%. Fifth, and separate: the Coldcard firmware RNG failure has now drained roughly 1,367 BTC ($88.6M) from 4,585 addresses across three waves — trivial as flow against $48.9bn of futures open interest, but a real hit to self-custody confidence and a stolen-coin overhang. The positive: valuation is genuinely cheap. MVRV-Z at 0.677 (22nd percentile), NUPL at 0.170 (20th), Reserve Risk at the 2.3rd percentile, and aSOPR under 1 at 0.961 meaning the average coin that moved did so at a loss. That is what a floor looks like from the on-chain side.

Neutral, with the long end doing the tightening the policy rate is not. The funds rate sits at 3.63% while the 10-year yields 4.75%, so the curve is pricing term premium and growth rather than imminent easing, and M2 growing 5.53% year over year is expansionary at the same time. Headline inflation ran near 3.5% in June on the Middle East oil shock, which blocks a fast cut path — though Brent falling more than 10% below $90 after the August 3 decision to hold off on new strikes against Iran is the first genuine disinflationary impulse in months. The next FOMC is September 15-16; between now and then July payrolls land August 7 (FactSet near +87,500, Continuum +120k with unemployment rising to 4.3% from 4.2%, versus +57k in June) and July CPI on August 12. Sentiment gauge: Fear & Greed reads 25, Extreme Fear — but that print is dated July 18, sixteen days stale, and almost certainly describes the late-July dip rather than today's tape, which shows 30-day implied volatility at the 2.5th percentile of the year and balanced liquidations. Treat it as background, not as a live contrarian trigger.

The 10-year at 4.75% is rising, not falling: 4.63% on July 22, 4.68% on August 2, then a 7 basis point jump on August 3 alone, +12bp over twelve sessions. That move lands directly into the Treasury Quarterly Refunding Announcement on August 5, which sets the coupon auction schedule and therefore the duration supply for the quarter. There is no credit-spread feed in this dataset, so I will not pretend to a credit read — but a rising long end alongside record equities describes a growth-and-term-premium regime rather than a liquidity-expansion regime, and long-duration non-yielding assets are the wrong vehicle for that mix.

DXY at 100.01, down from 101.46 on July 24 and back up off the 99.71 printed on August 2 — the first sub-100 reading of this stretch, immediately withdrawn. A roughly 1.4% dollar decline over ten sessions is normally a clean tailwind for a dollar-denominated scarce asset. Bitcoin gained 0.28% over the same period. Under normal conditions this dollar path is bullish; the fact that it produced nothing is the more important observation.

Unambiguous risk-on. The S&P 500 closed at 7,600.5, up 1.48% from Friday July 31's 7,489.72 and up 3.89% from 7,316.15 on July 29, with VIX down to 15.86 from 20.66 on July 29. About 71% of the index has reported Q2 with another 15% due this week. Bitcoin over that same July 29 to August 3 stretch went 63,874 to 63,550, down 0.51%. Equities made new highs; BTC made a lower high.

Seven-day one-sigma is about 3.8% on 7-day realized volatility of 27.19%, or 4.3% on the 30-day figure of 30.87% — call it 61,000 to 66,100, which is essentially the entire 30-day range. That framing matters: the honest modal outcome here is a grind, not a crash. First downside stop is 62,780, the August 1 close; then 61,849, the 30-day low, 2.7% below spot. A close beneath 61,849 inside the window is roughly a one-sigma event, and given the flow picture I would put it modestly better than even odds against — somewhere in the 35-45% range rather than the coin flip the volatility alone implies. The 60-day low at 58,519 is 7.9% down, a two-sigma move, and realistically needs a payrolls shock on August 7 or a hot CPI on August 12 to get there inside seven sessions. Upside is capped at 66,257 on the same logic in reverse. The genuine tail risk sits on the volatility axis rather than the level axis: an implied-vol index at the 2.5th percentile of a year is the market underwriting continued quiet, and the last 30 days contained exactly one move beyond ±4% (+4.71% on July 14) with the largest downside day only -3.0%. That is the setup in which a 6-8% single session is mispriced, and if it arrives it is more likely to arrive on the downside given who is currently not bidding.

low