Bitcoin 2026-08-25 daily brief — AI stance: bearish

Cool, and that is the most important fact in the derivatives complex. Weighted across venues by open interest, perpetual funding annualizes to 6.10% — roughly 56% of the ~11% a year that the standard 0.01%-per-8-hour default represents. Longs are paying below neutral carry after a 24% advance to a 90-day high. That is not a crowded book; it is a market that rallied without adding leverage. The largest single venue is cooler still at 5.18% annualized, 0.92 points under the all-venue weighted figure, which tells us the modest long crowding that exists sits on the other large-open-interest venues rather than that one — a distribution point, not a spread between two market-wide measures. The path matters as much as the level: the crowd did briefly pay up, with the weighted rate reaching 0.0114% per 8 hours on Aug 22, just above the default, and it has since cooled by roughly half while price held its highs. Longs stepped back at the top rather than piling in. The practical consequence is that the bearish case here has to be delivered by spot supply and event risk, because there is no liquidation cascade available to deliver it.

Deleveraged in linear risk, long convexity. Market-wide futures open interest across all venues is $55.86B, up 19.4% from $46.78B on Aug 13 in dollar terms — but bitcoin appreciated 24.3% over that same window, so coin-denominated open interest FELL from about 737,000 BTC to about 708,000 BTC, a 3.9% contraction, and it also declined 2.5% from Aug 24's $57.30B. Futures exposure shrank through a 24% rally, which is the arithmetic fingerprint of short covering rather than new directional risk, and it corroborates Bloomberg's Aug 20 read that the move left the rally hunting for real buyers. Where risk was genuinely added is options: notional open interest rose 50.6% to $40.51B, about +21% in coin terms, with $7.03B of volume on Aug 25 against roughly $2.6-2.7B on Aug 23-24. The market bought convexity into the Aug 28 monthly expiry instead of levering spot. That is a structurally healthier configuration than a funding-fuelled top — and it is also one whose supportive hedging flow has a known expiry date inside this horizon.

Still violent on both sides, but the fuel is switching. Total liquidations sit at the 94th percentile of the past year, with $175.1M of shorts and $87.8M of longs stopped out on Aug 25, a 1.99 ratio still favouring short pain. The trend inside that is decay: on the same venue feed, short liquidations ran $811.1M on Aug 19 and $470.1M on Aug 21, so Aug 25's $175.1M is roughly a fifth of the Aug 19 figure. Separately and market-wide, Bloomberg reports roughly $2.7 billion of crypto shorts liquidated on Aug 19, the second-largest such event on record. Meanwhile long liquidations doubled from $44.7M on Aug 24 to $87.8M on Aug 25. Fewer shorts remain to squeeze, and a growing population of late longs now sits above recent support.

The US-versus-offshore spot spread is -1.41 bps, essentially flat and well inside the ±10 bps that marks a decile extreme, so the snapshot itself is uninformative. The fourteen-day trend is not. From Aug 13 to Aug 21 the series closed a persistent US discount in a clean monotone: -10.76, -10.12, -10.85, -10.66, -10.07, -8.50, -6.81, -5.01, -2.21 bps. That is offshore de-risking ending, and it was a necessary condition for the rally. It reached +0.32 bps on Aug 24 — the only positive print in the window — and has since slipped back to -1.41. So the US bid repaired the discount but never established leadership, and it faded at the high. For an advance whose durability depends on institutional spot absorption, and on the evidence that fund absorption just fell below issuance, that is the wrong sequence in the wrong order.

CONFLICT — and the macro side is itself split, which is why this needs naming rather than averaging. The liquidity channel supports bitcoin: DXY 98.94, M2 +5.34% year-over-year, enlarged Treasury buybacks from Sept 9. The rates-and-growth channel does not: the 30-year at 5.196% near two-decade highs, preliminary August UMich sentiment at 51.0 from 55.2 in July, ADP private payrolls at 44,000 in July, Q2 GDP at 1.5% annualized. Against both, the local trend read is unambiguously long — 90-day high, above both moving averages, funding below its neutral baseline. Over the next five to seven sessions I side with the rates-and-growth channel, for one reason: it has a dated catalyst inside the window on Aug 26 and Aug 28, and the liquidity channel's next dated catalyst is Sept 9, outside it.

The demand engine decelerated below the supply line. On Aug 25 the fund complex absorbed 379 BTC against roughly 444 BTC of new miner supply — the first session of this streak in which spot funds bought less bitcoin than was mined. It remains the seventh consecutive positive session, but the figure fell 91% from 4,288 BTC on Aug 24 and the 30-day flow z-score flipped from +1.14 to -0.35. The fragility underneath is concentration: per Farside, Aug 24's $337.6M was 62% BlackRock's IBIT, with IBIT running 54% to 83% of daily inflows across the six-session run — a streak dependent on one issuer's creation desk. Miner economics are still deteriorating: the hash-ribbon ratio has fallen in every one of the last 13 sessions to 0.9859 (9th percentile) with Puell at 0.939, so the hash-rate rollover that registers as a cycle-bottom trigger is simultaneously a near-term forced-selling headwind. The regulatory tailwind is real but undated and therefore not a 5-7 day catalyst: the SEC's Aug 18 'Regulation Crypto Assets' proposal, and the Aug 19 White House push for the CLARITY Act, which passed the House in 2025 and is still stalled in the Senate.

Neutral in level, with hawkish event risk concentrated in a single 48-hour window. Fed funds at 3.63% against a 10-year at 4.70% is a +107bp positive slope: the long end is pricing term premium and duration supply, not further tightening, and M2 growing 5.34% year-over-year is outright expansionary. The asymmetry is in the calendar, not the setting. Kevin Warsh delivers his first Jackson Hole keynote as Chair on Aug 28 at 10:00 ET — the last set-piece communication before the Sept 15-16 FOMC — and July PCE, the Fed's preferred gauge, lands Aug 26 at 8:30 ET. Warsh carries a public reputation as a hard-money hawk and a critic of balance-sheet expansion; with M2 running +5.34%, the tail he can deliver is one-sided. Sentiment is offside for that tail: the Fear and Greed index at 74 is squarely in Greed, which is the worst starting position going into a speaker whose surprise distribution is skewed hawkish.

The 10-year at 4.70% has round-tripped its relief. Treasury announced on Aug 19 that it will at least double buyback operations to $4 billion in the 10-20 and 20-30 year sectors, which took the 10-year down 6bp to 4.647% and the 30-year down 9bp to 5.196% on the day; the 10-year printed 4.74% on Aug 24 and sits at 4.70%, above the 4.63% of Aug 14-16. So the fiscal fix bought roughly three sessions. The 30-year at 5.196% is at levels not seen in nearly two decades. There is no credit-spread feed in this dataset — I have no high-yield or investment-grade spread read and will not infer one from equity or vol proxies. The rates channel here is term premium and duration supply, not credit stress, and the enlarged buyback operations do not begin until Sept 9, outside this horizon.

A genuine tailwind that has stopped improving. DXY at 98.94 is down 0.96% from 99.90 on 2026-08-13 and holds below 99, which historically loosens the constraint on dollar-priced hard assets. But the last four readings are 98.84, 98.82, 99.03 and 98.94 — the dollar's decline flattened out over exactly the sessions in which the bitcoin advance also flattened, from 78,447 on Aug 21 to 78,879 on Aug 25. Gold at $4,715.90 shows the debasement trade is being expressed, and expressed elsewhere, more persistently and with less leverage attached.

Risk-off at the margin without any stress premium, which is the least-priced configuration. The S&P at 7,677.28 is down 1.56% from 7,798.99 on 2026-08-13, closing Aug 25 down 0.3% with the Nasdaq off 0.8% on a technology selloff, while VIX sits at 15.45. Equities are drifting lower on a quiet tape with US-Iran tensions supporting oil and global yields in the background. A cross-asset risk event into a 15.45 VIX repricies fast, and bitcoin's beta to that repricing is not zero.

Two downside paths with different shapes, and the absence of leverage argues strongly for the slower one. The grinding fade is the base case: with weighted funding at 6.10% annualized and coin-denominated futures open interest already 3.9% smaller than it was twelve sessions ago, there is no crowded long book to cascade, so the likely path is drift — a retest of the Aug 22 shelf at 76,998 (-2.4%), then the Aug 20 breakout base at 72,660 (-7.9%), which is where the impulse actually began and where a normal retest terminates. Below that, the 200-day at 69,135 (-12.4%) is the structural line, and it matters more than usual because the 50-day at 65,745 sits beneath it: the 65,745-to-69,135 band is territory price crossed in a straight line on Aug 19-21 and would re-enter with no intermediate structure to slow it. The fast path is event-driven and narrower: a hawkish Warsh keynote on Aug 28 into a Greed reading of 74, with 25-delta skew at the 7th percentile of its month meaning the crowd owns calls and almost nobody owns protection, is the configuration where a single session moves 6-8% — the mirror image of Aug 19's +6.92%. What I do not think is available is a 2022-style liquidation spiral: the positions that would fuel one were closed on the way up, which is precisely why this is a medium-confidence fade rather than a high-confidence one.

High on realized, moderate on implied, and the gap between them is the tradeable observation. Realized 7-day volatility is 69.44% annualized against 44.17% at 30 days and 41.41% at 90 days — 1.57x its own monthly baseline — built on three sessions above 4.8% in the past week (+7.66% Aug 21, +6.92% Aug 19, +4.85% Aug 20), with liquidations at the 94th percentile of the year. Implied has not followed: the Deribit 30-day index at 43.28 is only the 54th percentile of its own trailing year, and 30-day at-the-money implied at 39.44% sits roughly 30 vol points below trailing realized, even though that 39.44% is the 77th percentile of its own 90-day range. Options are cheap relative to what the market just did, which encodes an expectation that this vol burst decays quickly — an expectation that PCE, a first Jackson Hole keynote from a new Chair, and monthly settlement inside 72 hours are each capable of disappointing.

Medium rather than high for three specific reasons: the flow break is a single session, funding below neutral removes the mechanism that would make a decline fast, and fading a fresh 90-day-high breakout fights the momentum base rate.