Bitcoin 2026-07-28 daily brief — AI stance: bearish

Perpetual funding is running at 8.08% annualized on a cross-exchange average and 9.43% weighted by open interest. Both are below the ~11% annualized level that the exchange-default rate implies, so longs are paying less than the neutral baseline to hold exposure — this is not a crowded long book, and anyone reading it as stretched is misreading the calibration. The 1.35 percentage-point gap between the two views matters, though: it is well past the point where positioning can be called evenly distributed, and it means the long lean is concentrated on the venues carrying the largest open interest rather than spread across the market. If a forced-selling episode starts, it starts on the big books. The one detail that leans bearish is the path: funding at 0.007379% per eight hours sits near the top of its fourteen-day range — which spans a low of -0.00014% on 07-25 — and it got there on the same session that produced an eleven-day price low. Traders are adding leveraged length into a falling price. That is fuel below, not a floor.

Cleaner than a week ago and therefore less explosive, but also less supported. Futures open interest at $47.22bn is down 7.5% from the $51.06bn peak on 07-21, funding is under the neutral baseline on both measures, and two sessions of long liquidations have taken weak length out. The vulnerability in this market is not leverage — it is the absence of a bid, and deleveraging does nothing to fix that. On the options side, open interest of $33.91bn sits just below the fourteen-day peak of $35.09bn on 07-24, with the 07-31 monthly expiry clustered at $70,000 and $72,000 including a roughly $2.5bn call spread. At 63,687.66 that entire structure is 10-13% out of the money and expires worthless, removing a large block of upside positioning and whatever dealer hedging flow was tied to it. Thirty-day implied volatility at the 18.1st percentile of the past year with 25-delta skew of 4.60 vol points at only the 23rd percentile, and a term structure in contango at +2.90 points from 30 to 90 days, says the options market is priced for calm through a decision, an inflation print and an expiry. That is a cheap hedge and an underpriced tail.

Two consecutive long-dominant sessions after a short squeeze, which is how a range grinds lower. On 07-27, $71.68M of longs were liquidated against $21.67M of shorts — a 3.31x ratio and the largest single-day long liquidation in the fourteen-day window. On 07-28 it continued at $47.40M longs against $16.31M shorts, a ratio of 0.34. Set that against 07-20 and 07-21, when $52.23M and $48.09M of shorts were flushed into the 66,257 high. The market squeezed shorts at the top of the range and is now bleeding longs at the bottom of it. Crucially, total liquidation intensity sits at only the 50.7th percentile of the past year — despite the widely-quoted ~$700M of crypto derivatives liquidations across all assets, BTC-specific liquidation pressure is merely median. This is ordinary attrition, not a capitulation flush, which means the cleansing event has not happened.

The US-versus-offshore spot spread reads -9.79 bps, essentially at the -10 bps practical extreme that marks the bottom decile. More important than the snapshot is the persistence: it has been negative on all thirteen sessions with data, running from -6.83 bps on 07-16, through -4.22 bps on 07-23 at its shallowest, down to -12.33 bps on 07-26, and back to -9.79 bps now. There has not been a single session of US-side leadership in two weeks. Offshore is setting the price and the US institutional channel is a persistent net seller — which is the same fact the ETF flow data report through a different pipe, and two independent measurements of the same absence is what turns it from noise into a regime characteristic.

ALIGNED — both push bearish. The macro tape (a Fed whose only surprise is hawkish, DXY at 101.38 firming, the 10-year at 4.65% near range highs, gold 28% off its high) and the crypto-domestic read (30-day ETF flow at the 7th percentile, Coinbase premium negative for thirteen straight sessions, miners at the 16th-percentile Puell) point the same way. One genuine dissent belongs on the record: US equities are risk-on (S&P +0.21%, VIX 18.21) and the on-chain valuation panel is at accumulation percentiles with 4/8 bottom triggers and 0/8 top triggers. Those are the two facts that keep this a medium-confidence call rather than a high one.

The demand side is the whole story and it has three legs, all broken. Passive: the 30-day average ETF flow is -1,289 BTC per day, sitting at the 7th percentile of its own history — near the worst sustained stretch on record — with -780 BTC on 07-28 and the 7-day average now flipped negative at -81 BTC. ETFs are net sellers of roughly 1.97x daily miner issuance, so the vehicle built to absorb supply is currently adding to it. The 'third consecutive weekly inflow' headline for the week ending 07-24 is a mirage: $33.79M of net inflows for the week, against $465M of outflows across Thursday and Friday alone, of which roughly $415M was IBIT. Regional: the Coinbase premium is -9.79 bps, at the practical -10 bps decile threshold, and has been negative on every one of the thirteen days with data. Miners: Puell at 0.648 (16th percentile) and the hash ribbon ratio at 0.9901 (10th percentile) — the 30-day hash average is still below the 60-day, the recovery from 0.9739 on 07-16 stalled and ticked down on each of the last two days, and the cross back above 1.0 has not happened. The fee picture is the sharpest version of miner distress: 530 sats per transaction at the 11th percentile against 560,389 transactions at the 96th percentile — record throughput generating almost no fee revenue, so miner income is essentially pure subsidy at a depressed price. Structure and policy offer nothing: the Digital Asset Market Clarity Act was shelved before the Senate's summer recess with no floor time until at least September, and BitMart's wind-down on 07-26 is the third centralized-exchange closure in July after AscendEX and BitMEX. Cohort levels frame the range: the short-term-holder aggregate cost basis is near $67,870 (price 6.2% below it, so recent buyers are underwater and every rally meets loss-averse supply), while the long-term-holder cost basis is near $49,977, leaving that cohort 27% in profit with no forced-selling pressure.

Hawkish, and hawkish in the specific way that hurts most: the tail is a hike, not a cut. The FOMC decision on 2026-07-29 carries roughly 65-66% odds on a hold at 3.50%-3.75% and 34-36% on a 25bp HIKE, with nothing priced for a cut. Fed funds stand at 3.63% with core inflation running near 4.2% year-over-year, so the policy rate is barely at the inflation rate in nominal terms. M2 growth of 5.53% y/y is the one mildly supportive line, but it has not translated into a bid for duration or for crypto. This is a non-SEP meeting — no dot plot — so the whole information content is the statement and the press conference, and the June PCE print lands the very next morning, 2026-07-30. The live sentiment gauge reads 25 (Extreme Fear), though that stamp is 2026-07-18 and is therefore eleven days stale — it predates both the 66,257 high on 07-21 and the slide to the current 11-day low, so treat it as directionally consistent rather than current.

The 10-year sits at 4.65%, up from 4.55% on 07-16, having peaked at 4.71% on 07-24 and 07-25 before easing 6bp into the FOMC. That is near the top of its recent range with core inflation around 4.2%, so real yields are compressed but nominal financing costs are high and rising — the opposite of the condition under which long-duration, zero-cashflow assets re-rate upward. The Treasury quarterly refunding announcement on 2026-08-05 falls inside the following week and is a live risk to the long end via duration and size mix. There is no credit-spread feed in this dataset, so I am not making a credit read — I can see the level and direction of the 10-year and nothing about whether corporate risk premia are widening alongside it.

DXY at 101.38, up 0.62% from 100.75 on 07-17 and grinding higher into the decision. This is a firming, not a wrecking ball — a 0.6% move over eleven sessions is a headwind, not a shock. What makes it matter at this stage is the company it keeps: gold at $4,036 is defending $4,000 but sits roughly 28% below its January 2026 high, so the debasement/hard-money bid that would normally cushion BTC against a firm dollar is itself in drawdown. A rising dollar with gold broken is a worse configuration for BTC than a rising dollar with gold making highs.

Risk-on at the US index level with a violent rotation underneath, and BTC did not participate in the resilience. The S&P closed at 7,428.78 on 07-28, up 0.21% on the day, while VIX fell to 18.21 from 18.67 — and this was the session that contained the KOSPI's 10.84% collapse to 6,023.66, Samsung down 13.4% and SK Hynix down 14.5% on the Chinese immersion-DUV lithography report, their worst days in nearly two decades. US equities absorbed an AI-supply-chain shock without breaking. Over that same 07-27 to 07-28 stretch BTC made an 11-day low at 63,687.66. BTC underperformed a tape that was actively shrugging off bad news, which is the cross-asset evidence that its weakness is its own, not imported risk aversion. The S&P is only 1.4% below its 07-16 level of 7,533.77; BTC is 22.56% below its 90-day high.

Downside is layered and the first two shelves are close together. The 50-day at 63,276 is 0.65% below and is the only near-term technical support; beneath it there is nothing structural until the round 60,000 and then the 30-day low at 58,519, which is 8.1% down and is the level defining the current range. There is no on-chain support in that gap — the short-term-holder cost basis near $67,870 is overhead, not a floor. The next genuine on-chain shelf is the long-term-holder aggregate cost basis near $49,977, 21.5% below spot, which is roughly where the $52,000 target from the Nansen demand analysis sits; that is a deep-bear scenario, not a 5-7 day base case. My judgment on the distribution over the horizon, offered as judgment and not as a model output: the largest single branch, roughly half the weight, is a 58,500-65,500 grind that establishes another lower high; about three in ten is a break of 63,276 that tests 60,000 and challenges 58,519; about two in ten is a relief squeeze into 65,400-66,300 that I would fade unless the 7-day average ETF flow turns positive from -81 BTC alongside it. The genuine tail is that a daily close below 58,519 breaks a thirty-day range into a market with no passive bid and implied volatility that will already have repriced — that is the path where the low-50s arrive quickly rather than gradually. Offsetting all of it: long-term holders are 27% in profit and funding is below neutral at 8.08% annualized, so there is no forced-seller cohort and no leverage overhang large enough to manufacture a cascade on its own. Absent demand grinds; it does not usually crash.

low — with the explicit caveat that this is a coiled spring, not comfort. The Deribit 30-day implied volatility index sits at 37.24, the 18.1st percentile of the past year; 30-day at-the-money implied is 35.29% at the 21st percentile of the last ninety days. Realized volatility has compressed to 25.76% over 7 days against 34.06% over 30 days and 35.64% over 90 days. Implied sits above short-dated realized, so options are not mispriced against the recent past — they are cheap against the calendar. A decision on 07-29, an inflation print on 07-30, and a monthly expiry on 07-31 are being priced at bottom-quintile volatility, and 25-delta skew at the 23rd percentile says even the put side is not being paid up for. The largest single-day move of the last thirty days is only 4.71%; a market that has forgotten how to move is the one most exposed to remembering.

low