Bitcoin 2026-09-28 daily brief — AI stance: bearish

Open-interest-weighted funding across exchanges is 0.004529% per 8 hours, about 4.96% annualized, below the roughly 11%/yr exchange-default neutral baseline. That is soft, not crowded: longs are paying little, and CoinDesk on 2026-09-28 reports some venues flipped to shorts paying. The single major venue reads 8.2% annualized, 3.25 percentage points above the aggregate, so what modest long lean exists is concentrated at that one venue rather than market-wide. Over the 14-day window the OI-weighted rate ranged 0.00069% to 0.00935% per 8h (about 0.8% to 10.2% annualized) and never reached the baseline even on the 2026-09-21 squeeze day. Reading: leverage is not the fuel for a downside cascade, and the short lean is the main hazard to a bearish call.

Deleveraged and mildly short-leaning. Aggregate futures open interest is 53.84B, down 12.5% from the 61.56B peak on 2026-09-21 and back near the 2026-09-15 level of 51.81B; CoinDesk (2026-09-28) puts BTC-denominated OI near 652k BTC against an early-2026 peak near 800k. Options OI reset to 35.03B from 52.23B after the 2026-09-25 expiry (about $16B notional, 37% of the venue book), so dealer gamma around 83k to 85k is thinner into PCE and payrolls; options volume of 1.95B is the second-lowest of the window. The single venue in the feed shows the same pattern at its own scale, 6.04B from 6.70B on 2026-09-22 (-9.9%). Net: a market with little forced-selling fuel below and real squeeze fuel above, which caps bearish confidence at medium even though flows and macro point down.

The character flipped after the spike. 2026-09-18 (178M shorts vs 8M longs) and 2026-09-21 (392M shorts vs 56M longs) were the squeeze days that produced the +5.73% and +6.47% candles. Since 2026-09-22 longs have been the larger liquidated side on five of seven sessions: 88M on 2026-09-23, 51M on 2026-09-24, 32M on 2026-09-25 and 65M on 2026-09-28, against 18M to 29M of shorts each day. On 2026-09-28 the ratio was 0.45 (more longs liquidated) at the 68th percentile of 1y totals. Late longs from the 86k spike are being flushed on 1% dips, which is how a market grinds lower without a vol event; totals are not at capitulation scale.

Coinbase premium -1.06 bps: neutral on the plus-or-minus 10 bps extreme scale, but the 14-day pattern carries the information. It was negative on 12 of 14 days, ranging from -9.46 bps on 2026-09-18 to +1.24 bps on 2026-09-22, with a mean near -2.9 bps, and the only two positive prints landed the day after each short squeeze (2026-09-19 and 2026-09-22). Offshore has led both the rally and the stall; US spot has not been the marginal buyer even in the $2.4B ETF week, consistent with the ETF bid being executed via arbitrage rather than a US spot premium. That removes one bullish tell (a positive, widening premium) that would otherwise argue against the bearish tilt.

CONFLICT. The macro tape (10-year 5.18% at multiyear highs, DXY 101.2 rising, Brent $108, hike pricing into PCE and payrolls) pushes down. The local structural read (above both MAs, 7d and 30d ETF MAs at the 90th and 80th percentiles, sub-neutral funding, NEUTRAL cycle with two bottom triggers) is constructive. The brief sides with macro for the 5-7 day window because the local bullish inputs are lagging measures of a rally whose daily fuel has visibly run out.

ETF flows: the daily print decayed across six prints from +11,536 BTC (2026-09-21) to +8,293, +4,111, +2,260, +1,600 and then -285 BTC on 2026-09-28, the first net outflow since 2026-09-16; flow-to-issuance ratio -0.55 and excess absorption -801 BTC mean ETF demand did not cover miner supply on the day. The 7d MA of 4,696 BTC (90th percentile) and 30d MA of 2,408 BTC (80th percentile) are trailing measures of the $2.4B week The Block reported on 2026-09-26, which turned YTD flows positive. Miner economics: Puell 1.20 (56th percentile) and the hash ribbons ratio at 1.008, rising for 10 straight days since 2026-09-19 and above 1.0 since 2026-09-20, a miner recovery. Network: transaction count 780,095 at the 99.7th percentile with fee per transaction 439 sats at the 11th percentile, heavy count with no fee pressure. Regulatory: the Clarity Act is dead for 2026 after the 49-50 Senate cloture failure on 2026-09-15. Exchange risk: the Bitget $388M exploit of 2026-09-24 is contained, with withdrawals resuming and the protection fund absorbing the loss. Corporate bid: Strive added 1,107 BTC for $94.5M. Halving-cycle position: roughly 29 months past the April 2024 halving with MVRV-Z 0.98 (41st percentile) and no top triggers, mid-cycle rather than late-cycle.

hawkish. The macro feed puts fed funds at 3.63% (the calendar digest text states a 3.75% to 4.00% target range; the two are not reconciled here and the feed value is used). The 10-year at 5.18% sits about 155bp above that policy rate and is up 22bp since 2026-09-24; CNBC on 2026-09-28 reports mounting bets on a further Fed hike, and the digest describes the 2026-09-30 PCE print landing with markets pricing tightening rather than cuts. M2 is growing 5.66% YoY, so money supply expands while the bond market tightens, an inflation-worry configuration reinforced by Brent above $108. Fear & Greed 74 (Greed) says crypto sentiment has not absorbed that yet. In-window prints: JOLTS and Conference Board confidence 2026-09-29, PCE and ADP 2026-09-30, ISM manufacturing 2026-10-01, payrolls 2026-10-02, ISM services 2026-10-05 at the window edge. The FOMC minutes on 2026-10-07 and the 2026-10-27/28 FOMC meeting fall outside the 5-7 day horizon and are later context.

10-year 5.18% and rising: +24bp from the 4.94% low of 2026-09-19 to 21 and +22bp from 4.96% on 2026-09-24, flat at 5.18% for the last three prints. The 2026-09-28 tape marks 2007 highs for the 10-year and 2004 highs for the 30-year, driven by inflation, debt supply and the oil jump. The gap to the feed policy rate (3.63%) is about 155bp, which reads as the bond market pricing inflation and supply risk rather than growth. There is no credit-spread feed in this context, so no credit read is offered.

DXY 101.195, up from 99.683 on 2026-09-15 (+1.5% in two weeks), higher on 8 of 11 supplied day-over-day changes and above 101 on each of the last five prints. A firming dollar alongside a 5.18% 10-year is the classic headwind for a dollar-priced, non-yielding asset. BTC rose 10.4% through the same two weeks on idiosyncratic ETF and short-squeeze flows, so the usual inverse relationship broke temporarily. With the flow impulse fading, the working assumption is that the dollar reasserts as a drag this week; a DXY push above 101.5 on a hot PCE is the trigger to watch.

Risk-on but softening. S&P 500 at 7,683.69 is 1.0% below its 2026-09-21 close of 7,764.70 and back near the 2026-09-17/18 levels of 7,638 to 7,651; VIX 16.07, up from 14.21 on 2026-09-22 but still low in absolute terms. This is not yet a risk-off regime, but the first down-week in the supplied window coincides with the yield break to 5.18%. Nasdaq-100 data and a BTC-equity correlation series are not supplied here.

Downside ladder from 83,479: 81,200 to 81,300 (2026-09-19/20 closing shelf, about -2.7%); 80,863 (straddle floor through the Oct 2 payrolls close, -3.1%); 79,266 (Oct 9 straddle floor, -5.0%); 76,475 (50-day MA, -8.4%, also the 2026-09-15 to 17 base at 75,590 to 76,371); roughly 73,200 (short-term-holder cost basis implied by MVRV STH 1.14, -12.3%); 71,135 (200-day MA, -14.8%). Probability framing from the options book: by construction the straddle breakeven band is exited roughly 40% of the time in either direction, so about a 40% chance the Oct 2 close sits outside 80,863 to 85,137; the bearish tilt puts perhaps 55 to 60 of that 40 in the lower tail, not more. A test of the 50-day inside seven days is about a 1.8-sigma move on the Oct 9 implied daily sigma of 1.76%, a low-single-digit-percent event on the options math. The deep levels become live only if a hot PCE and a hot payrolls print push the 10-year through 5.25% while the ETF daily print stays negative; absent both, the base case is a test of 81,000 to 82,000 rather than a break.

low implied, compressed realized, high transition risk. DVOL 35.93 sits at the 8.2th percentile of its 1y range; ATM 30d IV 34.8% is at the 42nd percentile of 90 days and fell 1.5 vol points over five days. Realized vol is 16.35% (7d) against 41.58% (30d) and 38.66% (90d), the deepest short-window compression in the supplied history and unstable after a 43.5% quarter. Straddles charge plus-or-minus 1.4% to the Sept 30 expiry (81,838 to 84,162), plus-or-minus 2.6% through the Oct 2 payrolls close (80,863 to 85,137) and plus-or-minus 4.5% to Oct 9 (79,266 to 86,734). The front end is inverted (7d IV 2.65 points above 30d), so the market itself prices this week as rougher than the month. Those ranges say how far price is expected to travel, never which way.