Bitcoin leaves its September range with funding still cheap, and no flow data explains the move

Bitcoin is $85,793.94 on Binance spot at 12:19 PM ET on Mon Sep 21, up 5.7% on the day, 9.7% on the week and 11.3% on the month, a departure from the $76,500 to $78,300 band that contained the tape through the week of Sep 12. Ether is $2,762.85, up 4.4% on the day and 9.8% on the week. This arrives after the Fed raised rates on Sep 16; whether that decision caused any part of the gain is untested, since there is no consensus to grade the surprise against and no flow data. The central uncertainty is not direction. Roughly a ten percent weekly crypto gain has arrived with no spot volume, fund flow, exchange flow or liquidation data returned at all.^[1][2]

The big picture

This is revision 3 of the day. Against the edition filed Sat Sep 12 at 2:12 PM ET: the Fed met and raised rather than held, so that binary event has an answer; oil lost triple digits; and Bitcoin left the range rather than resolving inside it. The funding stand-down trigger touched at exactly 0.01% per 8h on Sep 12 is now untriggered, printing 0.002% per 8h on the same venue and interval, and both watched venues still lean long-paying.^[1]

The Fed hiked into a softening commodity tape

On Sep 16 the Fed raised its target range a quarter point to 3.75 to 4.00 percent, 12 to 0, with the interest rate on reserves at 3.90 percent. The finding reaches this letter through a relay and was not independently retrieved, so it is singly sourced; the size of any surprise cannot be graded because no consensus was supplied. EFFR at 3.88 percent on Sep 18 fits a full pass-through and corroborates transmission, not the surprise. Meanwhile WTI front-month crude is $91.75, down 4.5 percent on the day and 9.5 percent on the week, and Brent $95.52, down 3.8 percent and 9.6 percent, both off the Sep 12 references of $100.05 and $104.61. The level change is measured, the cause is not: no OPEC+ meeting, quota, outage or refinery event dated in the past week was returned, and no chokepoint transit or war-risk data within seven days. A dated report lays the softer tape on a firm dollar and fading hike-risk appetite, which is commentary about a mechanism, not a supply, demand and positioning decomposition.^[1][2][3][4]

Equities firm, the dollar firmer, gold softer

Equities took the hike in stride: the S&P 500 is 7,746.47, up 1.3 percent on the day and 1.7 percent on the week, and the Nasdaq Composite 27,016.55, up 1.9 percent and 3.2 percent. Gold is $4,389.8, down 0.8 percent on the day and 4 percent on the month, against a dollar index of 100.39, firmer on the day and week. The rate-driven reading of gold is a hypothesis, not a measurement: no dated haven-flow evidence either way was returned. The 10-year yield is 4.97 percent with the week flat, and a 5.01 percent peak print on Sep 16 did not hold as a daily level, so the binding-headwind criterion stays unmet rather than satisfied. A 2-year print of 4.40 percent is dated Sep 9, twelve days old, and cannot be read against the Sep 16 decision. The cross-asset picture is not a clean risk-on signal: stocks, dollar and crypto up, gold and crude down, with Bitcoin's 9.7 percent weekly gain far larger than the Nasdaq's 3.2. That suits an easing inflation impulse and equally suits crypto as the highest-beta expression of an equity bid; neither wins without flow evidence.^{[1][2][3][4][5]}

What it means for Bitcoin

OKX eight-hour funding on the BTC-USDT perpetual is 0.002 percent, seven-day average 0.0072 percent, the 17th percentile of 90 days, against exactly 0.01 percent at the 89th percentile on Sep 12. Funding is a periodic payment between long and short perpetual holders; positive means longs pay shorts. So leverage got cheaper on this one venue and interval while price gained about ten percent on the week. That is venue-specific confirmation, not a market-wide crowding measure. Hyperliquid is long-paying at 0.00166 percent per hour, a different interval that cannot be ranked against the eight-hour distribution. Open interest blurs the picture: CoinGecko-covered BTC derivatives show $83.67 billion with no seven-day trend, against $63.67 billion on Sep 12 from a differently scoped provider set, so those prints cannot be read as re-leverage. The one interval-matched growth statement is a single venue, OKX perp OI at $2.65 billion, up 19.9 percent over seven days. More contracts exist; who opened them cannot be inferred from a count. And with no spot volume, basis, ETF flow or liquidation totals for Sep 18 to 21, spot-led versus leverage-led is untestable.^[1][2][3]

Outlook: replaced. Confidence: low.

Bitcoin in the market

Snapshot taken Sep 21, 2026, 12:19 PM ET.

Bitcoin price history from this edition's recorded snapshot, not a forecast.

The market at a glance

Measurements from Sep 21, 2026, 12:19 PM ET.

Market	Level	Daily change	As of
Bitcoin	$85,793.94	+5.7%	2026-09-21
Ether	$2,762.85	+4.4%	2026-09-21
S&P 500	7,746.47	+1.3%	2026-09-21
Nasdaq	27,016.55	+1.9%	2026-09-21
Dollar index	100.39	+0.2%	2026-09-21
Gold	$4,389.8	-0.8%	2026-09-21
Brent	$95.52	-3.8%	2026-09-21
WTI	$91.75	-4.5%	2026-09-21

What comes next

Next meaningful test

The supplied calendar runs through the week: RBA Governor Bullock speaks late Sep 21, Australian employment and unemployment are due Sep 23, the SNB policy assessment with the rate forecast at 0.00 percent on Sep 24, US claims Sep 24 forecast 201K against 196K, revised University of Michigan sentiment Sep 25 forecast 47.5 against 47.8, and BOE Governor Bailey Sep 25. None of these is the Bitcoin test; no retrieved source compared them against anything beyond their priors, so none can be sized. The meaningful observation is the funding print. If OKX eight-hour funding stays below 0.01 percent while price holds above the old $78,300 ceiling, leverage is not being repriced upward into the gain on the venue that can be measured. If it crosses back above 0.01 percent, the strict-cross stand-down condition of the prior letter would fire on this instrument and interval. Both are conditional outcomes, not forecasts, and no liquidation data exists to gauge how much leverage is stacked either way.^[1][2]

Conditional scenarios

OKX eight-hour funding stays below 0.01 percent while price holds above the old $78,300 ceilingThe standalone long-perp cost stays cheap on the one venue and interval that can be measured; whether the wider market is uncrowded remains unestablished, and no flow evidence confirms real capital behind the move.

OKX eight-hour funding crosses back above 0.01 percentThe prior stand-down condition would be met on its original instrument, interval and strict operator, shifting the read toward an upward repricing of leverage in a market with no liquidation data to size it.

What remains uncertain

Whether the roughly ten percent weekly crypto gain was spot-led or leverage-led is untestable: no spot volume, CVD, basis, perp premium, ETF flow, stablecoin supply, exchange balance or liquidation total was returned for Sep 18 to 21, so no buyer or seller can be identified.
All options data is unknown: no implied volatility, skew, term structure or expiry and strike observation returned, so no conditional options structure may be named.
Cause of the Sep 17 to 21 oil unwind is unknown; the firm-dollar and fading-hike-risk attribution is commentary on a mechanism, and no shipping chokepoint transit, war-risk or freight data was retrieved.
FOMC consensus, dot plot, press conference content and minutes are unsupplied, so the Sep 16 hike's surprise cannot be graded; the statement itself is singly sourced. Fed plumbing (balance sheet, reserves, ON RRP, TGA, auctions) is unsupplied, and BLS CPI, PPI, payrolls and JOLTS actuals remain unsupplied for an eighth consecutive revision.

How the outlook has evolved

continuing

Bitcoin keeps the $76,500 to $78,300 range into the Sep 16 Fed.

No closing level at the Sep 16 2:00 PM ET horizon was supplied, so the criterion is unmet and the thread cannot be resolved or closed. Later evidence: price $85,793.94 on the Sep 21 read, above the ceiling, so the range has been left behind, but not at the graded timestamp. The frame is retired for the new letter; the original claim stays open until a Sep 16 close is available.

Original criteria and dates

First recorded 2026-09-12. Horizon: 2026-09-16T14:00:00-04:00.

Daily closes remain strictly inside $76,500 and $78,300 through the Sep 16 FOMC.

resolved

The cheap-deck read is dead; stop all adds if funding crosses 0.01 percent per 8h.

The stated outcome occurred: the operator was never satisfied. The Sep 12 print touched exactly 0.01 percent, not a strict cross, and the matched instrument and interval now prints 0.002 percent, 17th percentile. The branch that occurred is the reset below the threshold, not confirmation of a cross above it; the conditional stand-down never fired and the path between the two reads is unobservable.

Original criteria and dates

First recorded 2026-09-12.

A strict cross above 0.01 percent per 8h on OKX BTC-USDT-SWAP funding.

continuing

Both watched venues lean long-paying; treat the weekend bid as leverage, not demand.

Both remain long-paying: OKX +0.002 percent per 8h and Hyperliquid +0.00166 percent per hour. Intensity is not comparable across the two intervals. The leverage-not-demand half stays untestable with no flow data.

Original criteria and dates

First recorded 2026-09-12.

OKX and Hyperliquid funding both strictly above zero over consecutive reads.

contradicted

Rising funding with flat or shrinking open interest reads as existing longs paying more while some trim, not new leverage arriving.

Both inputs reversed: funding fell to 0.002 percent at the 17th percentile while OKX perp open interest rose 19.9 percent over seven days to $2.65B. The cross-date market-wide comparison ($83.67B versus $63.67B) is unusable because the provider sets differ and no seven-day trend exists, so the market-wide half of the original criterion cannot be evaluated at all. The original trimming reading is not supported by these prints; a contract count cannot establish who trimmed or added. A new criterion for a re-leveraging test, dated 2026-09-21, is a strict rise in the same venue set's open interest on a same-interval read, and would be a new thread rather than a revision of this criterion.

Original criteria and dates

First recorded 2026-09-12.

Whether market-wide open interest is flat or shrinking while funding rises, at the same venue set and interval.

contradicted

Oil holds triple digits and the inflation impulse stays alive.

WTI $91.75 and Brent $95.52, both dated Sep 21, so the level condition fails. Cause unestablished; the inflation-impulse half is not measured by this criterion.

Original criteria and dates

First recorded 2026-09-12.

WTI front month at or above $100 and Brent at or above $100 on a dated print.

contradicted

The dollar is flat and there is no haven confirmation of a fresh risk-off impulse.

DXY 100.39, up 0.9 percent on the week against a flat 99.1 on Sep 12, so the flat-dollar half fails. The haven half is not established either way. Cause unestablished; no haven-flow data was returned.

Original criteria and dates

First recorded 2026-09-12.

DXY unchanged within a small band versus the prior read.

continuing

A 10-year yield above 5.00 percent is a binding headwind.

A 5.01 percent peak print on Sep 16 did not hold as a daily level; the read here is 4.97 percent with the week flat. Under the original operator the criterion is unmet, not satisfied. Cause unestablished.

Original criteria and dates

First recorded 2026-09-12.

A strict 10-year yield above 5.00 percent, sustained rather than a single peak print.

resolved

The Sep 16 FOMC is the resolving event for the range, with hawkish and dovish branches.

The stated outcome occurred: a 25bp hike to 3.75-4.00 percent, 12-0, with EFFR at 3.88 percent on Sep 18. The observed path then moved upward rather than testing $76,500 or stopping at $78,300, matching neither named branch. Surprise size is not gradeable; no consensus was supplied.

Original criteria and dates

First recorded 2026-09-12. Horizon: 2026-09-16T14:00:00-04:00.

Compare the decision and its transmission against the hawkish and dovish branches named in the prior letter.

Every edition and failed attempt

Supporting evidence

OKX BTC-USDT-SWAP eight-hour funding 0.002 percent, seven-day average 0.0072 percent, 17th percentile of 90 days, against exactly 0.01 percent at the 89th percentile on Sep 12. Same venue and interval; the strict-cross stand-down trigger is not met at this read, and the path between the two endpoint reads is unobservable.^[1]

Open interest: CoinGecko-covered BTC derivatives $83.67B with no seven-day trend against $63.67B dated Sep 12, differently scoped so cross-date re-leverage cannot be inferred; OKX perp OI $2.65B, up 19.9 percent over seven days, the only interval-matched growth statement; Hyperliquid $3.90B at a mark of $85,845.^[1][2][3]

FOMC raised the target range 25bp to 3.75-4.00 percent on Sep 16, 12-0, IORB to 3.90 percent; EFFR 3.88 percent dated Sep 18, up 25bp. No consensus was supplied, so surprise size is ungaugeable, and the statement finding is singly sourced through a relay.^[1]

Calendar inside seven days, with no actual yet: AUD employment 20.9K forecast vs -15.8K prior and unemployment 4.5 vs 4.5 percent (Sep 23), SNB rate 0.00 vs 0.00 percent (Sep 24), USD claims 201K vs 196K (Sep 24), revised UoM sentiment 47.5 vs 47.8 and inflation expectations prior 4.6 percent (Sep 25), BOE Bailey (Sep 25).^[1]

Bitcoin $85,793.94, +5.7 percent day and +9.7 percent week; Ether $2,762.85; S&P 500 7,746.47, +1.3 percent day; Nasdaq Composite 27,016.55, +1.9 percent day; dollar index 100.39, +0.9 percent week; gold $4,389.8, -0.8 percent day and -4 percent month; 10-year 4.97 percent, week flat, against a 5.01 percent peak print on Sep 16 that did not hold as a daily level.^{[1][2][3][4][5]}

Technical details

Snapshot taken Sep 21, 2026, 12:19 PM ET. These measurements belong to this edition.

Bitcoin$85,794▲ +5.7% 1d

Funding 8h0.002%p17 of 90d

Open interest$83.67B

US 10Y4.97%0 bp 1w

Fed funds3.88%▲ +25 bp 1w

Dollar (DXY)100.39▲ +0.2% 1d

Brent$95.52▼ -3.8% 1d

Gold$4,390▼ -0.8% 1d

S&P 5007,746▲ +1.3% 1d

Additional measurements from this edition's recorded snapshot.

Sources

30 returned sources; citation presence does not establish that every claim is verified.

Snapshot sources

Successful readings saved with this edition; separate from researcher retrieval coverage.

How this edition was produced

3 × deepseek/deepseek-v4.1-flash

$0.12 · 6 m 48 s · run on a connected key

3 of 3 researchers returned notes; 2 took a position on the others' notes; deepseek/deepseek-v4.1-flash consolidated them into the one agreed document the brief was written from.

Editorial review: flagged. Model review is not independent verification.

967 narrative words; 1146 supporting words

Geopolitics and energy: 5 assigned-source citations

Economy and policy: 3 assigned-source citations

Crypto flows and positioning: 5 assigned-source citations

Research notes

AGREED

- Sep 16, FOMC raised the target range 25bp to 3.75–4.00%, 12–0, IORB to 3.90%, standing repo/reverse repo at 4.00%/3.75%; reported by Researcher B and relayed by Researcher A and Researcher C, but Researcher B's own position did not answer this round and its notes were not delivered, so this rests on a secondhand relay of one primary document, New York Fed FOMC statement, https://www.newyorkfed.org/medialibrary/media/markets/fomc-statement-20260916.pdf.
- Sep 18, EFFR 3.88%, up 25bp on the week; NY Fed via OBSERVED snapshot and Researcher B (Federal Reserve H.15, https://www.federalreserve.gov/releases/h15/); the two are the same published series, so this is corroboration of pass-through, not independent confirmation of the hike's surprise size.
- Sep 21, 12:19 PM ET, Bitcoin $85,793.94, +5.7% day, +9.7% week, +11.3% month; Binance spot via OBSERVED snapshot. Ether $2,762.85, +4.4% day, +9.8% week, +14% month.
- Sep 21, 12:19 PM ET, equities risk-on: S&P 500 7,746.47 (+1.3% day, +1.7% week), Nasdaq Composite 27,016.55 (+1.9% day, +3.2% week); OBSERVED snapshot.
- Sep 21, WTI $91.75 (−4.5% day, −9.5% week) and Brent $95.52 (−3.8% day, −9.6% week); Yahoo front-month futures via OBSERVED snapshot (https://finance.yahoo.com/quote/CL%3DF). Both are off triple digits from the Sep 12 references of $100.05 and $104.61.
- Sep 21, DXY 100.39, +0.9% week; Yahoo DX-Y.NYB via OBSERVED snapshot. Researcher A additionally cites a Reuters Sep 17 report of a seven-week high on the hike, https://www.reuters.com/world/china/global-markets-global-markets-2026-09-17/.
- Sep 21, Gold $4,389.8, −0.8% day, −4% month; OBSERVED snapshot.
- Sep 21, OKX BTC-USDT-SWAP 8h funding 0.002%, 7d average 0.0072%, 17th percentile of 90 days; OKX perp via OBSERVED snapshot. Interval and venue match the prior criterion's instrument.
- Sep 21, both watched venues are long-paying: OKX +0.002% per 8h, Hyperliquid +0.00166% per hour; OBSERVED snapshot, venue-named.
- Sep 21, open interest: CoinGecko-covered BTC derivatives $83.67B with no 7d trend; OKX BTC-USDT perp $2.65B, +19.9% over 7 days; Hyperliquid $3.90B at mark $85,845; OBSERVED snapshot, venue-named. OI rose; who opened and why is not established.
- Sep 17–21, no dated OPEC+, quota, outage, hurricane, pipeline or refinery event, no shipping chokepoint transit or war-risk data, and no Sep 14–21 trade or sanctions primary (USTR, BIS, OFAC, EU) was returned by any seat.

RESOLVED

- The prior letter's Sep 16 FOMC event is answered by an actual: a 25bp hike to 3.75–4.00%. The decisive evidence is the FOMC statement relayed from Researcher B (https://www.newyorkfed.org/medialibrary/media/markets/fomc-statement-20260916.pdf), with EFFR at 3.88% on Sep 18 as interval-matched corroboration. The event is resolved; the surprise size against consensus is not, because consensus was never supplied.
- The prior thread "oil holds triple digits, inflation impulse stays alive" is contradicted on level. WTI $91.75 and Brent $95.52 are dated measurements (https://finance.yahoo.com/quote/CL%3DF) against the Sep 12 references of $100.05 and $104.61. Level change resolved; cause not resolved.
- The prior "dollar flat / no haven confirmation" framing is superseded on the dollar: DXY 100.39, +0.9% week, versus 99.1 flat on Sep 12. Both A and C accept this on the dated snapshot row. Gold's −4% month alongside a firm dollar has no dated haven-flow evidence either way, so that half stays open.
- The prior stand-down trigger set Sep 12 at a strict cross above 0.01% per 8h OKX funding is not touched at this read: the instrument and interval are matched and the current print is 0.002%, 17th percentile. A and C agree on the operator reading. What that settles is the state at Sep 21, not the path between Sep 12 and Sep 21.
- Researcher A's arithmetic that the 10-year "topped 5.00% for the first time in three years" describes a Sep 16 peak print of 5.01% on H.15 is accepted over a sustained-cross reading; A and C converge, and the strict 5.00% criterion therefore remains unmet rather than satisfied. Evidence: https://www.federalreserve.gov/releases/h15/.

UNRESOLVED

- Researcher B's position is missing, not empty. Its FOMC and H.15 findings survive only as relayed by A and C, and no seat independently retrieved those primaries. Treat the Sep 16 hike as singly sourced via a secondhand relay.
- Cross-date market-wide open interest: $83.67B (Sep 21) versus $63.67B (Sep 12) cannot be called re-leverage. The two prints come from differently scoped CoinGecko provider sets and no 7d trend exists for the current one. Both A and C dispute the inference. The only interval-matched growth statement is OKX perp +19.9% over 7 days, one venue.
- Cross-venue funding comparison stays unresolved: converting Hyperliquid's 0.00166% per hour to ~0.0133% per 8h is arithmetic, not a same-interval market-wide print, and must not be ranked against the 90-day 8h distribution. A and C both dispute it.
- The prior criterion's hold-or-fade path between Sep 12 and Sep 21 is unobservable from two endpoint reads. A lower print on Sep 21 establishes fade at this read, not a clean cross or a hold in between.
- Whether the roughly +10% weekly crypto move was spot-led or leverage-led is untestable: no spot volume, CVD, basis, perp premium, ETF flow, stablecoin supply, exchange-flow or liquidation total was returned for the Sep 18–21 window. A +5.7% day with no liquidation print is itself a data absence both A and C flag. Mixed or unclear is the valid state.
- All options data for Sep 18–21 remains UNKNOWN; no IV, skew, term structure or expiry/strike observation returned, so no conditional structure may be named.
- Cause of the Sep 17–21 oil unwind is UNKNOWN. Researcher A's dollar-and-fading-hike-risk attribution is secondary commentary on a mechanism, and A itself lists the driver as UNKNOWN. No supply, demand or positioning decomposition was returned.
- Shipping and chokepoint evidence (Red Sea, Suez, Hormuz, Panama, Baltic transit counts, freight rates, war-risk premia) dated within seven days was not retrieved by any seat; Researcher A names this a material gap in its own assignment.
- Fed plumbing (WALCL, reserves, ON RRP, TGA, auction results and tails) remains unknown, so the bill-purchase directive cannot be sized; B's relay did not supply it.
- Post-meeting Fed communication, dot plot, press-conference content and minutes are unknown, so the Sep 16 hike's surprise cannot be graded.
- BLS CPI, PPI, payrolls and JOLTS actuals are unsupplied for an eighth consecutive revision; Sep 10 and Sep 11 CPI, PPI, ECB, claims, UoM and GBP GDP actuals likewise.
- Calendar items inside the next seven days are scheduled but not evaluated: SNB policy rate Sep 24 03:30 ET (forecast 0.00%, prior 0.00%), BOE Gov Bailey Sep 25 05:15 ET, USD claims Sep 24 08:30 ET (forecast 201K, prior 196K), Revised UoM Sentiment Sep 25 10:00 ET (forecast 47.5, prior 47.8) and Inflation Expectations (prior 4.6%). No seat returned a forecast or prior discrepancy for these; no actual exists yet.
- Whether any crypto-specific news event (regulatory, venue, ETF) explains the divergence from equities is unknown; neither A nor C returned one.
- BTC dominance and market-wide OI 7-day trend remain unknown, carried from the prior letter without resolution.

Contribute the next chapter

Choose one researcher or three in Run. New drafts are reviewed before publication. If material questions remain unresolved, the current edition stays live.