The number that decided the week was one tenth of one percentage point. August core consumer prices — the series that strips out food and energy because those move for reasons monetary policy cannot reach — rose 0.3% on the month against a 0.2% consensus, and against 0.2% in July. Headline inflation came in exactly as forecast at 0.4% on the month and 3.4% on the year. Core on the year was 2.4%, also as forecast. One line of one release missed by a tenth, and by Friday evening the market had moved from pricing a September rate hike at about 70% to pricing it at close to 90%.

And bitcoin rose. Bitstamp closed Friday 11 September at $77,212.48, up 0.893%, ending four consecutive declines. It then closed Saturday 12 September at $77,269.22, up 0.0735%, in a session whose entire high-to-low range was $420.00 — the fourth-narrowest of the 255 completed sessions of 2026. That is the shape of the weekend: a violent Friday that resolved almost nowhere, then a Saturday that barely happened at all.

The mainstream story, and where it actually touches bitcoin

It is worth being explicit about the direction of travel here, because a reader arriving from general news coverage may have the wrong cycle in mind. The Federal Reserve is not cutting. Its target range is 3.50% to 3.75%, the Chair is Kevin Warsh, and the Federal Open Market Committee meets on Tuesday 15 and Wednesday 16 September with an updated dot plot, in a meeting that the market now expects to raise rates. CBS News summarised the post-CPI consensus as a hike being “all but guaranteed”. On the probabilities the desk is printing every number it found rather than picking the most dramatic: Yahoo Finance put the market at “roughly 87%”, up from 72% a day earlier and 50% a week earlier; CME’s FedWatch tool was reported at “nearly 90%, up from 70% on Thursday”; Polymarket had a hike at 79.5% with a hold at 20.5%. Three venues, three numbers, all far above the 60% bar the desk set on Thursday.

That grades N1 a pass. The bar was FedWatch at 60% or better on the evening of Friday 11 September, confirmed by two outlets. CME FedWatch was reported at nearly 90%; Yahoo Finance put the market-implied probability at roughly 87%. Both are far above the bar, and the grade does not turn on which measurement you prefer.

Where this genuinely touches bitcoin is not the headline but the composition. The uptick was an energy story: on the Bureau of Labor Statistics release, fuel costs were up 28% from a year earlier, diesel up 52%, and gasoline rose 3.9% on the month, accounting for over a third of the entire monthly increase, while natural gas fell 1.1% and electricity fell 0.2%. Energy-led inflation is the kind a central bank cannot fix by tightening, and it arrives in a month of elevated crude prices following the strikes on Iranian tankers the desk covered on Thursday. That is the macro backdrop against which bitcoin is being asked to behave like a hedge, and the honest reading of this week is that it did not: it fell 3.821% across the six sessions from Monday 7 to Saturday 12 September, measured from Sunday 6 September’s close of $80,338.98 to $77,269.22, with a weekly high of $80,430.96 on Monday and a weekly low of $76,040.02 on Friday.

The rates verdict: the front end confirmed the hike and the long end did not

The Treasury constant-maturity curve for Friday 11 September, the final row published at the time of the snapshot, is the cleanest single statement of what the market concluded. Set against Thursday 10 September:

    • 2-year 4.63%, up 7 basis points from 4.56%.
    • 3-year 4.69%, up 6bp. 5-year 4.78%, up 3bp. 7-year 4.87%, up 3bp.
    • 10-year 4.96%, up 1bp — the highest of every constant-maturity row published so far in 2026, a series beginning 2 January.
    • 20-year 5.38%, down 1bp. 30-year 5.35%, down 2bp.
    • Bills: 1-month 3.93%, 3-month 4.07%, 6-month 4.12%, 1-year 4.35%.

That is a twist, not a shift. The front end went up because the market decided the Fed is hiking on Wednesday; the long end went down because a hike delivered into energy-led inflation is disinflationary further out. The 2s10s spread compressed to 33 basis points from 39, and 2s30s to 72 from 81. When a curve flattens from the front on a hot core print, the market is not telling you inflation has won — it is telling you it expects the central bank to win, and to pay for it later.

That grades O1 a pass. The bar, set on Thursday, was a 2-year CMT close on Friday 11 September at 4.61% or higher, which is 5 basis points above Thursday’s 4.56%. It printed 4.63%, 7 basis points up and 2 above the bar.

One discrepancy to print rather than hide, because it is exactly the trap Field Guide #42 was written about. The desk’s 10-year figure is 4.96%, up 1bp. Market reporting on Friday had the 10-year falling one basis point to 4.93% after touching 4.979% intraday, its highest since late 2023. Both are correct and they are not the same measurement: the constant-maturity yield is read off the par curve from bid-side quotations gathered at or near 3:30 PM in New York, while a market-close quote is a later print. A reader who mixes the two will conclude the 10-year both rose and fell on the same day. It did, depending on when you looked. Marker AB1 — the 10-year CMT closing at 5.00% or above on any day to Friday 18 September — is now 4 basis points away, and it is graded on the constant-maturity close, not the intraday touch.

The two markers that failed

U1 has settled a fail. The bar was a Bitstamp close on Friday 11 September at $80,318 or better — Strategy’s 4,603-coin tranche cost basis, and a level the desk noted on Thursday would require a 4.951% single-session move. Friday delivered 0.893%. The close was $77,212.48, $3,105.52 short. Worth noting how it failed rather than that it failed: Friday’s high was $79,836.83, the highest high since Monday 7 September, and its low was $76,040.02, the lowest low since 23 August. A range of $3,796.81 — the 29th-widest of 255 sessions — on volume of 2,773.55 coins, the 71st-highest of 255. The market moved a long way in both directions and finished up less than one percent. Post-CPI sessions often look like that.

W1 has settled a fail. The bar was Hyperliquid’s HYPE token at $82.60 or above at 00:00 UTC on Sunday 13 September. CoinGecko’s daily series prints $79.6473, $2.9527 short. The path was 78.5568 on Friday and 79.4341 on Saturday, so the token recovered from Friday’s low without getting near the bar, and it was trading at $79.31 at 06:14 UTC Sunday. The desk set that bar when HYPE had, on the desk’s own reporting, just printed a record high of $89.60 on the day its $820 million unlock landed; the lesson it now records is that a bar set 7.8% below a fresh high, four days out, in a week bitcoin itself fell 3.821% and HYPE fell further, was a bar set on momentum rather than on arithmetic.

With Thursday’s I1 pass, which was graded in Friday’s column, that is five markers settled across the four days from 10 to 13 September: I1 pass, N1 pass, O1 pass, U1 fail, W1 fail. Full grading history sits in Friday’s markers column.

Leverage: the bar moved under the marker

The most consequential change in the derivatives complex over the weekend is that positioning got smaller, not that it got more bullish. Binance USDT-margined perpetual open interest at 00:00 UTC:

    • Friday 11 September: 106,858.934 BTC.
    • Saturday 12 September: 103,385.135 BTC — down 3,473.799, or 3.251%, in one day.
    • Sunday 13 September: 103,299.613 BTC — down a further 85.522, or 0.083%.
    • Two-session change: −3,559.321 BTC, −3.331%. Dollar value $7,980.00 million, implying a price of $77,251. Live open interest at 06:10 UTC was 103,613.862, up 314.249 on the 00:00 print.

That flips X1. The bar, set on Friday, was Binance perpetual open interest at 00:00 UTC on Thursday 17 September coming in above 105,105.968 BTC — Thursday’s window low. On Friday the reading was above the bar and the marker looked comfortable. It is now 1,806.355 BTC below it, a shortfall of 1.719%, with four days to run. Roughly 1,750 coins of leverage arrived in the session to Friday’s 00:00 UTC print and roughly 3,560 left over the weekend — twice as much out as in, leaving open interest below where it stood on Thursday — and the marker went from passing to failing without the price doing anything at all. G1 — open interest below 100,000 at any September 00:00 print — is now only 3.300% away, having been 6.86% away on Friday.

The rest of the leverage picture is mildly constructive and internally consistent. Funding on the perpetual printed 0.006093% at 00:00 UTC Saturday (6.672% annualised), 0.004334% at 08:00 (4.746%), 0.005169% at 16:00 (5.660%) and 0.004788% at 00:00 UTC Sunday (5.243% annualised) — positive throughout, longs paying shorts, but nowhere near stressed. The quarterly basis improved sharply: against a 06:10 index of $77,285.185, the September contract at $77,446.2 is a 0.2083% premium over 12.08 days, 6.297% annualised, up from 3.977% on Friday; December at $78,363.0 is 1.3946% over 103.08 days, 4.938% annualised. That leaves G2 — September basis above 8.00% on Friday 25 September — 170 basis points short, having been 402 short on Friday. The desk described G2 on Friday as effectively dead. It is not dead. It is 170 basis points and twelve days away, which is a useful reminder that calling a marker before its settlement date is a habit worth breaking.

Account ratios went the other way, and this is the one number the desk would flag as a caution rather than a comfort. Binance’s global long/short account ratio went 1.6008 on Friday to 1.5471 on Saturday to 1.6539 on Sunday. The top-trader position ratio went 2.1224 to 2.2061 to 2.2324 — the highest since 26 August, when it was 2.2560. So the crowd is smaller but more one-sided: fewer coins of open interest, held in a more heavily long configuration, four days before a Federal Reserve decision that the market expects to tighten. Funding at 5.243% annualised sits 105 basis points below the September basis, against 69 on Friday, which is the market paying up for term exposure rather than for leverage.

The ETFs quietly halved September’s cushion

Farside’s table now has a complete September through Friday, and the two rows that were missing on Friday are both negative. In US$ millions: 1 September −236.5, 2 September +101.1, 3 September +730.8, 4 September +174.6 — then no row for Monday 7 September, which was Labor Day — 8 September −46.6, 9 September −120.2, 10 September −282.7, 11 September −13.2.

September month-to-date is +$307.3 million over eight sessions, against +$603.2 million over six when the desk last printed it. The month’s cushion has halved, and every dollar of the reduction came in the last four sessions, which were four consecutive daily net outflows totalling −$462.7 million. BlackRock’s IBIT ran +10.7, −19.5, −24.5, −19.2 across those four, so three straight negative days for the largest fund in the complex. H2 — September net flow positive, settling Wednesday 30 September — remains open and positive, but it is now a $307 million cushion with seventeen calendar days and thirteen trading sessions to go, and two days like 10 September would erase it.

The one genuinely encouraging detail: Friday’s −$13.2 million was the smallest of the four outflows, and it landed on the day bitcoin rose and the whole rates complex repriced. Cumulatively since launch the complex has taken in $55,224 million. IBIT alone has taken in $64,004 million, against Grayscale’s GBTC at negative $27,782 million — which is why the complex total is lower than the total of its largest fund.

Mining is about to deliver the third-largest retarget of 2026

The difficulty projection has moved a long way in two days and deserves more attention than it usually gets. If it settles near its current level it will be the third-largest upward retarget of 2026, behind +14.725% on 19 February at height 937,440 and +7.153% on 27 June at height 955,584. At retarget height 967,680 the epoch is 1,117 blocks of 2,016 complete (55.41%) with 899 to go, at an average block interval of 9.541 minutes. The projection is +4.902%, against +3.103% on Friday when the epoch was 39.73% through — a move of nearly 180 basis points in 48 hours. The estimated timestamp is 05:07 UTC on Saturday 19 September. The previous retarget settled at +1.307%.

Hashrate rows, which mempool.space stamps at the end of each period, read 1,025.39 EH/s for 12 September and 986.70 EH/s for 13 September, with a current-hashrate estimate of 958.04. Pool shares over the trailing week across 1,059 blocks: Foundry USA 24.7%, AntPool 18.5%, F2Pool 13.8%, ViaBTC 9.3%, SpiderPool 9.1%, MARA Pool 4.7%, SECPOOL 4.6%, Luxor 3.4%. Fees were 3 satoshis per virtual byte for the fastest tier and 1 for everything else at the time of the snapshot — which is why a token sale could put 1,110 transactions on chain this weekend for about a thousand dollars, as the desk’s lead article documents. Marker P1 — the retarget at 967,680 coming in positive — is on course by a wide margin.

Sentiment, for completeness: the Fear & Greed index printed 56 on Friday, 63 on Saturday and 61 on Sunday, the twenty-fifth consecutive day in Greed — a run that began on 20 August 2026, the day after a Fear reading of 46. Friday’s 56 remains the lowest of all twenty-five readings.

The marker table

Twenty-four markers were open after Friday. Four settled today and four are added, leaving twenty-four open. Every weekday named below was checked programmatically rather than by eye, following the error the desk made twice last week.

    • H2 — September Farside net flow above zero, settles Wed 30 Sep. Reading +307.3 over 8 sessions. Open, positive, cushion halved.
    • H3 — 30-year CMT at 5.25% or above on Thu 17 Sep. Reading 5.35% (11 Sep). Passing by 10bp.
    • G1 — perpetual open interest below 100,000 at any September 00:00. Reading 103,299.613. 3.300% above the bar.
    • G2 — September basis above 8.00% on Fri 25 Sep. Reading 6.297%. 170bp short, and not dead.
    • I1 — settled PASS on 10 September at a 2.55% real 10-year.
    • I2 — Bank of Japan hikes on Fri 18 Sep. Open.
    • K1 — CLARITY Act cloture reaching 60 votes by Fri 18 Sep. Open.
    • L1 — Bitstamp close on Wed 16 Sep at $81,265 or above. Reading $77,269.22, a gap of $3,995.78, needing +5.171%.
    • N1 — SETTLED PASS. FedWatch at 60% or better on the evening of Fri 11 Sep, two outlets: 87% and about 90%.
    • O1 — SETTLED PASS. 2-year CMT on Fri 11 Sep at 4.61% or above: printed 4.63%.
    • P1 — retarget at 967,680 positive, about Sat 19 Sep. Projection +4.902%. On course.
    • Q1 — Senate cloture roll call occurs on Tue 15 Sep. Open.
    • R1 — Trezor’s ShipMonk affected-customer count revised up by Wed 30 Sep. Open.
    • S1 — Trezor Anonymous Delivery live in at least one EU country by Wed 30 Sep. Open.
    • T1 — 3,500 BTC or more returned to the Liquid peg by 00:00 UTC Monday 14 Sep. Reading 3,400, unchanged. Failing by 100, grades tonight.
    • U1 — SETTLED FAIL. $77,212.48 against $80,318, short $3,105.52.
    • V1 — Liquid restart plus a federation peg-out by Wed 30 Sep. Blocks yes, peg-outs still suspended. Open, half met.
    • W1 — SETTLED FAIL. HYPE $79.6473 against $82.60 at 00:00 UTC Sun 13 Sep, short $2.9527.
    • X1 — perpetual open interest at 00:00 UTC Thu 17 Sep above 105,105.968. Reading 103,299.613. Now 1,806.355 below the bar, having been above it on Friday.
    • Y1 — a Blockstream or Liquid statement by Wed 30 Sep either sourcing full backing from non-whitehat funds or confirming a bounty paid or agreed. Open.
    • Z1 — the next Strategy 8-K, due by Tue 15 Sep, showing at least one bitcoin bought. Open. The tranche is $3,048.78 per coin underwater, $14.034 million across 4,603 coins.
    • AA1 — the whitehat address spending at least one output by 00:00 UTC Thu 17 Sep. No spend since 7 Sep. Open.
    • AB1 — 10-year CMT at 5.00% or above on any day to Fri 18 Sep. Reading 4.96%. 4bp away.
    • AC1 — Blockstream or Liquid publishing replayed-versus-discarded transaction counts by Wed 30 Sep. Open.
    • AD1 — Trezor publishing its Brevo-access findings by Wed 30 Sep. Open.
    • AE1 (new) — cumulative ico-20 mints paying the whitehat address reaching 2,000 by 00:00 UTC Wed 16 Sep. Reading 1,109. The desk expects this to fail.
    • AF1 (new) — the FOMC raises the target range at the meeting concluding Wed 16 Sep. Current range 3.50–3.75%; market 79.5% to about 90%.
    • AG1 (new) — the ico-20 share of the whitehat address’s lifetime transaction count, re-pulled at 06:10 UTC Wed 16 Sep, is published against today’s 77.8%. Set by today’s field guide as a test of its own fee arithmetic.
    • AH1 (new) — Revolut names the impersonated government agency or publishes an affected-customer count by Wed 30 Sep. Open.

What the desk is watching next

Monday is quiet by design: T1 grades at 00:00 UTC as Sunday ends, and Strategy’s 8-K is due Monday or Tuesday for Z1. The week is then front-loaded. Tuesday brings the Senate cloture roll call for Q1, the first day of the FOMC, and the end of the AE1 and AG1 windows, which both settle at 00:00 UTC Wednesday. Wednesday brings the decision and dot plot for AF1, the L1 and AG1 closes, and the House Ways and Means Committee’s vote on a crypto tax package — a vote the desk had not previously tracked and will now. Thursday brings X1, AA1 and H3. Friday brings the Bank of Japan for I2, the CLARITY deadline for K1 and the AB1 cut-off. Saturday brings the retarget for P1.

Two standing cautions, both of which the desk has earned. First: bitcoin closed Saturday $1,012.43 below Wednesday 9 September’s close, which is the last time it closed higher than it is now, and it is still 11.688% down for the year and 1.657% down for the month. A two-session advance streak — the thirty-third streak of two or more in 2026 — is not a trend. Second: Saturday’s $420 range and 499.05 coins of volume, the 23rd-lowest of 255 sessions, mean the price sitting on the screen this morning has been tested by almost nobody. The Sunday partial at 06:10 UTC was $77,285.94, up 0.0216%, on 105.78 coins.

Method: prices, funding, open interest, basis, mining and on-chain figures in this article are pulled directly by Bitcoin Mastery at the timestamp stated — Bitstamp BTC/USD daily candles for closes, Binance BTCUSDT spot and USDT-margined perpetual for intraday, open interest, funding and account ratios, Binance COIN-M quarterly contracts for basis, mempool.space for difficulty, hashrate, pool shares, fees, address balances and individual Bitcoin transactions, blockstream.info’s Liquid API for sidechain block heights, hashes, timestamps and transaction counts, alternative.me for the Fear & Greed series, CoinGecko for altcoin daily prices, Farside Investors’ table for ETF flows and US Treasury CMT par yields for rates. Transaction counts, fee totals, byte totals and OP_RETURN payloads are recomputed from the full confirmed transaction list of the address concerned, not read off a summary. Where a third-party figure is cited we name the source and its date; where two sources disagree we print both. Every streak or extreme figure is published with the first date of its series in the same sentence.

Disclaimer: This article is for informational purposes only and does not constitute investment advice. Cryptocurrencies are volatile and you can lose money. Nothing here is a recommendation to buy or sell any security, digital asset, token or exchange-traded fund, including MSTR, L-BTC, HYPE, ORDI or the LEAF token where discussed above. Token sales of the kind described in this article are unaudited, frequently anonymous and have no obligation to deliver anything in return for a payment; treat any coin sent to one as capable of going to zero. Do your own research and consult a licensed financial advisor before making investment decisions.