This edition was filed at 06:00 AEST, ten minutes after the US cash close, before the authoritative sources had posted. They have since posted and five figures are corrected here. The body text below is left as filed — the masthead's data-as-of discipline means an edition is a point-in-time document, and rewriting it silently would defeat the purpose. Everything in this box supersedes the corresponding figure below.
- The S&P 500 range is retired. Monday's close was 7,619.98, −37.00 points, −0.5% (AP tabulation). The narrative recap this edition used for the low end of the range was right to the point; the quote page that produced 7,629.64 was a CFD, not the cash index. Dow 52,421.20 (−152.09, −0.3%) and Nasdaq Composite 26,186.41 (−146.62, −0.6%) are both confirmed as published.
- The Russell 2000 figure published in the dashboard and the tape was wrong. The correct close is 2,892.24, −11.71, −0.4%, not the 2,898.27 / −0.20% shown below — the same CFD-versus-cash error, and it understated the decline by half. The characterisation still holds: the Russell outperformed the S&P (−0.4% against −0.5%) on a session led by mega-cap AI, which is the point the text makes.
- The official Treasury par curve for 14 September has posted: 2y 4.65%, 5y 4.80%, 10y 4.97%, 30y 5.34%. The 10-year is therefore 4.97% on the official par basis, one basis point below the 4.98% vendor mark published; the 5.00% intraday print is unaffected and independently confirmed.
- Material for the book: 5s30s is markable after all, and it is closer to its stop than this edition said. With the 5-year at 4.80% and the 30-year at 5.34%, 5s30s closed Monday at 54bp, three basis points flatter on the session — so the steepener (V003) is nine basis points from its published 45bp stop, not twelve, and the section 08 line describing it as unmarked is superseded. 2s10s closed 32bp. The shape is unchanged and the reading in theme 2 stands: the belly led again (5y +2bp) while the 30-year fell a basis point.
- Fed pricing settled higher than the snapshot published. The Fed Rate Monitor refreshed at 17:05 ET on 14 September to 91.5% for the 16 September hike, against the 88.1% pre-open snapshot this edition used and labelled as such. The direction of the argument is unchanged and the Polymarket wedge is wider, not narrower — roughly 12.5 points against the settled read rather than the nine published.
What this says about the process, and what has changed because of it. Four of the five corrections have one cause: at a 06:00 Sydney start on a Tuesday to Friday the US bell is ten minutes old, and the AP tabulation, the official par curve and the settled Fed-pricing refresh have not published. The edition flagged that timing problem in section 13 and published ranges rather than guesses, which was the right call — but a range that resolves to its low end, and one outright wrong index level, are still a worse product than waiting. The runbook has been amended: US-close figures are now fetched last, after the rest of the note is drafted, and the Fed Rate Monitor is not quoted before its post-17:00 ET refresh. The source library's note that the quote page was "usable as a second source" has been withdrawn — it was wrong, and this correction is the evidence.
The bottom line
- The hike is now 88% priced and the June dots sit roughly 30bp below where the strip ends the year — so Wednesday is a projections event, not a decision event. The 16 September hike moved 85.5% → 88.1% (Investing.com Fed Rate Monitor, stamped 14 Sep 07:45 ET — a pre-open Monday read, not a settled one). Cumulative ≥1 hike 93.6% by October, 97.8% by December; the modal December end-state is 4.00–4.25% at 48.7%, with the tail skewed to three hikes (26.5%) over one (22.6%). The June SEP median for end-2026 was 3.8% against a June committee split 9–9 on whether to hike at all. The median has to move to roughly 4.1% simply to validate what is already priced. A dot plot that confirms only one more hike is the dovish surprise, and almost nobody is positioned for it. Polymarket's event book prints 79% against the strip's 88.1% — a nine-point wedge, and the cheap side of the trade.
- The ten-year touched 5.00% intraday for the first time since 2023, and leveraged funds are short 6.86 million Treasury contracts across the curve. It closed 4.98%, the 2-year 4.65%. The position that matters is not directional: CFTC leveraged funds are net short −2.07m 5-years, −1.94m 10-years, −1.29m 2-years and −0.86m ultra-bonds (TFF, 8 Sep) — the basis trade at full stretch, with asset managers on the other side. That is a plumbing exposure, not a macro view, and a disorderly move in either direction out of Wednesday is the way it breaks. Note also what the day did not do: with the 10-year at 5%, the S&P fell only about half a percent and the Russell outperformed.
- The Salalah talks did not happen — they were postponed with no new date, and Saudi Arabia, not Bahrain, was the objector. Oman and Iran announced the deferral "in the interest of consensus"; Iran's foreign ministry said Riyadh submitted amendments fearing the wording would establish a status quo it could not accept, after drone strikes on its territory. What was on the table was temporary entry and exit corridors — inbound through Iranian waters, outbound mostly Omani — funded by voluntary tolls. Separately the East–West pipeline is shut (Abqaiq to Yanbu, ~7mb/d of nameplate routing, ~700kb/d actually lost, repairs estimated at over a month, no restart date), with Iraq confirming the launch platform was seized in Maysan province. Crude-carrier transits through Hormuz were zero on four of the last five observed days, a 7-day average of 0.29/day (Tankermap). Brent settled $107.13, +2.41%. The premium is now paying for the absence of an expected catalyst rather than a new attack, which makes it stickier, not more fragile.
- A new factor entered the tape on Monday and it is not a rates factor: the AI-capex de-rating. Anthropic's Dario Amodei published an essay urging firms to "pace the rate of capabilities advancement," echoed by OpenAI's Altman. Semis and AI infrastructure fell 5–8% — Teradyne −9.3%, Coherent −8.68%, Nvidia −2.5% — while cybersecurity rallied (CrowdStrike +10.45%, Palo Alto +8.97%) and the Russell 2000 held up. The cleanest expression was offshore: KOSPI −2.43% (SK hynix −6%, Samsung −4%) and a 162 basis-point split between the Nikkei (−0.81%) and TOPIX (+0.81%) in a single session. This is a factor rotation inside the equity market, not a risk-off, and the desk closed its Korea/Taiwan semis trim last edition — which now looks a session early.
- The S&P is a quarter of a percent above the gamma flip going into the largest expiry on record, and the corporate bid is gone. SpotGamma puts dealer gamma at −$8–10bn with the flip at 7,600 and no gamma trough until ~7,350; spot closed 7,619–7,630. Friday's expiry is $6.2trn — 23% of total US options exposure on one day, surpassing June's $7.7trn record (Rubner, Citadel Securities, 31 Aug, verified against the primary). Buyback blackout accelerated ~12 September against $1.1trn of authorisations. BofA's current systematic estimate is −$126bn of selling under down paths against +$35bn of buying flat (4 Sep vintage, which supersedes the −$163bn/+$9bn figure carried in earlier editions), with CTAs near maximum long and triggers starting at declines of 1.4%.
- China's August credit missed badly and nobody priced it — new yuan loans came in at 15% of consensus. ¥60bn against ¥400bn expected (prior −¥340bn); aggregate financing ¥1,660bn against ¥2,040bn; outstanding loan growth 4.9% from 5.1%. A positive print that small is effectively no net credit creation, and the whole of the beat versus the negative prior came through non-bank channels. Iron ore is at $97.55/t, a fifth consecutive session below $100. The August activity data — industrial production, retail sales and fixed-asset investment, consensus −7.2% YTD y/y — lands at 12:00 AEST today, and Monday's Materials sector in Sydney fell only 0.55%. The read-through has not been taken.
Overnight recap
United States — an AI scare, a 5% handle, and a market that barely moved
Monday was the rare session where the two largest headlines pointed the same way and the index still declined less than one per cent. The S&P 500 closed in a 7,619–7,630 band depending on source (−0.36% to −0.50%; see section 13 — this is the edition's principal unresolved conflict), the Nasdaq Composite at 26,186, −0.6%, the Dow at 52,421–52,455 and the Russell 2000 at 2,898.27, −0.20%. The session opened at its lows (9:38 ET: S&P −0.56%, Nasdaq −0.84%), traded to −0.65% and −1.17% respectively by early afternoon, then pared into the close as yields backed off the round number. The Nasdaq Composite held its 100-day at 25,982.27; the Nasdaq-100 did not.
Two drivers. The first was an AI-safety shock with a capex transmission: Amodei's public call to "pace the rate of capabilities advancement," partially endorsed by Altman and xAI, was read as an industry-led threat to the spending line that has been carrying index earnings. Semiconductors and AI infrastructure took it hardest — Teradyne −9.3%, Coherent −8.68%, Hewlett Packard Enterprise −9.3%, Nvidia −2.5%, with AMD, Marvell, Intel, Lumentum, SK hynix and Samsung all lower — while CrowdStrike +10.45% and Palo Alto Networks +8.97% led a defensive bid into software and security. The second was oil and the long end: Brent traded to a four-month high near $110 intraday on the Salalah collapse and the Saudi pipeline outage before fading, and the 10-year Treasury touched 5.00% — the first 5-handle since 2023 — closing at 4.98% with the 2-year at 4.65%. Bloomberg's framing named both inflation and supply; with a 25-year-high auction stop already printed in August, the concession side of the story is real.
The absence of a bigger equity reaction is the informative part. A 5% ten-year, a 2.4% oil move and an existential headline for the market's largest earnings engine produced roughly half a percent on the index and a positive print for most of the day in small caps. That is a market that has already bought its protection for Wednesday, not one that is complacent about it — CBOE SKEW at 154.5 (96th percentile over one year) says the same thing. No first-tier US data was released; the Empire State survey (consensus 14.1, prior 20.6) is the only print before the Fed, at 22:30 AEST today.
Europe — a two-factor session, and one index that could not be sourced
Europe took the AI headline squarely and the oil headline unevenly. Euro Stoxx 50 6,260.30 (−1.02%), DAX 25,432.00 (−0.53%), CAC 40 8,115.10 (−0.79%), FTSE MIB 51,855.62 (−1.25%), IBEX 19,658.00 (−0.91%), against FTSE 100 10,692–10,698 (+0.39% to +0.44%) — the London outperformance is energy weighting, not a UK story. Technology fell 1.4–2.0% with Soitec −12.6%, ASM International −8.7%, Infineon −7.6% and ASML −5.2%; energy rose 0.4%. Every one of those index levels was reconciled against Friday's close before publication; the Stoxx 600 and the SMI could not be sourced at a close and are not printed.
The bond market was the cleaner signal: Bunds bear-flattened hard, 2y +7.2bp to 3.25% against 30y −3.4bp to 3.86%, a ten-basis-point flattening in 2s30s on a day with no euro-area data. That is the market marking up the ECB's terminal, not its term premium, and it lines up with Monday's speakers — Kazāks said the case is building for rates to turn restrictive, Nagel that mildly restrictive territory cannot be excluded, while Šimkus pointed explicitly at December as the meeting with projections. October pricing stays at 28–29%. The OAT sold off 4.9bp to 4.50% and the derived OAT–Bund spread widened to 97.0bp, a fresh high on that series; the dedicated spread source has not rolled past Friday's 89.8bp and the gap between the two is now 7.2bp, isolated entirely to the French leg.
Asia — Australia flat on an auction print, Japan split, Korea wrecked
The ASX 200 closed 8,749.9, +0.10%, with breadth still negative (460 advancers against 604 decliners) and the A-VIX giving back only 2.1% of Friday's 9.2% spike to 14.05. Health Care led (+1.52%), Information Technology lagged (−1.14%), Materials fell 0.55% and Financials rose 0.43% — of which 0.36pp was printed in the closing auction alone, on what Market Index described as minimal gains during regular trading. That matters because Friday's +1.08% in financials was the first session against the mechanism of this desk's largest open view, and the follow-through turns out to be an auction artefact rather than accumulated demand. ACGBs bull-flattened modestly, with 2s and 3s both failing to hold 5% and closing back at 4.98%, retracing less than a fifth of Friday's selloff.
Japan split violently along factor lines: the Nikkei 225 fell 0.81% to 63,492.99 while the TOPIX rose 0.81% to 4,061.18 — a 162bp spread in one session, the price-weighted AI complex against the broad value and financial base, with a BoJ hike to 1.25% expected Friday. The 10-year JGB held 3.00%, near a three-decade high. Korea was the region's casualty at −2.43%, a third consecutive decline, hit by both the AI de-rating and a direct terms-of-trade blow from Brent. China was quietly weak (CSI 300 −0.67%, Shanghai −0.07%) with Hong Kong up 0.45%, and the credit data landed badly. India was closed for Ganesh Chaturthi and reopens today into two sessions of accumulated negative news.
Market dashboard
Monday 14 September — cross-asset change
| Equities | Close | 1d | Fri 11 Sep | Note |
|---|---|---|---|---|
| S&P 500 | 7,619–7,630 | −0.36 / −0.50% | 7,656.98 | Two sources, ~10pt apart — see §13. Below the 50d, 100d and 200d |
| Nasdaq Composite | 26,186 | −0.6% | 26,333.04 | Single source. Held the 100d at 25,982.27; the NDX did not |
| Dow Jones | 52,421–52,455 | −0.23 / −0.29% | 52,573.29 | Best of the majors — least AI weight |
| Russell 2000 | 2,898.27 | −0.20% | 2,903.94 | Outperformed all day (+0.45% at 12:47 ET). Not a risk-off |
| VIX / VIX3M | no close | ~17.4 at 09:38 ET | 15.84 / 18.60 | Neither source posted a 14 Sep close. IVTS 0.852 as of Fri, day 109 of contango |
| Stoxx 600 / SMI | not sourced | ≈ −0.3% intraday | 639.10 / 13,775.27 | No close obtainable for either at this hour. Not printed |
| Euro Stoxx 50 | 6,260.30 | −1.02% | 6,325.13 | Ties exactly to Friday. Page carried three different % fields |
| DAX | 25,432.00 | −0.53% | ≈25,569 | Soitec −12.6%, ASM Intl −8.7%, Infineon −7.6%, ASML −5.2% |
| CAC 40 | 8,115.10 | −0.79% | 8,179.77 | Corroborated at 8,118.90 intraday |
| FTSE 100 | 10,692–10,698 | +0.39 / +0.44% | 10,650.44 | Energy weighting, not a UK story |
| FTSE MIB / IBEX | 51,855.62 / 19,658.00 | −1.25% / −0.91% | ~52,512 / 19,838.50 | Both tie to Friday this edition |
| Nikkei 225 | 63,492.99 | −0.81% | 64,011.34 | Exchange archive. Range 62,726.18–63,691.53 |
| TOPIX | 4,061.18 | +0.81% | 4,028.30 | 162bp split against the Nikkei — factor rotation, not a market move |
| Hang Seng | 24,918 | +0.45% | 24,806 | A second source printed 25,005 at the same stated %; discarded |
| CSI 300 / Shanghai / Shenzhen | 4,480.08 / 3,885.33 / 13,384.6 | −0.67% / −0.07% / −0.64% | 4,510.16 / 3,888.11 / 13,471.3 | AI/semi-led. HSI outperformed the mainland again |
| KOSPI | 6,741.85 | −2.43% | 6,910 | Worst in the region. SK hynix −6%+, Samsung −4%, SK Square −7.1% |
| TAIEX | 45,990.78 | −0.67% | 46,184.85 | Page narrative was stale and disregarded |
| Nifty 50 / Sensex | closed | — | 23,398.10 / 74,782 (disputed) | Ganesh Chaturthi. Reopens today into a two-session backlog |
| S&P/ASX 200 | 8,749.9 | +0.10% | 8,741.20 | 460 adv / 604 dec / 403 unch · A-VIX 14.05 (−2.12%) · Small Ords −0.61% |
| ASX SPI 200 futures | no clean read | — | — | Contract rolled to Dec (APZ2026) on 13 Sep. Page fields mutually inconsistent; deliberately not quoted |
| Rates & credit | Level | 1d | Week | Note |
|---|---|---|---|---|
| UST 2y | 4.65% | +2bp | +26bp (to Fri) | Vendor mark; official par curve for 14 Sep not yet posted |
| UST 5y | 4.78% (Fri) | n/a | +24bp | No Monday mark obtained — so 5s30s cannot be rolled forward |
| UST 10y | 4.98% | +2bp | +18bp (to Fri) | Touched 5.00% intraday — first since 2023. Four sources corroborate |
| UST 30y | 5.35% (Fri) | n/a | +11bp | Official par curve, 11 Sep. One vendor page carried 4.61% — discarded as corrupt |
| 2s10s · 5s30s | +33bp · +57bp (Fri) | ≈ unch · n/a | from +41 · +70 | Friday's session bear-flattened both: 2s10s −6bp, 5s30s −5bp |
| Bund 2y / 10y / 30y | 3.25% / 3.53% / 3.86% | +7.2 / +2.5 / −3.4bp | 2s30s flattened 10.6bp — terminal repricing, not term premium | |
| OAT 10y · OAT–Bund | 4.50% · 97.0bp derived | +4.9bp · +2.4bp | Dedicated series 89.8bp but stamped 11 Sep — a 7.2bp gap, all in the French leg | |
| BTP 10y · BTP–Bund · BTP–OAT | 4.41% · 88.0bp · −9.0bp | +6.0bp | Italy still trades through France | |
| Gilt 2y / 10y / 30y | 4.80% / 5.41% / 5.94% | n/a / +4.9bp / n/a | 10y near the highest since August 2007. CPI Wednesday, MPC Thursday | |
| JGB 2y / 10y / 30y / 40y | 1.85% / 3.00% / 4.07% / 4.11% | +1.2 / +1.1 / +2.1 / +0.1bp | 10y above 3% for the first time since 1996; near a three-decade high | |
| ACGB 2y / 3y / 5y / 10y | 4.98% / 4.98% / 5.01% / 5.34% | −2.3 / −2.5 / −2.6 / −2.4bp | 2s and 3s failed to hold 5%; 20y 5.71% (−3.6), 30y 5.78% (−3.8); 3s10s 36bp | |
| Canada 10y · Switzerland 10y | 3.972% · 0.56% (Fri) | +3.2bp · — | Swiss page stale at 11 Sep | |
| US IG / HY / CCC OAS | 80 / 270 / 1,070bp | obs 10 Sep | Four sessions stale. The 11 Sep observation does not exist — FRED, ALFRED and the corroborant all terminate at 10 Sep. See §13 |
| FX | Mon 14 Sep | 1d | Fri 11 Sep | Note |
|---|---|---|---|---|
| DXY | 99.43 | +0.31% | 99.12 | Three sources spanned 99.43–99.54 and "+0.1%"; the one reconciling to Friday is used |
| EUR/USD | 1.15418 | −0.50% | 1.1600 | Below the 1.1563 range floor — the house view's invalidation level |
| USD/JPY | 154.86 | +0.85% | 153.55 | Worst G10 performer into a Friday BoJ hike. Carry over policy |
| GBP/USD | 1.3480 | −0.34% | 1.3526 | CPI Wednesday, MPC Thursday |
| AUD/USD | 0.71408 | −0.42% | 0.71711 | Fell while ACGB yields also fell — trading the commodity complex and the dollar, not rates |
| NZD/USD | 0.5784 | −0.51% | 0.58135 | NZ Q2 GDP Thursday, consensus +0.1% from +0.8% |
| AUD/NZD | 1.2346 derived · 1.2353 vendor | +0.09 / +0.17% | 1.2335 | Cross-computed first: the vendor page was again serving Friday's data as live |
| AUD/JPY · EUR/JPY | 110.58 · 178.74 | +0.43% · +0.35% | 110.16 · 178.16 | Both derived from legs; both vendor cross pages stale |
| USD/CAD · USD/CHF | 1.3911 · 0.8181 | +0.28% · +0.19% | 1.3872 · 0.8165 | |
| USD/CNY · PBoC fix | 6.7068 · 6.7698 | −0.02% · 45 pips stronger | fix 6.7743 | Fix set 615 pips WEAKER than the 6.7083 estimate. Spot sits 630 pips strong of the fix — implausible; treat the spot as suspect |
| USD/MXN · INR · KRW | 17.1315 · 95.834 · 1,341.53 | +0.98% · +0.25% · — | 16.9657 · 95.599 · 1,341.53 | MXN the day's outlier loser. KRW page stamped 13 Sep — a Friday mark |
| Commodities & digital assets | Last | 1d | Week | Note |
|---|---|---|---|---|
| Brent (Nov) | $107.13 | +2.41% | wk to Fri +8.65% | O 107.54 / H 108.47 / L 106.87. Traded near $110 intraday. Path: 101.21 → 107.63 → 104.61 → 107.13 |
| WTI (front) | $102.44 | +2.39% | wk to Fri +9.37% | O 102.53 / H 103.58 / L 102.09 |
| Brent–WTI | $4.69 | +$0.13 | from $4.56 | Settle-based. Intraday vendor marks span $4.30–5.04 — an open house view, so the range matters |
| Diesel crack | $107.72 | obs 10 Sep | from $101.1 (4 Sep) | A record closing price, and the ten-day data gap is closed. 3-2-1 crack $62.73, gasoline crack $40.88 (both 9 Sep) |
| Henry Hub / TTF | $2.90 / €83.15 | +2.39% / +4.57% | −0.59% / +13.43% | European gas is the week's biggest energy mover |
| Gasoline / Heating oil | $3.43 / $5.07 | +3.83% / +2.29% | +5.57% / +11.05% | A second source ran 2–3% below on both; range noted |
| US retail diesel | $5.967/gal | +36.8c w/w | w/e 7 Sep | Correction: the "first time ever above $6" claim is refuted — the EIA ULSD print is three cents short. See §13 |
| Gold (spot, derived) | ≈$4,327 | −1.33 / −1.47% | $4,385.61 | Comex Dec settle $4,352.76 (−1.27%). Basis ≈$25–32, a normal carry — a face-value read gives $60–66 and repeats a known vendor artefact |
| Silver / Platinum | $62.91 / $1,767.90 | −2.11% / −1.65% | −4.88% / −4.06% | Silver's sign conflict did not reproduce this edition |
| Copper LME 3M | $14,065/t | −1.18% | −5.9% from the record | Three-week low. Section 232 refined-copper decision delayed again — the tariff premium is being priced out |
| Comex–LME arb | −$66 to −$198/t | sign flip | from ≈$0–30/t | Computed, not quoted. Comex now trades at a DISCOUNT to LME. Cash–3M a $36/t contango, from $436/t backwardation in mid-August |
| Aluminium / Zinc / Nickel | $3,256 / $3,807 / $16,380 | +0.1% / −1.9% / −0.5% | Two sources agree on levels within 0.15% | |
| Iron ore | $97.55/t | −0.48% | −7.37% y/y | Fifth consecutive session below $100. China activity data 12:00 AEST today |
| Lithium / Uranium | ¥138,650/t / $90.15 | −2.53% / — | −6.00% / — | A second consecutive −6% week in lithium. Uranium has no Monday print |
| Wheat / Soybeans | $7.30 / $12.97 | +0.60% / +0.07% | disputed | Levels do not reconcile with Friday's carried marks — see §13. Still no food-security bid |
| Bitcoin | $79,072.61 | +2.5% | +2.33% from Sun | ~1% below the twice-rejected $79,800–79,890. Dominance 57.2% |
| Ether | $2,540.45 | +1.4% | +1.36% from Sun | A weekday print above $2,500 on $79.9bn volume — the house view's pre-committed trigger. Closed, see §08 |
| SOL / XRP / BNB | $103.16 / $1.46 / $725.63 | +2.6% / +9.0% / +0.9% | XRP is the only idiosyncratic move; everything else traded as a beta bundle. Total cap $2.779trn | |
| Crypto derivatives | OI $64.7bn · liq. $143.6m | funding +0.0055% | Four editions of failure ends. Liquidations 0.22% of OI and near-neutral funding: the rally is not leveraged | |
| Spot ETF flows | BTC −$13.2m · ETH +$216.4m | 11 Sep | 14 Sep not published — the ETH table's zeroes are the placeholder, not a zero |
Conventions: 1d = change on Monday 14 September, recomputed from verified Friday 11 September closes rather than taken from vendor change fields — three vendors carried a phantom Sunday 13 September row that corrupts every printed daily percentage today. Yields in %, changes in bp. Gold is spot with the Comex December settle shown separately. Brent and WTI are front-month settles; the Brent front contract is November 2026 and has not rolled. "≈" marks a derived value; ranges are printed where sources disagree rather than a midpoint being invented.
What is driving markets
1. The energy shock has become a monetary shock — and the off-ramp just closed
The change since Friday is not a new attack; it is the removal of an expected de-escalation. The Iran–GCC meeting in Salalah was postponed with no new date, and the objector was Saudi Arabia, not Bahrain — Riyadh submitted amendments fearing the corridor wording would establish a status quo it could not accept, after drone strikes on its own territory. What died with the meeting was a concrete mechanism: temporary inbound corridors through Iranian waters and outbound through Omani, funded by voluntary tolls for navigation, environmental protection and search-and-rescue. Meanwhile the East–West pipeline is shut — Abqaiq to Yanbu, ~7mb/d of nameplate routing capacity and the primary Hormuz bypass — with a damaged pumping station, ~700kb/d of actual loss, repairs estimated at over a month and no restart date. Crude-carrier transits through Hormuz were zero on 10, 11, 12 and 14 September and one on the 13th, a seven-day average of 0.29 per day against a disputed pre-war baseline of 100–140 vessels. And the diesel crack made a record close at $107.72 on 10 September, participating fully in the crude rally rather than compressing — against US distillate stocks that ended August at the lowest level for the month since 1951.
2. The long end is where the stress lives — and now it has a 5-handle
The 10-year touched 5.00% intraday on Monday, closing 4.98%, with four independent sources corroborating the print. That is the first 5-handle since 2023, and it arrived on a day with no US data — which makes it a supply and term-premium event rather than an inflation event. A Treasury auction stop already hit a 25-year high in August. The global picture is the same trade in different currencies: JGB 10s above 3% for the first time since 1996 and near a three-decade high; gilt 10s at 5.41%, the highest since August 2007, into Wednesday's CPI; OAT–Bund at 97.0bp on the derived series, a fresh high, against Germany's 2027 budget carrying €118.7bn of core net new borrowing of which €85.4bn runs through the constitutional defence exemption — that exemption line, not the €33.4bn inside the debt brake, is the fiscal-dominance number. The one exception is Australia, where yields fell 2–4bp on the same day the US printed 5%.
What makes this dangerous rather than merely expensive is the positioning underneath it. Leveraged funds are net short 2.07m 5-years, 1.94m 10-years, 1.29m 2-years and 0.86m ultra-bonds — roughly 6.86 million contracts across the curve, reconstructed from the raw CFTC file and summed back to printed open interest on both sides. That is the basis trade, with asset managers holding the other leg. It is not a directional macro view and it does not care which way the market goes; it cares about the speed.
3. The crowd is long risk, short bonds, and — for the first time this month — has bought some protection
The positioning picture has moved in two directions at once. On the one hand the CFTC shows no equity de-risking at all as of 8 September: E-mini S&P leveraged-fund net short was essentially flat on the week, CTAs sit near maximum long with sell triggers that only begin at declines of 1.4–4%, and BofA's systematic monitor projects −$126bn of selling under down paths against +$35bn of buying in a flat market — a 3.6:1 asymmetry with no dry powder on the buy side. Gold is the single most crowded position in the entire report at 56.4% of open interest with gross longs 9.0× shorts, and it got more crowded on the week. On the other hand, options tell a different story: CBOE SKEW printed 154.5, the 96th percentile over one year and the 98th all-time, and the VIX traded up near 17.4 on a session where the index fell half a per cent. Somebody has paid up for the left tail.
Two corrections belong here. First, the "skew at the first percentile" figure carried since edition No. 001 is formally withdrawn — at 154.5 CBOE SKEW is nowhere near a low percentile, and the original number was most plausibly a 25-delta risk reversal, a different part of the surface entirely. Second, the flow data does not net: Lipper reported −$32.27bn of US equity fund outflows for the week to 9 September, including a record −$40.44bn in large-cap, while BofA's own client book saw +$7.0bn, its sixth-largest buying week since 2008. Those are different universes measuring different people. Fund investors sold; institutions bought.
4. AI capex is the earnings engine — and on Monday the threat moved from the financing side to the earnings side
This theme has run since edition No. 001 with the financing leg as the crack: Oracle at BBB−, hyperscaler CDS at records, capex outrunning free cash flow. Monday was the first session where the demand leg was attacked instead. Anthropic's Dario Amodei published an essay urging the industry to "pace the rate of capabilities advancement," partially echoed by OpenAI's Altman and by xAI, and the market priced it immediately as a threat to the capex line: Teradyne −9.3%, Coherent −8.68%, Hewlett Packard Enterprise −9.3%, Nvidia −2.5%, with the same trade wrecking Korea at −2.43% (SK hynix −6%+, Samsung −4%, SK Square −7.1%) and splitting Japan 162 basis points between the Nikkei and the TOPIX. The financing side meanwhile is about to be marked in public: SoftBank is roadshowing $10–20bn of dollar high-yield in New York from 14–17 September, BB+ rated, possibly with a euro tranche, while repaying a $25.9bn bridge on the 15th — and it will not price before the Fed. No tranches, tenors, spread talk, order book or leads have been disclosed.
The bullish counterweight is real and should not be waved away. FactSet's 11 September edition has Q3 earnings growth at +28.7%, revised UP 2.1pp intra-quarter, revenue growth 11.9%, a guidance ratio of 72 positive to 42 negative, and a forward P/E of 19.1× against a five-year average of 19.8×. The multiple is below its own recent average. The bear case is not valuation; it is that 19.1× on +28.7% growth is being defended with the ten-year at 5%.
5. The dots, not the decision
Everything above resolves into one number on Wednesday. The decision is 88.1% priced; the projections are not. The June SEP put the end-2026 median at 3.8% off a committee split 9–9 on whether to hike in 2026 at all — one participant at +75bp, five at +50bp, three at +25bp, eight unchanged, one at −25bp — while raising 2026 headline PCE to 3.6% from 2.7% and core PCE to 3.3% from 2.7% in a single meeting. Three months later the strip's modal December end-state is 4.00–4.25%. Arithmetic, not opinion: the new median has to print near 4.1% just to ratify current pricing. The 29 July hold was 9–3 with Hammack, Kashkari and Logan dissenting to hike; if the Committee moves this week, the interesting question is whether dissents appear on the dovish side for the first time in the cycle. Blackout was verified empirically rather than assumed — the last Board speech on the Fed's own 2026 listing is Waller on 3 September.
Central bank watch
Fed funds pricing — implied probabilities by meeting
| Bank | Policy rate | Last move / vote | Next decision (AEST) | Market pricing | Bias |
|---|---|---|---|---|---|
| Fed | 3.50–3.75% | Held 29 Jul, 9–3 (Hammack, Kashkari, Logan to hike) | Wed 16 Sep 14:00 ET · 04:00 Thu AEST · SEP + dots · presser 04:30 | 88.1% hike (14 Sep 07:45 ET); 83.2% cbw (11 Sep); 79% Polymarket event book. Cumulative ≥1 hike 93.6% Oct, 97.8% Dec; modal Dec 4.00–4.25% at 48.7%, 26.5% on three | Hawkish |
| ECB | DFR 2.50% / MRO 2.65% | +25bp 10 Sep, unanimous; no forward guidance | Thu 29 Oct 14:15 CET · 00:15 Fri 30 AEDT | October 28–29%, unchanged. Kazāks 14 Sep: case building for restrictive rates. Šimkus points at December. The 92.4% tracker remains discarded — it still shows a 2.25% rate | Hawkish on terminal, not October |
| BoJ | 1.00% | Held 31 Jul 8–1 — the dissent was hawkish (Takata, wanted 1.25%) | Fri 18 Sep · ~13:00 AEST (MPM runs 17–18) | ~72% (OIS, 14 Sep), with a 97% outlier. Prior 62% reading discarded — that page shows a 0.50% policy rate. Reuters: 1.25% with "no preset idea on the terminal rate" | Hike base case, dovish-hike risk |
| BoE | 3.75% | Held 30 Jul, 6–3 | Thu 17 Sep · 21:00 AEST · no MPR | Hold 73.1% / hike 26.9% (as of 11 Sep — has not refreshed). ~53bp priced over three meetings; Dec 17 at 100% higher. CPI Wed: 3.1% headline, 2.7% core | Hawkish hold; the vote is the trade |
| RBA | 4.35% | On hold since 11 Aug | Tue 29 Sep · 14:30 AEST | 76% (centralbank.watch, 14 Sep, direct-fetch) — unchanged from 13 Sep. Nov 52% higher. The 85% cached tracker again uncorroborated and again discarded. Terminal 4.60% | Hike base case |
| RBNZ | 2.75% | +25bp 2 Sep (2nd straight) | Wed 28 Oct · 12:00 AEDT | Roughly even odds of more. NZ Q2 GDP Thu, cons. +0.1% from +0.8% | Tightening, decelerating |
| BoC | 2.25% | Held 2 Sep — 6th consecutive | Wed 28 Oct (low confidence — one source says 19 Oct, which is the survey release) | Summary of deliberations Wed | Neutral, energy-alert |
| SNB | 0.00% | Held 18 Jun | Thu 24 Sep · ~17:30 AEST (date unconfirmed from the SNB) | Conditional forecast assumes 0% throughout; first hike ~mid-2027 | Extended hold |
| Norges | 4.25% | Last decision 13 Aug | Thu 24 Sep · ~18:00 AEST + MPR | Confirmed from the bank's own site — the carried-unverified line is closed | Hawkish hold |
| Riksbank | 1.75% | Held 20 Aug: "the probability of a rate increase later this year remains" | Thu 24 Sep · ~17:30 AEST (two secondary sources; riksbank.se would not render) | Meets four to five days after Swedish certification, into a one-seat parliament and possibly no government | Hawkish hold, political risk |
| PBoC | LPR 3.00% / 5y 3.50% | 15th month unchanged | LPR Mon 21 Sep | Fix 6.7698, 615 pips weaker than estimate, into a badly missed credit print — a deliberate stability signal | Easing bias |
| Emerging markets | |||||
| Brazil (Copom) | Selic 14.00% | 4th straight −25bp | Meets 15–16 Sep · decision Thu 17 ~07:30 AEST | ~95% for −25bp to 13.75% (B3 options, 10 Sep) | Easing |
| Mexico (Banxico) | 6.50% | Held 6 Aug; last cut 7 May, 3–2 | Thu 24 Sep · ~05:00 Fri AEST | MXN spec long +82,101 LF / +94,732 legacy. Worst G10-plus performer Monday at −0.98% | Restrictive hold |
| India (RBI) | Repo 5.25% | 4th hold 5 Aug, neutral stance | Wed 7 Oct | USD/INR 95.83 is the constraint. Market reopens today | Neutral |
| Korea (BoK) | 3.00% | +25bp 27 Aug — first back-to-back since early 2023 | Thu 22 Oct | Sep 1–10 exports +83% y/y, chips +270%, against a −2.43% equity session | Tightening |
| Indonesia (BI) | 5.75% | Held 19 Aug, 2nd straight | Wed 23 Sep · ~17:20 AEST | Rupiah-defence hold | Hold |
| Turkey (CBRT) | 37.00% | Held 10 Sep — 5th consecutive | Thu 22 Oct | Blocked from cutting by the energy shock; banks eye October | Restrictive hold |
Fed detail. See theme 5 for the dots arithmetic. Three procedural points. Blackout is confirmed, not assumed: the Fed's own 2026 speech listing ends at Waller on 3 September, with Barr on 1 September and Warsh at Jackson Hole on 28 August before that. The presser time of 14:30 ET is convention — the Fed publishes the date, not the hour. And the status of Governor Cook is genuinely unresolved and is the most consequential open item on this desk: the Supreme Court ruled 5–4 in her favour on 29 June, holding that a governor is removable only for cause and entitled to notice and an opportunity to respond, after which the administration served that notice in early August with a three-week response window. That window expired in late August and the outcome is unsourced. We cannot confirm she is seated and voting this week.
BoJ detail — the sixth consecutive edition of the same third-party error. The Bank's own schedule states the September MPM runs 17–18 September and the statement is released on Friday the 18th. Both calendars checked today — one of them a major aggregator, the other a central-bank tracker — printed Thursday the 17th, which is the meeting start. Substantively: the July vote was 8–1 with the dissent hawkish (Takata, for an immediate 1.25%), which inverts the framing carried in earlier editions; Asada remains the board's named dove and has set an explicit bar of demand-driven inflation. The binding constraint is not the decision but the calendar around it — Tokyo is shut Monday to Wednesday 21–23 September, so a Friday hike repricing has to clear through offshore venues into a four-day domestic gap.
RBA detail. Pricing is 76% and has not moved between 13 and 14 September, which is itself worth noting in a week where the Fed repriced. NAB alone calls September; ANZ, CBA and Westpac all call November, and all four sit at a 4.60% terminal. Deutsche Bank, UBS and Morgan Stanley call the 29th. Westpac's Luci Ellis has said a September move "remained possible" but that the Board would "strongly favour November." The Bullock event is now pinned: a parliamentary committee appearance on Friday 18 September — the single source printed "Friday (September 20)", but 20 September is a Sunday, and the same article's demonstrable two-day error on the Hunter event (it said "Monday (September 16)" for remarks actually delivered on Monday the 14th) resolves the weekday as correct and the date as wrong. Time and committee remain unconfirmed. It is the last scheduled Governor communication before the decision. Meanwhile rba.gov.au has served roughly three-month-stale content for a fifth consecutive edition; every RBA item here is routed through news coverage, and silence on that site is not evidence of absence.
Regional briefs
United States
A 5% ten-year, an AI scare and an oil spike moved the index half a per cent — the market is already hedged for Wednesday. No first-tier data landed Monday; the Empire State survey (cons. 14.1, prior 20.6, at 22:30 AEST today) is the only print before the Fed, and it has beaten by six points or more in four of the last six months, which makes a hot number thirty hours ahead of a live meeting asymmetric for the front end. Retail sales follow Wednesday (cons. +0.8–0.9% from −0.6%). The Cleveland Fed's September nowcasts have core CPI at +0.20% m/m but core PCE at +0.28% m/m and 3.49% y/y — the SEP's variable is running a full point above the one that made the headlines on 11 September. Fiscally there is no autumn cliff: CRFB confirms funding runs to 11 December, alongside VA extenders, farm bill provisions and surface transportation. Tariffs continue to migrate from the struck-down IEEPA authority to Section 232, with steel, aluminium, copper and patented pharmaceuticals all now under 232 orders — and the refined-copper decision was delayed again on Monday, which is why copper fell 5%.
Euro area
The Bund flattened ten basis points in 2s30s on a day with no data — that is the terminal being marked up, not the term premium. Kazāks said the case is building for rates to turn restrictive; Nagel that mildly restrictive territory cannot be excluded; Dolenc that markets do not exclude a 3.25% deposit rate by mid-2027. Against that, Šimkus pointed at October as an assessment meeting and December as the one with projections, and Econostream's tone meter shows the hawkish bias easing post-decision. Both readings are consistent: the hawkishness is about the terminal, not the next meeting, and 28–29% for October looks well calibrated. Germany's 2027 budget is the fiscal story — €555.4bn of spending, €118.7bn of core net new borrowing split €33.4bn inside the debt brake and €85.4bn through the constitutional defence exemption, plus €54.9bn of infrastructure fund and €30bn of Bundeswehr special fund, for ~€203.6bn in total. Defence spending is €109.7bn, up €27.1bn. Euro-area industrial production (cons. −0.5 to −0.7% m/m) and ZEW land today and tomorrow.
United Kingdom
Gilt 10s at 5.41% are the highest since August 2007, and the MPC vote on Thursday matters more than the level. A hold at 3.75% is 73.1% priced — though that reading is stamped 11 September and has not refreshed through two sessions — with roughly 53bp of tightening across the next three meetings and December fully priced higher. Wednesday's CPI is the input: consensus 3.1% headline from 2.9%, and 2.7% core from 2.6%. Today brings the labour market: unemployment 4.9–5.0% (sources disagree), average earnings ex-bonus 3.5% from 3.4%, claimant count +8.3k from −11.0k. There is no Monetary Policy Report at this meeting, which is confirmed from the Bank's own calendar — the 2026 MPR meetings are February, April, July and November. Fiscally the Budget is 28 October and the headroom question is settled enough to trade: against the OBR's last official £22bn, the credible working range is £8–11bn, with the Resolution Foundation at £8bn in August and Bloomberg describing it as halved in September. Gilt yields have risen since both estimates, so £8bn is the live end.
Japan
A 162 basis-point Nikkei–TOPIX split in one session is the cleanest picture of the AI de-rating anywhere in the world. The Nikkei's price-weighted technology complex fell 0.81% to 63,492.99 on the Amodei headline while the TOPIX's broader value and financial base rose 0.81% to 4,061.18 on the BoJ hike expectation. For AUD/JPY and regional risk, TOPIX is the signal and the Nikkei is the AI beta. JGBs sold modestly across the curve with 10s holding 3.00% for the first time since 1996 and 30s at 4.07%. Friday's decision is expected to take the policy rate to 1.25%, the highest since April 1995, with US Treasury Secretary Bessent having publicly pressed for more aggressive tightening and the imported energy cost from the pipeline outage compounding the domestic case. The yen was the worst G10 performer on Monday at −0.85% despite that — carry demand and the dollar bid dominating a majority-priced hike. Per the standing rule, no Japanese single-stock or sector-index relative value is opened or quoted: those inputs have been unsourceable for six consecutive editions and the sector proxy served ten-month-stale data.
China & Hong Kong
August credit came in at 15% of consensus and the market has not taken the read-through. New yuan loans ¥60bn against ¥400bn expected (prior −¥340bn); total social financing ¥1,660bn against ¥2,040bn; M2 7.5% against 7.6%; and outstanding loan growth down to 4.9% from 5.1%. A positive print that small is effectively no net credit creation, and the improvement on the negative prior came entirely through non-bank channels. Credit impulse is the binding constraint on the steel and iron-ore complex, which is why iron ore is at $97.55/t on a fifth consecutive sub-$100 session. The August activity data lands at 12:00 AEST today: industrial production cons. 4.8% (prior 4.5%), retail sales 0.8% (0.6%), fixed-asset investment −7.2% YTD y/y from −6.7% — marked down further from the −6.7 to −7.0 band the desk was working with — plus house prices at 11:30. Watch the gap: consensus has IP running six times the pace of retail sales, the production-over-consumption imbalance that has capped the complex all year. Hong Kong outperformed the mainland again (HSI +0.45% against CSI 300 −0.67%), with Tencent +0.7%, Lenovo +3.5% and Xiaomi +2.7% against AI names Z.AI −7.1% and MiniMax −4.5%.
Emerging Asia & LatAm
Korea took the worst of both shocks and is the region's cleanest AI proxy. KOSPI −2.43%, a third consecutive decline, driven by the semiconductor de-rating (SK hynix −6%+, Samsung −4%, SK Square −7.1%, with SK Group's chairman himself warning that excessive AI spending becomes a problem if returns do not materialise) and by a direct terms-of-trade hit from Brent to an oil importer. Hyundai Motor −3.5%. That is the same index whose first-ten-day September exports rose 83% y/y with chips up 270% — the tension this desk closed a view on last week, and the argument has not changed even though the price has. Taiwan −0.67% to 45,990.78. India was closed for Ganesh Chaturthi and reopens today into two sessions of accumulated negative news, with the Nifty at 23,398.10 near three-month lows, the Sensex level disputed between 74,782 and 74,872 on a single page, and USD/INR at 95.83. Mexico was the day's outlier FX loser at −0.98%, with spec MXN length large on both CFTC series. Brazil's Copom meets today and tomorrow with ~95% priced for a −25bp cut to 13.75%.
Australia & New Zealand
The bank question — and the answer is "artefact"
Friday's +1.08% in Financials was the first session against the mechanism of this desk's largest open view, and the honest test was whether it followed through. It did, nominally: Financials +0.43% on Monday, a second consecutive gain. It does not survive inspection. Market Index reports that +0.36pp of that 0.43% was printed in the closing auction alone, "despite minimal gains during regular trading" — roughly 84% of the move manufactured in the final match rather than accumulated through the session. And the mechanism that supported Friday's bid reversed: Friday's bank rally coincided with ACGB yields up 11–14bp, while Monday's coincided with yields down 2–4bp. So the yield-artefact explanation is dead, and what replaces it is worse — an auction print is not demand at all. Only one individual bank close is independently sourced (Westpac +0.9%); CBA, NAB and ANZ are not. Two sessions do not make a turn, and the second of the two is not a real session. The underweight is not yet challenged by the evidence.
Elsewhere the tape was two-tier. The index rose 0.10% to 8,749.9 on negative breadth (460 advancers, 604 decliners) with the Small Ordinaries −0.61% against a flat large-cap index — the recovery is confined to index-level defensives. Health Care led at +1.52% (CSL +2.7%, Telix +4.3% on an FDA approval); Information Technology lagged at −1.14%. The A-VIX gave back only 2.1% of Friday's 9.2% spike, to 14.05, and remains at the top of its three-month range. Materials fell 0.55% with BHP −0.5%, but the sector split hard by commodity: gold miners bid (Northern Star +2.5%, Catalyst +6.1%), lithium wrecked on GFEX futures more than 35% below the May peak (Mineral Resources −4.4%, IGO −3.4%, Pilbara −2.4%), and uranium worse still (Deep Yellow −8.8%, Paladin −6.9%, NexGen −4.5%). Idiosyncratic: Core Lithium +13.0%, FleetPartners +12.1%, Lovisa +5.7%; Star Entertainment −10.3%, Megaport −8.3%, Chorus −5.6% to a 52-week low. A Calvary-led consortium won approval to operate the 25 remaining Healthscope hospitals.
Rates, the currency, and no usable SPI
The 5% handle did not hold. ACGB 2s and 3s both closed back at 4.98% after Friday's first-of-cycle break, retracing less than a fifth of the 13–14bp selloff; the curve bull-flattened modestly (30y −3.8bp against 2y −2.3bp) and 3s10s sits at 36bp from 35bp. The striking relative-value fact is that Australian yields fell on the same session the US ten-year printed 5.00% — a gap unlikely to survive a Fed on Wednesday and a BoJ on Friday. The 10-year at 5.34% remains the highest since May 2011.
AUD/USD fell 0.42% to 0.71408 while ACGB yields also fell, which says the currency is trading the commodity complex and the dollar rather than the rate differential. AUD/NZD ground higher to 1.2346 derived and 1.2353 vendor — and the vendor cross page was, once again, serving Friday's data as live, caught only because the cross was computed from its two legs first. AUD/JPY is flat at 110.58 into Friday's BoJ, which makes it the most exposed Australian cross on the book.
There is no usable SPI indication and none is printed. The contract rolled from September to December (APZ2026, expiry 18 December) on 13 September. The futures page's fields are mutually contradictory — last 8,805.0, previous close 8,805, change +17.0, day range a single point, timestamp "Closed at 12/09" — and the 8,805 print against an 8,749.9 cash close is December carry, not a +55-point indication. The alternative venue showed no trades at levels 250–300 points stale. The honest substitute is the overnight direction: a soft-to-flat open with renewed pressure on technology and long-duration names.
The RBA, and the last data before the decision
Pricing is 76% for 29 September and has not moved in two sessions. Chief Economist Sarah Hunter spoke Monday — businesses and households "well placed," most mortgagees holding good buffers, with AI-related price pressure flagged on consumer electronics. That is a hawkish-tolerant framing consistent with the priced move. The correction from No. 006 is confirmed and matters: August Labour Force publishes Thursday 24 September at 11:30, five days before the meeting — consensus employment +15.1k from −15.8k, unemployment 4.5% unchanged, participation 66.9%. The Monthly CPI Indicator publishes 30 September, the day after. So Labour Force is the last major domestic input the Board sees, and a repeat of July's −15.8k is the only realistic dovish off-ramp. Nothing is scheduled for Australia today; the Westpac Leading Index (Wednesday, prior 0.0%) is second-tier, and both the consumer sentiment and NAB business surveys already printed on Friday at 84.4 (−5.2%) and −5 (first negative in six years).
Property and New Zealand
Clearance rates and prices are pointing in opposite directions, and the clearance number is the less reliable one. National preliminary clearance for the 12–13 September weekend was 58.5%, a 19-week high; apply the standard 5–10pp preliminary-to-final haircut and the final lands at 49–54%, still below a balanced market. Domain's own city rows give a combined 53.2% (Melbourne 60%, Sydney 53%, Brisbane a standout 29%). But that is a flow measure responding to collapsed supply: 1,615 homes were scheduled, up 12.9% on the prior week but 32.8% below the same week a year ago. The stock measure is still deteriorating and broadening — Cotality's August index was −0.9% m/m, −3.1% q/q and −3.6% from the March peak, a fifth consecutive monthly fall, with 93% of capital-city suburbs recording a decline through winter and only Darwin rising. With a hike three-quarters priced, a 19-week high in a preliminary clearance rate is not a floor.
New Zealand was the week's worst major at −0.51% on the day to 0.5784. The RBNZ is at 2.75% after a second consecutive hike on 2 September and does not meet until 28 October. Q2 GDP on Thursday is the event, with consensus at +0.1% from +0.8% — an eight-tenths deceleration that is the substance behind the trans-Tasman policy gap this desk is long.
| Australia — the run into 29 September | Latest | Consensus / prior | Date (AEST) |
|---|---|---|---|
| Cash rate | 4.35% | 76% priced for +25bp | Tue 29 Sep 14:30 |
| Westpac Leading Index m/m (Aug) | 0.0% | +0.2% f'cast | Wed 16 Sep |
| RBA Governor Bullock — parliamentary committee | — | last Governor communication before the decision | Fri 18 Sep (time and committee unconfirmed) |
| S&P Global flash PMIs (Sep) | Mfg 52.0 / Svcs 53.2 | 51.7 / 52.5 f'cast | Tue 22 Sep |
| Labour Force (Aug) | −15.8k · 4.5% | +15.1k · 4.5% · part. 66.9% | Thu 24 Sep 11:30 — the last major input |
| Monthly Household Spending (Aug) | — | — | Tue 29 Sep 11:30, meeting morning |
| Monthly CPI Indicator (Aug) | 3.5% hdl · 3.6% trimmed | — | Wed 30 Sep 11:30 — the day after |
| Consumer sentiment · business conditions | 84.4 (−5.2%) · −5 | first negative conditions in six years | both printed Fri 11 Sep |
| Iron ore · Cotality home values (Aug) | $97.55/t · −0.9% m/m | 5th sub-$100 session · 5th monthly fall | China activity 12:00 today |
House views & tactical framework
| Asset | Bias | Conv. | Horizon | Rationale | What changes the view |
|---|---|---|---|---|---|
| Rates | |||||
| US 5s30s | Steepener | Med | 1–3 mo | Unmarked Monday — no 5y or 30y print obtained and the official par curve has not posted. Last mark 57bp from 61bp a week ago and 70bp on 4 Sep. The structural fiscal case is intact; the near-term driver has switched off | A close inside 45bp — twelve basis points away, and the stop will be honoured |
| ACGB 3s10s | Flattener | Low | 1–2 mo | Working, 36bp from a 43bp entry, though one basis point against on the session. Monday bull-flattened the long end (30y −3.8bp vs 2y −2.3bp) as 2s and 3s failed to hold 5% | Dovish RBA with a sticky 10y; a China stimulus impulse steepening the long end |
| OAT–Bund | Widener | Low | 1–3 mo | Working on the derived series, 97.0bp from ≈94bp, +2.4bp Monday. The dedicated spread source has not rolled past 89.8bp (11 Sep) and the 7.2bp gap is isolated entirely to the French leg — different benchmark selection, not a data error. "Widest since 2012" remains withdrawn; a one-year high is verified | A compression inside 80bp; a credible French consolidation; a dovish ECB October |
| Equities | |||||
| S&P 500 | Neutral, hedged; cut beta 12–18 Sep | Med | 2–4 wk | Working — 7,619–7,630, roughly −1.2% from a 7,718.60 entry, and this is the window. Below the 50d (7,669.98), 100d (7,675.80) and 200d (7,704.32), which are compressed into a 34-point band directly overhead. Gamma flip 7,600, now ~0.3% below spot; $6.2trn expires Friday; blackout on; −$126bn systematic selling against +$35bn buying | Clean break above 7,817 with breadth; a dot plot stopping at 4.00–4.25% |
| ASX 200 | Underweight tactically | Med | 2–4 wk | Still the best view on the book — 8,749.9, −2.84% from 9,005.9. And the caveat raised on Friday resolves in the view's favour: Financials' +0.43% was 84% printed in the closing auction with minimal regular-session gains, while the yield support that explained Friday's bid reversed. Breadth still negative; A-VIX 14.05 barely off a three-month high; Small Ords −0.61% | RBA hold 29 Sep; iron ore reclaims $100; banks stabilising on real intraday demand, not an auction print |
| China / HK | Neutral; H over A | Low | 2–4 wk | Zero of three on the trigger count, again. Monday the HSI (+0.45%) outperformed CSI 300 (−0.67%), Shanghai (−0.07%) and Shenzhen (−0.64%) — a second consecutive session of H-over-A. But August credit missed at 15% of consensus and the activity data lands at noon today | Three consecutive H-underperforming sessions — from zero |
| FX | |||||
| AUD/USD | Bias higher (0.7250–0.73) | Low | 2–4 wk | Worse — 0.71408, −0.78% from a 0.7197 entry, a fourth session below. And the diagnosis from Friday is confirmed rather than repaired: the currency fell while ACGB yields also fell, so it is trading the commodity complex and the dollar, not the differential the thesis rests on. RBA pricing unchanged at 76% | A close below 0.7100 — 41 pips away. A reclaim of 0.7226 restores Med |
| USD/JPY | Short — size cut | Med | 1–2 mo | Gave back most of the gain — 154.86, only −0.72% from 155.98 after −1.56% on Friday. The yen was the worst G10 performer into a ~72%-priced hike, which is exactly what the broken positioning predicted: LF shorts halved to −49,098 and legacy non-commercials flipped net long +10,796, a +103,023 swing. The fuel is gone and the price is now telling you so | A hawkish Fed paired with a dovish-hike BoJ. 152.00 is the level. Size for the 21–23 Sep Tokyo closure — gap risk, not drift |
| AUD/NZD | Long | Low | 1–2 mo | Working — 1.2346 derived / 1.2353 vendor against a ≈1.2315 entry. Cross-computed from legs first, which again caught the vendor page serving Friday's data as live. NZ Q2 GDP Thursday, cons. +0.1% from +0.8%, is the event | An RBA hold 29 Sep; a hawkish RBNZ 28 Oct; a China shock hitting Australia harder |
| DXY | Neutral, two-way | Low | 2–4 wk | 99.43, +0.31%. The dollar rose on a day the 10-year touched 5% — a more normal rates–FX correlation than the seven-basis-point response to a 26bp move in 2s that this view flagged on Friday. Three sources spanned 99.43–99.54 and "+0.1%" | Cook removal action; a dot plot that extends the path materially |
| Commodities | |||||
| Brent | Residual call spread only — no new risk | Low | 1–3 mo | $107.13, and the $105 add-condition is now exceeded on price as well as on the facts. Friday's note said the condition was "arguably met on the facts and refuted by the price"; the price has stopped refuting it. But adding at a four-month high, thirty hours before an 88%-priced Fed and with the re-own level $12 below, is a level trade — and this ledger's own pattern is that level trades do not work. Not adding, and saying so | Re-own outright at $92–95. The $105 add-condition is retired rather than triggered |
| Brent–WTI | Long the spread | Low | 1–3 mo | Working, marginally — $4.69 from a $4.56 entry. The thesis got stronger overnight: Salalah postponed, the East–West pipeline shut, crude-carrier transits at zero on four of five days. Honest caveat: intraday vendor marks span $4.30–5.04, so the 13-cent gain is inside the noise band of the inputs | A Salalah agreement restoring transits (spread to ~$3); or evidence US distillate tightness is bidding WTI on its own merits — which the record $107.72 diesel crack makes more plausible, not less |
| Gold | Long | Low | 1–3 mo | Back below water and close to the stop — ≈$4,327 derived against a $4,355.80 entry, −0.66%, with the $4,300 invalidation only ~$27 away. Comex Dec settled $4,352.76; the basis is ≈$25–32, a normal carry, and a face-value vendor read giving $60–66 repeats a known artefact. Crowding got worse, not better: 56.4% of open interest, gross longs 9.0× shorts, net +3,836 on the week | A close below $4,300 without a bid |
| Copper | Neutral — deliberate no-position | Low | 1–3 mo | Vindicated a second time, and harder. $14,065/t, −1.18% and −5.9% from Thursday's record, on the Section 232 refined-copper decision being delayed again. The arb has now flipped sign to −$66 to −$198/t — Comex trades at a discount to LME — and cash–3M is a $36/t contango against $436/t of backwardation in mid-August | The arb re-widening through +$300/t; Commerce setting an actual report date. A negative arb makes the 232 option cheaper still; still not opening it — no report date, and two prior errors building copper trades on policy optionality |
| Iron ore | Fade above $100 | Low | 1–3 mo | Working — $97.55/t, a fifth consecutive session below the handle, −7.37% y/y. The catalyst arrived early and confirmed the thesis: August new yuan loans at ¥60bn against ¥400bn, outstanding loan growth down to 4.9% | Pre-National Day restocking sustaining $105+; property stimulus. Today's FAI print at 12:00 AEST, cons. −7.2% YTD |
| Credit & digital assets | |||||
| US credit | UW HY/CCC; prefer 3–5y IG | High | 1–3 mo | The highest-conviction view on the book is now unmarked for four sessions and that is a problem. IG 80 / HY 270 / CCC 1,070bp, all observation 10 September. FRED, ALFRED and the independent corroborant all terminate there — verified as a genuine data absence, not a fetch failure. The thesis is intact (CCC wider at every observation since 1 Sep with zero retracement, against IG tighter) but it spans neither the post-CPI session nor the pre-FOMC session | The mark is SoftBank's $10–20bn HY book — and it will not price before the Fed. Concentrate the expression in CCC; the HY index is dead weight |
| Bitcoin | Range $78–83k; buy $76–78k | Low | 2–4 wk | $79,073, back inside the range and ~1% below the twice-rejected $79,800–79,890. Derivatives finally marked and they are quiet: funding +0.0055%, liquidations $143.6m against $64.7bn of open interest — 0.22%. The move is not leveraged. Two binaries inside 24 hours: CLARITY cloture 04:15 AEST today, FOMC 04:00 AEST Thursday | A volume break above $83k; an FOMC hike toward $74k; cloture failing |
| Closed this edition | |||||
| EUR/USD V011 | CLOSED — wrong | Low | opened 7 Sep | Opened neutral in a 1.1563–1.1700 range at 1.1613, with "a close outside the range" as the published invalidation. It closed at 1.15418 — 21 pips through the floor. The range survived the ECB and the CPI and broke on a day with no euro-area data at all, on a broad dollar bid | Closed at its published trigger. Not re-opened as a directional short — the view was that the range would hold, and the lesson is not "therefore sell euros" |
| Ether V019 | CLOSED — wrong | Low | opened 7 Sep | Opened neutral, capped at $2,500, at $2,518, with a pre-committed trigger written in No. 006: "a weekday close above $2,500 on normal volume closes this view." ETH is at $2,540.45 on a Monday with $79.9bn of total crypto volume against Sunday's $54.5bn. That is a weekday, above the cap, on normal volume. Supported by a $216.4m ETH ETF inflow on 11 September, its strongest September day | Closed on the trigger as written, with no argument about whether the volume "really" counted. Eight rejections of $2,500 and then a clean break is exactly the shape a cap view is supposed to respect |
Nothing was opened, and that is the decision. Two candidates were live. The diesel crack is finally markable — RBN published on Monday and the series now has a record close of $107.72 on 10 September plus an independent $106.44 for the 9th, against the $101.1 of 4 September this desk has been unable to score for three editions. It is not opened, because the freshest mark is still four days old and because entering a trade at its record high, thirty hours before an 88%-priced Fed, is a level trade. The copper 232 option is cheaper than ever now that the arb has gone negative, and is not opened for the reason given in the table. More broadly: opening new risk into an SEP meeting where the median has to move 30bp to ratify what is already priced is the definition of paying for someone else's catalyst.
Scorecard. Seventeen open, nine closed: two right, four wrong, three scratch — two of six on decided views, down from two of four. That is a deterioration and it should be read as one. Both of this edition's closes were losses, both were at published triggers, and neither was argued with. Working: V006 ASX (−2.84%, and the counter-evidence dissolved on inspection), V016 iron ore (fifth sub-$100 session with the catalyst confirming), V005 S&P (−1.2% into its own window), V022 copper (a no-position that has now avoided two separate drawdowns), V004 ACGB flattener, V023 AUD/NZD, V025 OAT–Bund, V026 Brent–WTI. Against: V003 5s30s (twelve basis points from its stop and unmarkable on the session), V009 AUD/USD (a fourth session below entry, 41 pips from invalidation), V014 gold (·$27 from invalidation and the most crowded position in the market).
Portfolio-level read. Three of the four views closest to their stops — gold, AUD/USD, the 5s30s steepener — are not failing because the analysis was wrong but because the expression was a level. That is the third consecutive week the same pattern has shown up, and it now has enough repetitions to be treated as a rule rather than an observation: this desk's mechanism trades work and its level trades do not. The corollary for tonight is specific. The Fed's decision is priced; the projections are not; and the only outcome with no positioning behind it is a median that confirms one hike rather than two. Carry less gross into 04:00 AEST Thursday than you did into Monday, keep the convexity, and note that the view most in need of a smaller number — gold at 56.4% of open interest — is also the one closest to its invalidation. If it goes, it will go fast, and not because of anything in the dot plot.
These are analytical framings for a professional reader, expressed in the vernacular of a macro desk. They are not personalised investment advice, carry no position sizing, and do not consider any individual's objectives or circumstances. Every view is logged, triggered and scored in the project's views ledger.
Positioning, flows & sentiment
| Indicator | Latest | Change / context | Read |
|---|---|---|---|
| CFTC — report date Tue 8 Sep, published Fri 11 Sep. Every contract below was reconstructed from the raw file and summed back to printed open interest on both sides before publication | |||
| Japanese yen — LF (TFF, CME) vs legacy non-comm. | −49,098 vs +10,796 | from −102,188 and −92,227 | The week's event. LF covered +53,090; legacy swung +103,023 (gross longs +61,622, shorts −41,401). The structural short is gone, which removes the shock absorber into the Fed |
| Australian dollar — LF vs legacy | +49,779 vs −34,870 | opposite signs | The cleanest illustration of the series trap — legacy aggregates LF plus other reportables and index traders. Same divergence in GBP (+34,627 / −58,836) and NZD (−17,350 / +6,232) |
| EUR · CAD · CHF · MXN (LF) | −33,285 · −55,448 · −13,440 · +82,101 | legacy agrees in sign on all four | MXN long on both series (+94,732 legacy) into a −0.98% peso session |
| US dollar index (ICE Futures U.S., legacy) | +17,604 | OI +7,838, of which +5,405 spreading | Directional add only +579 net — not a real dollar-length build |
| Treasury futures — LF net (TFF, CBT) | −6.86m contracts | 2y −1.29m · 5y −2.07m · 10y −1.94m · ultra-10 −0.43m · 30y −0.28m · ultra-bond −0.86m | The basis trade at full stretch, asset managers on the other leg. Not a macro view — a plumbing exposure, and the 5-year carries the largest single short with LF adding +68,990 of length into the week |
| E-mini S&P · Nasdaq mini (LF vs legacy) | −341,104 / −76,036 · −31,872 / +20,895 | both essentially flat w/w | No equity de-risking at all as of 8 September — consistent with CTAs near maximum long. This is the vulnerability |
| Gold (COMEX legacy) | +231,960 = 56.4% of OI | net +3,836 on the week; gross longs 9.0× shorts | Still the most crowded position in the report, and it got worse. Longs +522, shorts −3,314. No speculative buyer left; the marginal flow has to be ETF or official |
| Silver · Copper (COMEX legacy) | +26,049 (25.2%) · +92,476 (31.1%) | copper net +11,607 w/w | But commercial copper shorts rose +16,067 — producers hedging into the spec bid. The classic late-stage configuration in a policy-driven squeeze |
| WTI — NYMEX legacy vs NYMEX managed money vs ICE Europe MM | +136,579 · +111,731 · −9,687 | NYMEX net +6,668 w/w | Three real numbers, not a data error: taxonomy differs between the first two, exchange and contract between the second and third. Speculative length was being added into the escalation |
| Henry Hub natural gas (NYMEX legacy) | +268,533 = 67.0% of OI | gross 14.0× | Flagged, not asserted. Both sides tie to printed OI, but the contract code and the OI level suggest this may not be the headline contract. If verified it displaces gold as the most crowded position in the report |
| Flows — three providers, three universes, never netted | |||
| LSEG Lipper, w/e 9 Sep | US equity −$32.27bn | largest in nine months | Large-cap −$40.44bn is a record weekly outflow. Mid −$682m; multi-cap +$3.52bn; tech +$1.71bn. Bonds +$6.56bn, a 21st straight inflow week. MMF −$10.41bn |
| BofA client flows, 8 Sep | US equity +$7.0bn | 6th-largest week since 2008 | Directly contradicts Lipper and both are right. Lipper measures US-domiciled funds; BofA measures its own client book. Fund investors sold, institutions and hedge funds bought. Concentrated in Technology |
| ICI long-term, w/e 2 Sep · MMF, w/e 9 Sep | +$8.12bn · $7.973trn (−$6.10bn) | domestic equity −$5.14bn; bond +$12.68bn | Mutual fund outflows −$25.11bn against ETF issuance +$33.23bn. ICI's MMF change (−$6.10bn) differs from Lipper's (−$10.41bn) — again, different universes |
| Spot crypto ETFs (Farside) | BTC −$13.2m · ETH +$216.4m | 11 Sep; 14 Sep not published | ETH's strongest September day, and the flow behind the view that closed above |
| Sentiment, surveys and options | |||
| BofA Global FMS, September | not yet published | due mid-month, i.e. today or tomorrow | A calendar fact, not a sourcing failure — and the highest-leverage data point of the week. Cash below 4.0% triggers the house sell rule, on the eve of the Fed. Watch cash, equity overweight, most-crowded trade, biggest tail risk |
| BofA Bull & Bear | 9.5 (10 Sep) | 9.6 on 3 Sep; 9.7 on 7 Aug | Vintage improved from 28 Aug to 10 Sep. ≥8.0 is a contrarian sell; six weeks in sell territory |
| AAII, w/e 9 Sep | Bulls 38.0% · Bears 39.3% | spread −1.3pp, from +2.1 | Prior weeks: −11.5 (26 Aug), −4.4 (19 Aug). The dated table carries no long-run averages and none are quoted |
| CBOE SKEW | 154.5 (11 Sep) | 96th percentile 1yr · 98th all-time | The "skew at the 1st percentile" figure carried since No. 001 is formally withdrawn — see §13. Deep OTM puts are expensive. All-time average 123; record 183 (Feb 2025) |
| VIX · VIX3M · IVTS · put/call | 15.84 · 18.60 · 0.852 · 0.83 | 11 Sep; no Monday close posted | Day 109 of contango (backwardation on 7.6% of days since 2010). Put/call 9-day MA in the 16th percentile — complacency in the wings, extremity in the tail |
| Dealer gamma | −$8 to −$10bn, flip 7,600 (SpotGamma, 12 Sep) | vs BofA +$10.9bn, 94th pct (4 Sep) | Both published: the conflict is convention and vintage — eight days apart, and the index traded through the flip in between. Below 7,600, no gamma trough until ~7,350 and "20-plus VIX quickly". ~76% of SPX volume is 0DTE |
| The 18 September expiry | $6.2trn — 23% of total US options exposure on one day | $9.6trn / 35% cumulative through 18 Sep | Verified against the Citadel Securities primary (Rubner, 31 Aug). Correction to the carried phrasing: 23% is of total exposure, not "of $9.6trn". Surpasses June's $7.7trn record. ~40% of positions roll off, muting gamma toward ~$4bn rather than clearing it |
| Systematic flows (BofA, 4 Sep) | −$126bn down / +$35bn flat | supersedes −$163bn / +$9bn (28 Aug) | CTAs near maximum long; triggers begin at declines of 1.4–4%. Leveraged/inverse ETF rebalance $1.2bn per 1% S&P, $2.7bn per 1% NDX. Asymmetry narrowed to ~3.6:1 but the inventory is full |
| Buybacks & seasonality (Citadel, 31 Aug) | blackout from ~12 Sep vs $1.1trn authorised | 67% of the largest authorisations sit outside Technology | September: 55% of years lower since 1928, avg −1.1%; second half −0.91%, the weakest fortnight of the year; midterm years avg −1.5% with a −6.2% drawdown. Retail dip-buying in September runs at ~half its monthly average |
| Breadth, technicals, valuation | |||
| % of S&P above the 200-day | 59.50% (11 Sep) | long-run mean 65.34% | % above the 50-day could not be sourced for a second consecutive edition — one page served ten-month-stale data. Recommend replacing the source or dropping the line |
| New highs / lows · McClellan · Hindenburg | 67 (1.41%) / 227 (4.79%) · −35.40 · inactive, 3 of 4 | universe ~4,750, not NYSE-only | New lows are 3.4× new highs; the signal fails only on the new highs leg. That is unambiguous breadth deterioration that happens not to meet this indicator's symmetry test |
| S&P moving averages · RSI | 50d 7,669.98 · 100d 7,675.80 · 200d 7,704.32 · RSI 48.96 | from 7,672.84 / 7,677.59 / 7,703.87 / 51.2 | The three averages are compressed into a 34-point band directly overhead — that is the ceiling; 7,600 and 7,580 are the floor |
| FactSet, 11 Sep | Q3 EPS +28.7% · fwd P/E 19.1× | revised up 2.1pp from 26.6% on 30 Jun | Revenue +11.9%; net margin 14.9%, second-highest on record; guidance 72 positive / 42 negative; bottom-up target 9,251.61 (+22%). 5-yr average P/E 19.8×, 10-yr 19.0× — the multiple is below its own recent average. The bear case is not valuation; it is defending 19.1× on +28.7% with a 5% ten-year |
The week ahead
| Day | AEST | ET | Event | Cons. | Prior | Imp. |
|---|---|---|---|---|---|---|
| Today — Tuesday 15 September · FOMC and Copom day one · India reopens · all in-scope markets open | ||||||
| Tue | 11:30 | Mon 21:30 | 🇨🇳 China house price index y/y (Aug) | −3.1% f'cast | −3.2% | M |
| Tue | 12:00 | Mon 22:00 | 🇨🇳 China August activity — industrial production y/y | 4.8% | 4.5% | H |
| Tue | 12:00 | Mon 22:00 | 🇨🇳 Retail sales y/y · fixed-asset investment YTD y/y · surveyed unemployment | 0.8% · −7.2% · 5.2% | 0.6% · −6.7% · 5.2% | H |
| Tue | 14:30 | 00:30 | 🇯🇵 Tertiary industry index m/m (Jul) | +0.4% | −0.2% | L |
| Tue | 16:00 | 02:00 | 🇬🇧 Labour market overview — unemployment rate · avg earnings ex-bonus 3m/y · claimant count | 4.9–5.0% · 3.5% · +8.3k | 4.9% · 3.4% · −11.0k | H |
| Tue | 16:45 / 17:00 | 02:45 / 03:00 | 🇫🇷 Final CPI y/y (Aug) · 🇪🇸 final HICP y/y (Aug) | 2.4% · 4.5% | 2.1% · 3.9% | M |
| Tue | 19:00 | 05:00 | 🇩🇪 ZEW economic sentiment (Sep) · 🇪🇺 ZEW · 🇪🇺 trade balance (Jul) | 37.0–42.7 · 39.5–39.9 · €3.7bn | 34.2 · 31.4 · disputed | H |
| Tue | 22:30 | 08:30 | 🇺🇸 Empire State manufacturing (Sep) — the only US print before the Fed | 14.1 | 20.6 | M |
| Tue | — | — | 🇺🇸 FOMC meeting day one · 🇧🇷 Copom day one · 🇯🇵 SoftBank $25.9bn bridge repaid | 3.50–3.75% · 14.00% | H | |
| Tue | 04:15 Wed | 14:15 | 🇺🇸 CLARITY Act — Senate cloture vote on the motion to proceed (60 needed; 53 R seats, ≥7 Democrats required) | enactment 24–25.5% | H | |
| Tue | 09:50 Wed | 19:50 | 🇯🇵 Trade balance (Aug) · core machinery orders m/m (Jul) | ¥−1,052.6bn · −1.1% | ¥−634.5bn · +9.7% | M |
| Tomorrow — Wednesday 16 September · the meeting | ||||||
| Wed | 10:30 | Tue 20:30 | 🇦🇺 Westpac Leading Index m/m (Aug) | +0.2% f'cast | 0.0% | L |
| Wed | 16:00 | 02:00 | 🇬🇧 CPI y/y (Aug) · core CPI y/y · RPI | 3.1% · 2.7% · 3.5% | 2.9% · 2.6% · 3.2% | H |
| Wed | 19:00 | 05:00 | 🇪🇺 Industrial production m/m (Jul) · labour cost index y/y final (Q2) | −0.5 to −0.7% · 3.0% | 0.0% · 3.2% | H |
| Wed | 22:30 | 08:30 | 🇺🇸 Retail sales m/m (Aug) · core retail sales · import prices | +0.8/+0.9% · +0.5% | −0.6% · −0.3% | H |
| Wed | 00:00 / 00:30 Thu | 10:00 / 10:30 | 🇺🇸 NAHB housing index (Sep) · business inventories · EIA crude inventories | 34 · +0.3% | 35 · 0.0% · −0.4M | M |
| Wed | 04:00 Thu | 14:00 | 🇺🇸 FOMC DECISION + SEP AND DOT PLOT — the median must reach ~4.1% to ratify the strip | 88.1% hike to 3.75–4.00% | 3.50–3.75%; June median 3.8% | H |
| Wed | 04:30 Thu | 14:30 | 🇺🇸 Chair's press conference (time is convention; the Fed publishes the date only) | H | ||
| Wed | 07:30 Thu | 17:30 | 🇧🇷 Copom decision | −25bp to 13.75% (~95%) | 14.00% | H |
| Wed | 08:45 Thu | 18:45 | 🇳🇿 GDP q/q (Q2) — the AUD/NZD event | +0.1% | +0.8% | M |
| Thursday 17 and Friday 18 September — the rest of the cluster | ||||||
| Thu | 21:00 | 07:00 | 🇬🇧 BoE decision + minutes + vote split — no MPR. The vote is the trade, not the level | hold 73.1% / hike 26.9% | 3.75%; 6–3 in July | H |
| Thu | 22:30 | 08:30 | 🇺🇸 Initial claims · Philly Fed (Sep) · housing starts · building permits | 209k · 28.9 · 1.32M · 1.40M | 206k · 47.4 · 1.24M · 1.43M | M |
| Thu | — | — | 🇯🇵 BoJ MPM day one — NO DECISION TODAY (two calendars wrongly print the decision here; see §13) | L | ||
| Fri | ~13:00 | Thu ~23:00 | 🇯🇵 BoJ DECISION — per the Bank's own schedule. Presser ~15:30. Tokyo then shut 21–23 Sep | ~72% hike to 1.25% | 1.00%; 8–1 hawkish dissent | H |
| Fri | 09:30 | Thu 19:30 | 🇯🇵 National core CPI y/y (Aug) · 🇦🇺 RBA Governor Bullock — parliamentary committee (time unconfirmed) | 1.8% | 1.8% | H |
| Fri | 16:00 / 23:15 | 02:00 / 09:15 | 🇩🇪 PPI m/m · 🇬🇧 retail sales m/m · 🇺🇸 industrial production m/m (Aug) | +0.6% · −0.2% · +0.1% | +1.1% · −0.5% · +0.2% | M |
| Fri | all day | all day | 🇺🇸 QUARTERLY OPTIONS EXPIRY — $6.2trn, 23% of total US options exposure on one day | June record $7.7trn | H | |
| The sessions after — decisions and closures through early October | ||||||
| Mon 21 | ~13:15 | Sun 23:15 | 🇨🇳 PBoC Loan Prime Rate (1y & 5y) · 🇯🇵 Tokyo closed (Respect for the Aged Day) | 3.00% / 3.50% | 15th month unchanged | M |
| Tue 22 · Wed 23 | 🇯🇵 Tokyo closed both days · 🇭🇰 Hong Kong closed Tue · 🇦🇺 flash PMIs Tue · 🇮🇩 Bank Indonesia Wed | Mfg 51.7 / Svcs 52.5 | 52.0 / 53.2 · BI 5.75% | M | ||
| Thu 24 | 11:30 · ~17:30 · ~18:00 | Wed 21:30 · 03:30 · 04:00 | 🇦🇺 LABOUR FORCE (Aug) — the last major input before the RBA · 🇨🇭 SNB · 🇸🇪 Riksbank · 🇳🇴 Norges + MPR · 🇲🇽 Banxico · 🇰🇷 Korea closed (Chuseok) | +15.1k · 4.5% | −15.8k · 4.5% | H |
| Fri 25 | 🇰🇷 Korea closed · 🇨🇳 China closed (Mid-Autumn Festival) | L | ||||
| Tue 29 | 14:30 | 00:30 | 🇦🇺 RBA DECISION · household spending indicator 11:30 | 76% hike to 4.60% | 4.35% | H |
| Wed 30 | 11:30 | Tue 21:30 | 🇦🇺 Monthly CPI Indicator (Aug) — the day after the meeting · building approvals · 🇨🇳 Golden Week begins, to 8 Oct | 3.5% hdl · 3.6% trimmed | H | |
| Oct | 🇧🇷 Copom · 🇮🇳 RBI Wed 7 Oct · 🇪🇺 ECB accounts Thu 8 Oct · 🇰🇷 BoK Thu 22 Oct · 🇬🇧 Budget Wed 28 Oct · 🇨🇦 BoC + 🇳🇿 RBNZ Wed 28 Oct · 🇪🇺 ECB Thu 29 Oct · 🇯🇵 BoJ 29–30 Oct + Outlook | headroom £8–11bn vs OBR £22bn | M | |||
Consensus figures are drawn from ForexFactory, Trading Economics and Newsquawk as of 14–15 September and can move. Where calendars disagree the range is printed. Two structural warnings carried from the verification pass: the BoJ decision is Friday 18 September per the Bank's own schedule, and third-party calendars printing Thursday are quoting the meeting start — a sixth consecutive edition of the same error; and every pricing figure in week-ahead material published before 11 September is stale, because pre-CPI Fed odds of 60–66% became a post-CPI cluster of 83–88%.
Risk radar
| # | Risk | Trigger / timing | Probability | Cheapest hedge / expression |
|---|---|---|---|---|
| 1 | The dots have to move 30bp just to ratify the strip — and a median confirming one hike is the unpriced outcome | SEP and dot plot, 04:00 AEST Thu | Hike 88.1%; modal Dec 4.00–4.25% at 48.7%, 26.5% on three. June median 3.8% off a 9–9 split | 2y payers against the hawkish tail; a small dovish-tail expression is the trade nobody wants. Polymarket 79% vs futures 88.1% is the cheap side |
| 2 | The ten-year at 5% breaks the basis trade rather than any macro view | Touched 5.00% Monday; auction concession already at a 25-year high | LF net short 6.86m contracts across the curve; 5y −2.07m the largest single leg | Long volatility rather than any cash-bond expression. Speed, not direction, is the risk |
| 3 | Gamma flip 7,600 is 0.3% below spot, into a record expiry with the buyback bid withdrawn | Spot 7,619–7,630; $6.2trn Friday; blackout from ~12 Sep | No gamma trough to ~7,350; "20-plus VIX quickly" below the flip; 2H-September averages −0.91% | Reduce gross through Friday; own convexity. VIX at 15.84 on Friday's close did not price it, though Monday's ~17.4 partly does |
| 4 | Systematic capacity is 3.6× against you with CTAs at maximum long | Declines of 1.4–4% begin tripping sell levels | −$126bn down vs +$35bn flat (BofA, 4 Sep); $1.2bn per 1% S&P, $2.7bn per 1% NDX from levered ETFs | Long convexity — the second derivative is the trade. CFTC shows zero equity de-risking as of 8 Sep |
| 5 | Salalah collapsed with no new date, and Saudi Arabia was the objector | Talks postponed 13–14 Sep; East–West pipeline shut, repairs >1 month | Crude-carrier transits 0 on four of five days, 7-day avg 0.29/day; all-vessel range 0.9–14/day. No closure probability published | Long Brent–WTI. The diesel crack is now markable at a record $107.72 but is not opened — see §08 |
| 6 | Gold is the most crowded position in the market and sits $27 from its invalidation | Real yields rising into a hike | 56.4% of OI; gross longs 9.0× shorts; crowding increased on the week | Conviction already cut; the $4,300 stop stands. There is no speculative buyer left to absorb an exit |
| 7 | China's credit impulse is gone and the activity data lands at noon today | New loans ¥60bn vs ¥400bn; loan growth 4.9% | FAI cons. −7.2% YTD y/y, marked down from −6.7/−7.0 | UW ASX materials; short iron ore. Monday's Materials at −0.55% has not taken the read-through |
| 8 | The AI-capex de-rating moves from sentiment to guidance | Amodei/Altman "pace the frontier"; semis −5–8%, Korea −2.43% | Semiconductor levered-ETF AUM already −50% from the June peak; technology −38% | The leveraged-retail channel has largely unwound — less forced selling left, but no leveraged bid to catch it either |
| 9 | SoftBank's $10–20bn HY book is the AI-credit mark, and it prices after the Fed | Roadshow 14–17 Sep; bridge repaid 15th | BB+ rated. CCC 1,070bp ≈ 4× the HY index — but all credit marks are four sessions stale | UW CCC specifically, not the HY index. The desk's highest-conviction view is currently unmarkable |
| 10 | CLARITY cloture is tonight and the odds have RECOVERED, not fallen | 04:15 AEST today; 60 votes, ≥7 Democrats needed | 24–25.5% enactment (Polymarket/Kalshi, 13 Sep) — a correction: No. 006 carried 16–19.5% and falling | Only two senators publicly identified (Gallego for, Gillibrand against). No leverage into the vote–FOMC pair |
| 11 | The BoE vote splits hawkish into a 3.1% CPI print | CPI Wed, MPC Thu 21:00 AEST | Hold 73.1% (stamped 11 Sep, not refreshed); Dec 100% higher | Short the gilt front end into CPI. The level is priced; the vote is not |
| 12 | The BoJ hikes into a three-day Tokyo closure | Friday decision; Tokyo shut Mon–Wed 21–23 Sep | ~72% (OIS), with a 97% outlier and a discarded 62% | Size for a gap, not a drift. Any repricing clears offshore into a four-day domestic gap |
| 13 | Governor Cook's status is unresolved going into a live meeting | SCOTUS 5–4 for Cook 29 Jun; removal notice served early Aug with a three-week window that has expired | — outcome unsourced; we cannot confirm she is seated and voting | Long gold on dips (but see #6); steepeners. This is the largest single unknown on the US desk |
| 14 | The Riksbank meets into a one-seat parliament | Swedish count 175–174 at 87.2% reporting; final count from Wed 16, certification 19–20 Sep; Riksbank 24 Sep | — | SEK and Swedish rates carry a political premium that does not clear before the meeting |
| 15 | The bull case, as a risk to the bears | FactSet, 11 Sep | Q3 EPS +28.7%, revised UP 2.1pp; guidance 72 positive / 42 negative; fwd P/E 19.1× vs 19.8 five-year | Earnings revisions are going the right way intra-quarter. Do not treat the bear case as settled |
Key levels
| Instrument | Last | Support | Resistance | Comment |
|---|---|---|---|---|
| S&P 500 | 7,619–7,630 | 7,600 (gamma flip) · 7,580 · 7,350 | 7,669.98 (50d) · 7,675.80 (100d) · 7,704.32 (200d) | The three averages sit in a 34-point band overhead; RSI 48.96 from 51.2; 59.5% above the 200d vs a 65.3% mean |
| UST 2y | 4.65% | 4.50 · 4.43 | 4.75 · 4.85 | The instrument the dots move, and the one this desk is deliberately flat |
| UST 10y | 4.98% | 4.85 · 4.79 | 5.00 (touched Monday) · 5.10 | Cleared the round number intraday for the first time since 2023 and closed back below |
| UST 30y · US 5s30s | 5.35% · +57bp (both Fri) | 5.25 · 45 (view closes) | 5.40 · 70 (4 Sep) | Unmarked Monday. Twelve basis points from the stop |
| DXY | 99.43 | 98.5 · 98.0 | 99.5 · 100.0 | Rose on the day the 10y touched 5% — a more normal correlation than last week's |
| EUR/USD | 1.15418 | 1.1500 · 1.1450 | 1.1563 (the broken floor) · 1.1650 | Closed 21 pips through the range that defined the view. Closed, not reversed |
| USD/JPY | 154.86 | 152.00 · 150 · 148 | 155 · 155.98 (entry) · 156 | Gave back most of Friday's move into a ~72%-priced hike. Squeeze fuel gone |
| AUD/USD | 0.71408 | 0.7125 · 0.7100 (invalidation) | 0.7197 (entry) · 0.7226 | 41 pips from the stop. Fourth session below entry |
| AUD/NZD | 1.2346 / 1.2353 | 1.2315 (entry) · 1.2278 | 1.2400 · 1.2500 | NZ Q2 GDP Thursday is the event |
| Brent (Nov) | $107.13 | 104.61 · 100 · 92–95 (re-own) | 108.47 (Mon high) · 110 · 122 (RBC) | Four-month high intraday. The $105 add-condition is exceeded and retired rather than triggered |
| Brent–WTI | $4.69 | 4.00 · 3.00 (view fails) | 5.16 · 6.00 · 8.00 | Input range $4.30–5.04 intraday — the 13-cent gain is inside the noise |
| Gold (spot, derived) | ≈$4,327 | 4,307 · 4,300 (invalidation) | 4,385 · 4,500 · 4,541 (200d) | $27 from the stop. Dec settle $4,352.76; basis ≈$25–32, normal carry |
| Copper LME 3M · Comex–LME arb | $14,065/t · −$66 to −$198/t | 14,000 · 13,800 | 14,875 (record) · +300 (view trigger) | The arb has flipped sign — Comex at a discount. Watch the arb, not the price |
| Iron ore | $97.55/t | 95 · 90 | 100 · 105 | Fifth session below the handle; China activity at noon |
| Bitcoin | $79,073 | 78,000 · 76,000 · 74,000 | 79,800–79,890 (twice rejected) · 82,500 · 83,000 | Sitting ~1% under the rejection zone with two binaries inside 24 hours |
| Ether | $2,540.45 | 2,500 (the broken cap) · 2,441 | 2,600 · 2,700 | Weekday, above $2,500, normal volume — the trigger, and the view is closed |
| ASX 200 | 8,749.9 | 8,699.80 · 8,600 | 8,819.40 · 8,900 · 9,000 | A-VIX 14.05, barely off a three-month high; breadth still negative; no usable SPI |
| ACGB 3y / 10y · 3s10s | 4.98% / 5.34% · 36bp | 4.89 / 5.28 · 30bp | 5.00 (failed twice) / 5.45 · 43bp (entry) | The 5% handle did not hold in 2s or 3s |
| OAT–Bund · BTP–OAT | 97.0bp derived / 89.8bp dedicated · −9.0bp | 80 (view closes) | 100 · 110 | 7.2bp between the two series, all in the French leg |
| US IG / HY / CCC OAS | 80 / 270 / 1,070bp | CCC 900 (view compresses) | CCC 1,100 · 1,200 | All observation 10 September — four sessions stale and spanning the two that matter |
Data notes & sources
Corrections to previous editions — four, three material
(1) "US retail diesel $6.00/gal, the first time ever above $6" is refuted. The EIA weekly ULSD retail series prints $5.967/gal for the week ending 7 September — three cents short of the threshold. The direction was right and dramatic (+36.8c week on week, the largest move in the series shown, from $5.599 on 31 August) but the threshold claim is withdrawn pending identification of which series produced it. (2) The CLARITY enactment odds were carried in the wrong direction. No. 006 published "16–19.5%, down from 24%"; the current quotes are Polymarket 25.5% and Kalshi ~24% as of 13 September 09:05 ET — at or above the level they were said to have fallen from, and recovering into the vote, possibly on Senator Lummis's updated 10 September text. The two readings may be different instruments; either way the risk-radar line is restated upward. (3) The record-expiry phrasing was wrong in its denominator. Verified against the Citadel Securities primary: $6.2trn is 23% of total US options exposure expiring on 18 September alone; the $9.6trn / 35% figure is the cumulative amount expiring through that date. The carried formulation "23% of $9.6trn outstanding" mis-stated it. (4) The BoJ's July dissent was hawkish, not dovish. The 8–1 vote on 31 July had Takata dissenting for an immediate move to 1.25%. Asada remains the board's named dove but was not that dissenter.
Conflicts resolved, and one that was not
The principal unresolved conflict is Monday's S&P 500 close. Two sources tie their change to the same verified Friday base of 7,656.98 and disagree: a narrative recap gives 7,619 (−0.50%, "down 37 points"), a quote page gives 7,629.64 (−27.34, −0.36%). The official par curve, the AP tabulation and every historical table had not posted at fetch time; the two live blogs were still intraday. Rather than pick one, the range is printed throughout and the dashboard says so. The same applies to the Dow (52,421 vs 52,454.91). The Nasdaq Composite at 26,186 is single-sourced. No VIX close exists for 14 September from either the historical series or the term-structure source; the intraday mark of ~17.4 at 09:38 ET is labelled as such. No Stoxx 600 or SMI close could be obtained and neither is printed — only intraday marks of 638.95 and 637.50 exist for the former.
Resolved: three vendors carried a phantom "Sunday 13 September" row in their Brent, WTI and gold tables, which corrupts every printed daily percentage on the Monday row — all changes in this edition are recomputed from verified Friday settles, and Brent's $107.13 was read from the dated table with the 10 September settle re-checked and confirmed at $107.63 (no further restatement). Gold's basis is published as a derived ≈$25–32 rather than the $60–66 a face-value read gives: the vendor's spot series ran ~$35 below the corroborated spot on 11 September, and applying its own daily percentage to the verified Friday spot gives ≈$4,327, which a CAD cross-check corroborates to within a dollar. The Comex–LME copper arb is computed, not quoted — no source states it — and the published range of −$66 to −$198/t reflects three Comex inputs, none of which is an exchange settle; note also that the Friday baseline is itself disputed, with one source implying +$207/t against the ≈$0–30/t this desk published. USD/CNY spot sits 630 pips stronger than the 6.7698 fix, which is implausible for onshore CNY; the fix is double-sourced and used, the spot is flagged. Two vendor pages described the PBoC fix as "stronger than expected" when 6.7698 against a 6.7083 estimate is unambiguously weaker; numbers taken, prose rejected. Hong Kong carried 24,918 on one page and 25,005 on another at the same stated +0.45%; the internally consistent figure is used. European index percentages again contradicted themselves — the Euro Stoxx 50 carried three values on one page and the AEX's change field implies a prior that does not tie; every level printed here was reconciled against Friday before publication and those that failed are omitted rather than printed. The Sensex level is disputed 74,782 against 74,872 within a single article; India was closed Monday so only the percentage is used. Wheat and soybean levels do not reconcile with Friday's carried marks by 3.3% and 1.3%; direction is unaffected. BoJ hike pricing narrows to ~72% (OIS, 14 Sep) with a 97% outlier; a 62% reading was discarded rather than averaged because the page publishing it also shows a 0.50% policy rate and the wrong meeting date.
Cleared from No. 006 — five, and two of them had run for four editions or more
The diesel crack is marked. RBN published on 14 September with a record close of $107.72/bbl on 10 September, corroborated independently at $106.44 for the 9th from an EIA-derived series — against the $101.1 of 4 September this desk has carried, unable to score, for three editions. Staleness falls from ten days to four. Crypto derivatives are obtained after four consecutive failures: BTC funding +0.0055%, OI $26.6bn; ETH +0.0056%, $17.5bn; aggregate OI $64.7bn with $143.6m of 24-hour liquidations. The Bullock event is pinned to Friday 18 September by weekday-correcting a single source whose two-day date error was independently demonstrated on the same article's Hunter item. Norges Bank's 24 September date is confirmed from the bank's own site and the Riksbank's from two secondary sources, closing what was the least-verified line in the table. Oracle's total debt is resolved at ~$125.34bn for the quarter ended June 2026, settling the contested $122–129bn range; the $160bn headline and S&P's $95bn are measuring different things.
Still unverified
The 11 September credit observation does not exist, for a fourth consecutive edition — and this time it is confirmed as a genuine data absence rather than a fetch failure. FRED, ALFRED and the independent corroborant all terminate at 10 September; FRED's own stated 14 September release had not produced observations by mid-afternoon Chicago time. IG 80 / HY 270 / CCC 1,070bp are therefore four sessions stale and span both the post-CPI session and the pre-FOMC session. This is the desk's highest-conviction view and it is unmarked. Oracle's CDS remains the single most important missing number: the freshest dated mark is ~203bp from 24 July, now roughly seven and a half weeks old, with an 11 August item saying only "again above 200bp"; the most promising unexploited lead is robots-blocked. Governor Cook's status is unresolved — the three-week response window served in early August has expired with no sourced outcome, and we cannot confirm she is seated and voting at this meeting. Also open: Monday's official Treasury par curve (not posted; 5y and 30y have no Monday mark, so 5s30s cannot be rolled); Monday's US sector performance, NYSE advance/decline and composite volume; the ASX 200's Monday range and volume, and individual closes for CBA, NAB, ANZ, RIO and FMG (only Westpac +0.9% and BHP −0.5% are sourced, which matters because the sector-level financials number is an auction artefact); any usable SPI indication after the December roll; Hang Seng Tech's Monday close; the % of the S&P above the 50-day for a second edition, one source having served ten-month-stale data; SoftBank's tranches, tenors, spread talk, order book and leads, none disclosed; 14 September ETF flows, not published; the reported Morgan Stanley ~90% no-hike call, whose headline exists in the search index but whose article body will not render — for a second edition, and material if true; a current strategist target table, the tracker being both four months stale and now login-gated; the SNB's September date from the SNB itself; the Trump–Xi summit date, still only "September, White House"; the 17 June memorandum's status, whose 60-day window expired 16 August with the public record predating the expiry; Sweden's official count, at 175–174 on 87.2% reporting; uranium's Monday print; and Deutsche Bank's vol-control percentile, which could not be re-sourced and should be dropped or re-dated.
Traps caught
Five, and one is a new species. A page correctly dated "Gold Rate Today USA, September 14, 2026" quoted spot gold at $1,975/oz — wrong by roughly $2,350, about 55% — and ranked highly; discarded in full. This is more dangerous than the year-traps caught in prior editions because the date is right and only the levels are fabricated. A Reuters European close dated 7 September ranks highly on the 14 September query and carries a week-stale Stoxx 600 of 649.9. A 2025 SNB poll and a 2025 triple-witching piece citing $6.5trn both surfaced and were rejected on year. A widely-syndicated market-update page and a major personal-finance site both carried Friday's closes under a Monday headline ("Grim Monday: AI Fears Spook Tech" over S&P 7,656.98 +0.86%). And the geographic trap recorded last edition recurred in a live form: the Saudi pipeline attack's launch platform was seized in Maysan province, Iraq, roughly 2,000km from Mayun (Perim) Island, Yemen — both theatres are active this week and the names must not be merged.
Tomorrow's first verification targets
Monday's S&P close, from the AP tabulation or the official par curve, to retire the range published here. The 11 and 14 September credit observations, whenever FRED resumes. The BofA Global Fund Manager Survey, due today or tomorrow and the highest-leverage single data point of the week. China's August activity at 12:00 AEST. The FOMC dot plot itself — specifically whether the end-2026 median reaches 4.1%, and whether any dissent appears on the dovish side. The CLARITY cloture result. Whether crude-carrier transits through Hormuz move off zero. Oracle's CDS. And whether Australian financials can produce a gain inside the regular session rather than in the closing match.
Distribution
Published to the standing artifact URL as this edition's version. Stored in Supabase (macro_editions, edition_no 7) with a character-count and structural integrity check against the published file. PDF delivered. The public site and newsletter remain non-operational — no Vercel project exists and no broadcast has been sent — so nothing here is publicly readable and no subscriber receives it.
United States, the Fed & positioning
- US Treasury — daily par yield curve, September 2026
- Investing.com — Fed Rate Monitor · centralbank.watch — Federal Reserve · Polymarket — September Fed decision
- Federal Reserve — 2026 speeches (blackout verified empirically) · FOMC calendar
- June 2026 SEP and dot distribution · Kiplinger — September Fed live blog
- Cleveland Fed — inflation nowcasting (core PCE) · BLS — September release schedule
- Stocks and News — Monday 14 September close · Trading Economics — US indices · TheStreet — 14 September live blog
- Investing.com — S&P 500 technicals · VIX historical data
- Supreme Court — Trump v. Cook, 29 June 2026 · CRS — LSB11449
- CRFB — congressional fiscal deadlines
- CFTC — Traders in Financial Futures · legacy CME · COMEX · NYMEX · petroleum disaggregated · ICE Futures U.S.
- Citadel Securities — September Setup (Rubner, 31 Aug) · SpotGamma — September opex, 12 Sep
- BofA research index (Flow Show, Bull & Bear, systematic flows monitor) · LSEG Lipper weekly flows · ICI — long-term fund flows · ICI — money market assets
- AAII — sentiment survey results · CBOE SKEW · VIX term structure · put/call · Hindenburg / McClellan · Fear & Greed replica · S&P breadth
- FactSet — Earnings Insight, 11 September 2026
Rates, FX & central banks
- Trading Economics currency pages — US · euro area · UK · Australia · New Zealand · Canada · China · Mexico · India
- Investing.com — USD/JPY · AUD/NZD (both crosses cross-computed from legs first)
- Trading Economics bond pages — Germany · France · Italy · UK · Japan · Australia · Canada
- Ideal Investisseur — OAT–Bund spread series
- Bank of Japan — MPM schedule (the authoritative decision date) · BoE — upcoming MPC dates · BoE pricing
- centralbank.watch — RBA pricing (primary) · Canstar — big-four forecasts · Michael West — RBA speakers · Oz Arab Media — bank calls
- Econostream — ECB speakers and tone meter · BBVA — ECB Watch, 10 September
- Norges Bank — policy rate and decision dates · Riksbank — August decision · central bank meeting calendar
- Rio Times — Copom pricing · Tradingpedia — PBoC fix
Australia, New Zealand & Asia
- Market Index — evening wrap, 14 September (close-basis sectors; the closing-auction note)
- Investing.com newswire — ASX close, breadth and A-VIX · ASX 200 historical data · ASX 200 futures (December roll)
- ABC — markets live, 14 September
- ABS — future release calendar (Labour Force 24 September confirmed) · Trading Economics — Australian calendar
- Domain — auction results · MacroBusiness — auction volumes · Cotality home value index tracker
- Nikkei Indexes — official archive · TOPIX historical data · News On Japan — 14 September close · JPX — holiday calendar
- Trading Economics — China calendar (August credit actuals; activity consensus) · China indices · Hong Kong · Korea · Taiwan · India
- Business Standard — Indian market holiday
Europe & geopolitics
- Trading Economics — European indices · Eurasia Business News — 14 September · Reuters via Investing.com — European open
- Newsquawk — headlines and data actuals · ForexFactory — calendar, week of 14 September
- Bundestag — 2027 budget introduction · House of Lords Library — UK fiscal outlook · Swedish general election — preliminary count
- Al Jazeera — Salalah talks deferred, 14 September · Free Press Journal — postponement
- GlobalSecurity — Iran War OPREP · Tankermap — Hormuz transits (crude-carrier series) · Windward — vessel counts · Al-Monitor — Kpler and Bab el-Mandeb
- Commons Library — CBP-10637, the 17 June memorandum · CSIS — Trump–Xi 2026 summits
- Euronews — overnight drone attacks, 13 September · RFE/RL — Poland border incidents
Commodities, credit & digital assets
- Investing.com — Brent dated settle table · WTI · gold (Comex December) · Trading Economics — commodities board
- RBN Energy — diesel crack, 14 September · crack spread series · EIA — weekly ULSD retail price
- EnergyConnects — OPEC+ holds October output · IEA news · SPR inventory · gCaptain — tanker and VLCC rates
- Reuters via Business Recorder — LME close, 14 September · Westmetall — LME copper official settlements · Gold Stock Canada — CAD cross-check
- FRED — IG OAS · HY OAS · CCC OAS · ALFRED — vintage check confirming the absence · Convex Trade — HY corroboration
- SoftBank — bridge repayment and HY roadshow · Oracle — total debt
- CoinGecko · Coinalyze — funding, open interest, liquidations · Farside — BTC ETF flows · ETH ETF flows · DeFi Rate — CLARITY Act fact sheet