The bottom line
- Friday's rally was bought on an Iran deal that was publicly dead twenty-eight hours after the settle, and Sydney opens with the gap risk. Crude fell hard into the Friday close — Brent −2.14% to $104.32, WTI −2.33% to $92.41 — on Araghchi's seven-day UNGA roadmap: fighting stops, the US lifts its naval blockade and unfreezes assets, and the Strait of Hormuz reopens at the end of day seven. On Saturday 26 September Trump rejected it outright — "they made a proposal but I rejected it" — and the same day the Houthis fired two drones at Riyadh and a ballistic missile at Khamis Mushait, the first air-raid alerts over the Saudi capital since the March–April hostilities. On Sunday he told Axios he expects more talks this week, via Qatari mediators, but "it is not the deal that I want to make", and on resuming strikes, "I am always thinking about it." Friday's settles are therefore stale as a risk marker — and the positioning correction below makes the asymmetry worse, not better.
- A carried positioning claim was wrong, and correcting it reverses the trade implication on the single most headline-sensitive asset in the book. This desk has been publishing that crude is net short on both COT cuts, and therefore that escalation would squeeze. It is not. NYMEX WTI legacy non-commercials are net LONG 141,106 contracts and got longer on the week (+5,201) on the 22 September report. The only crude contract that is net short is the far smaller ICE Futures Europe listing, at a trivial −4,834 on a third of the open interest — the carried claim took the first crude block on the disaggregated page and attributed it to NYMEX. The consequence: an escalation has far less short-squeeze fuel than this note has been assuming, and a genuine de-escalation now carries real long-liquidation risk. Own optionality, not futures, remains right — but for the opposite reason to the one previously given.
- The RBA decides at 14:30 tomorrow, the market trimmed its conviction into the weekend, and for the first time in several editions the tracker and the physical curve agree. Hike pricing has come back from 96.5% to 88–90%, and this time it is corroborated rather than contradicted by the bonds: the ACGB 2-year rallied 7bp to 5.01% and the 3-year 4bp to 5.00%, taking the embedded premium over a 4.35% cash rate from ~73bp to ~66bp. 3s10s steepened roughly 4bp into the meeting — a market marking this as near-terminal rather than the start of a run. Three further strands support that: Bullock's own test is already satisfied (she told CEDA on 22 September that unemployment "between 4.5 and 5" would "probably take enough heat out of the labour market"; August printed 4.6%); the ABS itself caveated that print for Labour Force Modernisation effects and an unusually high flow straight from outside the labour force into unemployment; and — new to this note — a voting Monetary Policy Board member publicly dismantled the wage-price-spiral premise five sessions before the decision. Iain Ross, 22 September: "there is no evidence of the emergence of a wage-price spiral in the present circumstances and recent data suggest such an outcome is unlikely." The hike is not the risk; the guidance is. The market prices a further ~40bp beyond tomorrow and CBA says 4.60% is the peak.
- Friday's US tape was a twist steepener dressed as a rally, and the long bond made a new cycle high on a day everything else went up. Treasury's official par rows for both the 24th and the 25th have posted. The front end bull-steepened hard on the oil relief — 2-year −6bp to 4.81% — while the 30-year rose 2bp to 5.49% and the 20-year to 5.54%, both new cycle highs. 5s30s went 41bp (23 Sep) to 44bp to 51bp, +7bp on Friday alone and +3bp on the week. The uncomfortable part for this desk is in §08: V003's steepener was stopped out at 41bp on the 23rd and the spread has since retraced ten basis points back through its own 45bp stop. And the macro point is larger than one view: Treasury has been running doubled long-end buybacks — from $2bn to at least $4bn per operation in the 10–20y and 20–30y buckets — since 9 September, and the long end has gone straight up anyway. Either the sizing escalates materially at the 4 November refunding or the market concludes the tool does not work. The 30-year reopening on 8 October is the test.
- Credit widened beta-first into the largest high-yield deal ever printed, which vindicates the house view that was downgraded on Friday. All three series posted together at 24 September: IG +2bp to 79, HY +12bp to 280, CCC +37bp to 1,112. CCC widened eighteen times as much as IG. That is not a rates move passing through — it is credit risk repricing at the low-quality end, and it is the cleanest risk-off signal anywhere in this note. It also lands directly on V017: SoftBank's $11.1bn BB+ priced through talk on 23–24 September, and over exactly those two sessions the index cheapened 12bp at the HY level and 37bp at the high-beta end. Pricing through talk bought the issuer a good outcome; the secondary market paid for it. The view is now 38bp from the 1,150bp level that confirms it rather than 25bp from the level that closes it. Friday's conviction cut from Med to Low was premature and is reversed.
- Two structural reads changed direction, and both matter more than the week's price action. First, Japan's "dovish hike" thesis is dead. A BoJ official signalled the Bank expects to raise roughly once every three months, potentially to 2% by around June 2027. Friday's price action is that signal exactly: JGB 2-year +3.1bp to 1.94% while the 30-year fell 0.8bp — a front-end-led bear flattener, not a term-premium event — with the yen up 0.99% and the megabanks up ~4% on a +1.30% index. The carried interpretation, built on Thursday's yields-up-yen-down signature, did not survive one session. Second, the demand leg of the regime tag is now in hard data. Germany's October consumer climate fell to −30.6 against −27.4 expected, a near-two-year low driven specifically by collapsing income expectations; US Michigan sentiment finished September at 48.1 with one-year inflation expectations at 4.6%, up from 3.4% before the Iran conflict; and only 49.1% of S&P 500 constituents sit above their 200-day moving average — through 50% for the first time in the carried series — with 308 new lows against 46 new highs on a day the index closed 0.93% from its record.
Overnight & weekend recap
Asia and Australia — Friday 25 September
Tokyo was the region's standout and the exchange's own archive has now settled the numbers. The Nikkei 225 rose 1.30% to 66,364.20, opening at its low of 65,639.62 and rallying all session to close 46 points off the high — a clean trend day, and +2.07% on the week, the best of any major index. The TOPIX rose 1.31% to 4,128.59, which matters: a broad index matching the Nikkei point-for-point means a market-wide re-rating rather than a few heavyweights. Japanese megabanks were the tell — Mitsubishi UFJ +4%, Mizuho +4.2%, Sumitomo Mitsui +3.6% — and banks up four per cent on a 1.3% index day is a rate-path trade, not beta.
The rate path is what changed. A BoJ official indicated the Bank expects to raise rates roughly once every three months, potentially lifting them to 2% by around June next year. The JGB curve priced it immediately and in the right shape: 2-year +3.1bp to 1.94%, 5-year +2.2bp, while the 10-year eased 1.2bp to 3.07% and the 20s, 30s and 40s all edged lower. A front-end-led bear flattener with the long end bid is what "more hikes sooner, inflation credibility improving" looks like, and it is the opposite of Thursday's parallel rout. The 10-year did trade as high as 3.11% intraday, a fresh thirty-year high, before retreating. At 1.94% the 2-year embeds roughly 69bp over the 1.25% policy rate — about 2.75 further hikes, materially more than the ~30% for October that has been carried.
Australia closed at a three-month low on a Melbourne public holiday. The ASX 200 fell 0.43% to 8,665.00, verified across four sources, opening at Thursday's close of 8,702.00 — which was also the day's high — and grinding to 8,639.90 before a small recovery. Volume was 589.9m units, roughly 19% below Thursday, with the wrap disclosing "light corporate newsflow" because of the AFL Grand Final holiday; treat Friday's internals as thin. Only two of eleven sectors rose — Consumer Staples +0.73% and Financials +0.27%, the latter on a third of the index weight and the only reason the fall was not worse — against Information Technology −1.66%, Consumer Discretionary −1.37% and Utilities −1.25%. Breadth was 336 advancers to 719 decliners. The A-VIX fell 1.74% to 11.53, which ties exactly to Thursday's 11.73 and is notably complacent into a central-bank decision. The wrap's own commentary is worth carrying: Tech, Real Estate and Consumer Discretionary have all entered correction territory, the average ASX 200 constituent sits 22.9% below its 52-week high, and the median tech name is 43% off. Movers: Netwealth −8.42% to $16.97 and Hub24 −4.35% to $64.49, both on a First Guardian class action against Netwealth subsidiaries — one theme, two platform names; lithium names fell as Chinese lithium carbonate futures dropped 4.8%.
The Australian front end rallied into the RBA — 2-year −7bp to 5.01%, 3-year −4bp to 5.00% — while the 10-year held at 5.39%, steepening 3s10s about 4bp. Hong Kong fell 1.01% to 24,510.09 on surging global yields, with technology and financials leading declines; the arithmetic ties to within 0.13 of a point against Thursday's close, which retrospectively validates a figure that would not tie last week. The mainland was shut (Mid-Autumn, 25–27 September) and reopens today. India rebounded but the streak held: Sensex +0.43% to 73,895.74 and Nifty +0.34% to 23,140.50, both tying exactly — yet the Nifty logged a seventh consecutive weekly loss, down 5.8% over the seven weeks, its longest run since the Covid crash of early 2020. Korea and Taiwan were both shut; Indonesia's IDX Composite was the region's worst on the week at −3.09%.
Europe — a three-week losing streak snapped, and the FTSE correction
Europe had its best week since early August. The Stoxx 600 rose 0.35% to 638.65 for +0.50% on the week, ending three consecutive weeks of roughly −3%; the Euro Stoxx 50 +0.48% to 6,302.82 and +1.07% on the week. Both figures close a gap that could not be published at all in the last edition, and both chains tie back to the verified 23 September anchor — the fabricated front row that produced the trap last week is gone. Sector composition was a textbook lower-oil rotation: banks +1.3%, financial services +1.1%, travel and leisure +0.8%, energy −1.3%. Switzerland led the week (SMI +1.15%) and France lagged (CAC 40 +0.16%) — the cyclical-bank complex over the luxury complex, all week.
A correction that changes a published characterisation: the FTSE 100 did not gain on Thursday and Europe had no gainer at all. This note reported the FTSE up 0.21% to 10,728.00 as "Europe's only gainer". The dated row and an independent point-change both put Thursday's close at 10,679.99, down 0.24% — a 48-point error with the sign inverted. Every major European index fell on Thursday 24 September. The CAC and FTSE MIB Thursday closes are corrected in §13 as well. On Friday the FTSE rose 0.14% to 10,695.25, a figure that reconciles to the penny from Thursday's corrected close plus an independently published point change.
The week's most important European print was the German consumer, and it was bad in the way that matters. October consumer climate fell to −30.6 against −27.4 consensus and a −26.8 prior — a 3.8-point fall, a 3.2-point miss and a near-two-year low — driven by a significant drop in income expectations and rising savings intentions, with economic expectations virtually unchanged. That is an energy-cost real-income shock rather than a growth scare: the German household is not forecasting recession, it is forecasting a smaller pay packet. It sits directly against Wednesday's PMI beat and it is the more reliable of the two for the consumer. Bailey turned hawkish on Friday at Oxford: "It's going to get harder to maintain that stance the longer we have high energy prices", adding that the Bank "could not afford to wait for the full evidence" on energy pass-through into expectations while calling the existing evidence "quite subdued". He was in the 6–3 majority to hold on 17 September; this is the most hawkish Bailey of the cycle and it validates rather than fights the ~81% priced for November.
The US session — first winning week in three, and a long bond that did not participate
AP's tabulation posted and it reconciles arithmetically to two independent anchors, so for the first time in several editions the US cash closes are verified rather than derived. The S&P 500 rose 0.51% to 7,743.41, the Dow 0.93% to 51,828.62, the Nasdaq Composite 0.48% to 27,068.72 and the Russell 2000 0.07% to 2,837.55 — the first winning week in the last three, with the S&P 0.7% from its record. The driver was exogenous and singular: oil coming off on the Iran roadmap. AP's own framing — "a cooldown in oil prices helped release some of the pressure that's built up on Wall Street" — is the whole explanation, and it is why the composition looks nothing like a growth rally.
Underneath, the bond market split. The front end bull-steepened on the oil relief and the Fed-pricing fade: 2-year −6bp to 4.81%, 3-year −5bp, 5-year −5bp, 7-year −4bp, 10-year −1bp to 5.17%. The 20-year rose 1bp to 5.54% and the 30-year 2bp to 5.49% — both new cycle highs. On the week the whole curve is higher (2y +5bp, 10y +16bp, 30y +15bp) and 5s30s steepened from 48bp to 51bp. The 10-year's 5.18% on the 24th was the highest since July 2007 and is independently corroborated; it was the long bond, not the ten-year, that set a new high on Friday. Term premium at the very long end is still widening, and it is doing so while Treasury buys it back.
Friday's data leaned hawkish on capex and grim on the consumer. August advance durable goods orders were flat at −0.0% with ex-transportation +0.3% and core capital goods +1.6% — that last figure directly validates the FOMC's "capital investment is robust" language. Against it, final September Michigan sentiment printed 48.1, its lowest in four months and 15% below January, with year-ahead inflation expectations at 4.6%, the highest since June and up from 3.4% "before the Iran conflict." Thursday's claims were 197k with continuing claims at 1.719m — labour-market stasis. Cleveland's Beth Hammack, a 2026 voter, spent Friday warning that "the biggest risk with inflation is the formation of an inflationary mindset" and explicitly declined to treat 5%-plus yields as a policy problem, saying current yields reflect growth, US debt and the rate path.
Fed pricing softened on both venues and this edition has a genuinely settled read. The rate monitor, stamped "Sep 26, 2026 12:35AM EDT" — after Friday's 17:00 ET futures close, so the settled state rather than a snapshot — puts the 28 October hike at 66.6% against 33.4% for a hold, down from the 71.2% pre-close read carried on Friday. December's modal outcome is still two cumulative hikes, but only just, at 51.6%. Polymarket's event page agrees within 1.6 points at 65%. Coherent with a 2-year 6bp lower: Friday's oil relief took some hawkishness out of the front end without changing the modal path.
The weekend — 26–27 September
Iran is the weekend, and it went the wrong way for anyone who bought Friday's tape. The sequence is dated and verified. On Friday and Saturday Araghchi tabled a seven-day roadmap at the UNGA: fighting stops on all fronts, the US lifts its naval blockade of Iranian ports and oil sanctions, frozen assets are released, a regional ceasefire covers Lebanon and Yemen, and the Strait of Hormuz reopens at the end of day seven, with nuclear talks following. Araghchi said the US steps were completable in four to five days. On Saturday 26 September Trump rejected it — the offer "would not be acceptable", suggesting Iran was motivated by losing "so badly", and telling Axios it was what the US "would have maybe agreed to a year ago". The same day the Houthis fired two drones at Riyadh and a ballistic missile at Khamis Mushait, with alerts also for Abha and Jazan; all were intercepted, and they were the first air-raid alerts over Riyadh since the March–April hostilities. On Sunday 27 September at 14:55 ET Trump told Axios he expects more talks this week — indirect, via Qatari mediators, Witkoff and Kushner with Araghchi — but that "it is not the deal that I want to make", and on resuming strikes, "I am always thinking about it." Araghchi spent the weekend claiming Tehran had not received an official US position and that the US ambassador "has not read" the proposal. No ceasefire and no memorandum is in force; the 17 June memorandum's collapse over shipping in July is now corroborated rather than single-source.
The physical picture did not improve, and one claim about it does not survive arithmetic. The Saudi East–West pipeline restarted on 22 September, eleven days after the strikes launched from Maysan province, Iraq — which vindicates the US Energy Secretary's "within days" on the restart and the three-to-five-week camp on full throughput, with initial flows well below the ~4m bpd pre-outage level and a return to pre-strike volumes still weeks away. But the Red Sea terminus at Yanbu is now the exposed node, which means the workaround has acquired the vulnerability it existed to avoid. Tracked transits stayed in the low single digits per day: Tankermap's seven-day tanker average was 1.3 vessels/day with LNG at zero, Windward counted 13 vessels of all types on 26 September, and VLCC rates are at a record $1.27m/day. Against that, Trump told Axios that "over 22 million barrels" transited on Friday night, the largest volume since the war began, with a US defence official confirming it. That figure requires roughly eleven VLCC-equivalents in one night and is incompatible with single-digit daily transits of all vessel types — and a freight market clearing at a record price is pricing a closed strait, not a record one. This is the third consecutive US-government volume claim to fail against vessel tracking, after "above 10 million bpd" and "40 ships a day under American protection". It is published as a claim, not a figure.
Elsewhere the weekend was quiet, with three items worth carrying. Korea reopens today into a two-session backlog, and the Sunday domestic commentary named its own downside triggers explicitly — oil above $100 and US Treasury yields above 5%, the pair that previously drove the KOSPI to the 6,600 region. Both are live. Foreign investors bought a net ₩400bn on 18 September, the first net buying in eight sessions, and the bull case is 7,500–7,700 on semis plus summit expectations; the index's last close was 7,080.92. Expect a gap-and-fade rather than a clean catch-up to Tokyo's +2.07% week. In India, the electoral-roll controversy escalated with Congress protests across thirteen states and threatened nationwide protests from 2 October — a domestic overhang building on a seven-week losing streak. No substantive Chinese policy development was identified; the mainland reopens today for exactly three sessions before Golden Week. Saturday's Australian auctions cleared a recomputed 48.6% nationally — 523 sold of 1,076 reported — with a 27.6% withdrawal rate and Melbourne's volumes down about 70% on the long weekend, so that city's 52% is a thin sample and not evidence of improvement.
Market dashboard
Week to 25 September — cross-asset change
| Equities | Close | 1d | WTD | Note |
|---|---|---|---|---|
| S&P 500 | 7,743.41 | +0.51% | +1.21% | ⭐ VERIFIED, not derived — AP's tabulation posted and locks arithmetically. 7,743.41 − 92.91 = 7,650.50, the verified 18 Sep close to the cent. Settled SPY ($771.35) implies 7,745.9, a 3bp tracking difference. 0.7% from the record |
| Nasdaq Composite | 27,068.72 | +0.48% | +2.06% | VERIFIED twice — AP, and a settled cash print stamped "Sep 25, 5:15:59 PM GMT-4" giving the identical level and point change. 27,068.72 − 129.34 = 26,939.38 against Thursday's verified 26,939.37. The week's best US index |
| Dow Jones | 51,828.62 | +0.93% | +0.28% | +478.64pt, the day's strongest major. ⚠ Implies a Thursday close of 51,349.98 — correcting the ≈51,378 unsettled figure published Friday by ~28pt |
| Russell 2000 | 2,837.55 | +0.07% | −0.80% | +1.98pt. Settled IWM ($281.97) implies 2,838.7, a 4bp difference. ⚠ A syndicated feed published 2,844.32 / +0.31% and was rejected — IWM's settled +0.11% refutes it and AP's weekly is internally consistent. The only major US index lower on the week |
| VIX / VIX3M | 14.87 / 17.93 | −0.80 | +0.06 | ⭐ THE THREE-EDITION GAP IS CLOSED. Dated series, tie-checked on three points: 22 Sep 14.21 matches FRED's terminal value exactly; 23 Sep 15.18 matches the carried vendor term structure; 25 Sep 14.87 is confirmed by a second unaffiliated vendor. Path 14.81 → 14.87 → 14.21 → 15.18 → 15.67 → 14.87. IVTS 0.8293, contango day 119 — deeply entrenched calm. ⚠ FRED itself still stops at 22 Sep |
| Stoxx 600 | 638.65 | +0.35% | +0.50% | ⭐ THE PRIORITY GAP IS CLOSED AND A EUROPEAN WEEKLY CHANGE EXISTS AGAIN. +2.22pt; the chain ties from the newest row back to the verified 639.92 of 23 Sep. Biggest weekly gain since early August, ending three consecutive ~−3% weeks. Last week's fabricated front row was transient |
| Euro Stoxx 50 | 6,302.82 | +0.48% | +1.07% | +30.32pt. Chain internally consistent across two rows. Gap closed after one edition |
| DAX | 25,408.64 | +0.56% | +0.4% | +142.11pt, verified on three sources; implied prior 25,267.14 against the carried 25,266.53 — an exact tie. 18 Sep level derived from the published weekly |
| CAC 40 | 8,077.80 | −0.04% | +0.16% | −3.63pt — Europe's only faller on Friday and its weakest market on the week. ⚠ Thursday's close corrects to 8,081.43 (−0.52%) from the 8,100.10 published |
| FTSE 100 | 10,695.25 | +0.14% | +0.34% | +15.26pt, reconciling to the penny from the corrected Thursday close. ⚠⚠ Thursday corrects to 10,679.99 (−0.24%) — the sign published Friday was wrong, and "Europe's only gainer" was false |
| IBEX 35 · SMI | 19,700.10 · 13,945.71 | +0.65% · +0.29% | +0.95% · +1.15% | IBEX a three-way tie; the SMI led Europe on the week and its source states the prior close explicitly |
| FTSE MIB | 51,866.93 | +0.63% | +0.62% | +324.61pt. ⚠ Thursday corrects to ≈51,542 (−0.85%) from ≈51,629 (−0.7%). FTSE Italia Mid Cap +0.21%, STAR −0.10% — large-cap financials, not breadth |
| Nikkei 225 | 66,364.20 | +1.30% | +2.07% | ⭐ VERIFIED POINT CLOSE FROM THE EXCHANGE ARCHIVE — the priority is closed. +850.21pt; OHLC 65,639.62 / 66,410.27 / 65,639.62. Opened at its low and closed 46pt off the high. The best major index on the week. ⚠⚠ The carried 24 Sep figure was WRONG: the true close was 65,513.99, BELOW the entire 65,647–65,828 range published, after an intraday spike to 66,249.11 reversed |
| TOPIX | 4,128.59 | +1.31% | — | ⭐ Four-edition gap closed, arithmetic-checked on both 24 and 25 Sep. Thursday was −0.39% to 4,075.30 — irreconcilable with a Nikkei up 1%, which is the independent evidence that the carried Nikkei figure was an intraday snapshot. TOPIX +1.31% against Nikkei +1.30% means the rally was genuinely broad |
| Hang Seng | 24,510.09 | −1.01% | ≈−0.95% | −251.04pt. ⭐ Arithmetic ties to within 0.13 of a point against Thursday — which retrospectively validates the Thursday close that would not tie last week. Technology and financials led declines on surging global yields. ⚠ Hang Seng Tech not obtainable |
| CSI 300 · Shanghai · Shenzhen | 4,439.14 · 3,888.37 · 13,317 | closed Friday | — | Levels are THURSDAY 24 September — the mainland was shut 25–27 September. Shanghai −1.22% and Shenzhen −2.34% on the last print. Reopens today for exactly three sessions before Golden Week shuts it 1–7 October |
| KOSPI | 7,080.92 | closed | — | Level is WEDNESDAY 23 September — shut Thursday and Friday for Chuseok. Reopens today into a two-session backlog of exactly the conditions its own analysts name as the 6,600 trigger: oil above $100 and UST above 5% |
| TAIEX | 48,024.60 | closed | — | Level is THURSDAY 24 September — shut Friday and again today. Reopens Tuesday with the largest information backlog in the region, having ended a two-day record run |
| Sensex · Nifty 50 | 73,895.74 · 23,140.50 | +0.43% · +0.34% | −0.54% · lower | Both tie exactly to Thursday. ⚠⚠ The Nifty logged a SEVENTH consecutive weekly loss — down 5.8% / 1,430pt over the seven weeks, its longest streak since the Covid crash. Realty and Auto led, IT and Pharma lagged; domestic 10y −1.1bp to 7.100% |
| S&P/ASX 200 | 8,665.00 | −0.43% | −0.76% | −37.02pt, verified on four sources. Lowest close since 12 June and a fourth straight weekly decline. Range 8,639.90–8,702.00 — it opened at Thursday's close, which was the day's high. Volume 589.9m units, ~19% below Thursday on the Melbourne public holiday. A-VIX 11.53 (−1.74%), tying exactly. Breadth 336/719/405. SPI unquotable, eleventh edition |
| NZX 50 | 13,811.11 | −0.10% | +0.5% | ⭐ The recurring gap is closed with a proper 16:45 close, not an intraday read. −13.72pt. A second consecutive weekly gain — one of the few regional markets up on the week. Weakness in the same rate-sensitive complex that led Australia lower |
| Indonesia IDX · Thailand SET · Philippines PSEi | 6,241.89 · 1,607.63 · 5,825.97 | — · +0.32% · +1.67% | −3.09% | Indonesia was the region's worst on the week while Bank Indonesia held at 5.75% for a third month. The Philippines posted the strongest single-day regional gain. All single-sourced |
| Rates & credit | Level | 1d | WTD | Note |
|---|---|---|---|---|
| US Treasuries — ⭐ Treasury's OFFICIAL par rows for BOTH 24 and 25 September have posted. Every figure below is the primary series; weekly changes are official 18 Sep → official 25 Sep | ||||
| UST 2y | 4.81% | −6bp | +5bp | The front end led the rally on the oil relief and the Fed-pricing fade. ⚠ Thursday's official 4.87% corrects the ≈4.92% vendor read published Friday |
| UST 3y · 5y | 4.94 · 4.98% | −5 · −5bp | +11 · +12bp | Thursday official 4.99 · 5.03. The 5-year has still not closed above 5% on the official series — the break remains intraday-only |
| UST 7y · 10y | 5.06 · 5.17% | −4 · −1bp | +13 · +16bp | ⚠ Thursday's official 10y was 5.18%, the highest since July 2007 — independently corroborated. Friday did NOT set a new high in the ten-year |
| UST 20y · 30y | 5.54 · 5.49% | +1 · +2bp | +16 · +15bp | ⭐⭐ BOTH NEW CYCLE HIGHS, set on a day the whole rest of the curve rallied. The 30-year is the highest of the cycle and the 20-year likewise. Term premium at the very long end is still widening — while Treasury buys it back (see §04 theme 2) |
| 5s30s | 51bp | +7bp | +3bp | ⚠⚠ Official path 48 (18 Sep) → 46 → 46 → 41 (23rd) → 44 (24th) → 51 (25th). V003's steepener was stopped out at 41bp and the spread has retraced ten basis points back through its own 45bp stop in two sessions. Disclosed in full in §08 — this is the least comfortable number in the note |
| 2s10s · 2s30s | 36 · 68bp | +5 · +8bp | +11 · +10bp | A clean twist steepener: 2s30s widened 8bp in a session and 10bp on the week. Official 2s10s path 25 → 20 → 25 → 26 → 31 → 36 |
| Rest of world — Friday 25 September. Weekly bond changes are unavailable outside the US and Australia: no verified 18 September levels were obtainable | ||||
| JGB 2y · 10y · 30y | 1.94 · 3.07 · 4.16% | +3.1bp | −1.2 · −0.8bp | ⭐⭐ THE SHAPE IS THE NEWS AND IT INVERTS THURSDAY. A front-end-led bear flattener: 2y +3.1bp, 5y +2.2bp, while 10s, 20s, 30s and 40s all edged lower. The 10y touched 3.11% intraday, a fresh thirty-year high, then retreated. At 1.94% the 2-year embeds ~69bp over the 1.25% policy rate — about 2.75 further hikes, far more than the ~30% carried for October. All tenors live-stamped |
| Bund 10y · 30y · 2y | 3.621 · 3.92 · 3.28% | +6.7 · +8.0bp | — | ⚠⚠ BUNDS HAVE STOPPED BEING THE EXCEPTION — they joined the rout one session late. The 10-year is described as the highest since June 2009, with ~100bp of ECB hikes priced by late 2027. ⭐ The 2-year's 2024 stale-stamp defect is cleared. ⚠ Magnitude depends on the prior accepted: +6.7bp against the carried Thursday, +1.1bp against the vendor's own prior. Direction agrees either way |
| OAT 10y · BTP 10y | 4.7255 · 4.5481% | +2.4 · +3.4bp | — | ⚠ The vendor's BTP prose ("eased to 4.50%") contradicts its own level and was discarded. BTP–Bund 92.7bp on same-vendor legs — Italy now trades ~17.7bp INSIDE France, which is the cleanest fiscal-divergence signal in Europe |
| OAT–Bund | 105.4bp | −4.5bp | — | ⭐ The dedicated same-page series, now twice-validated: its own legs reconcile to its headline, and it reproduces the carried 23 Sep 101.7bp to the decimal. A fresh 52-week WIDE of 109.9bp was set on Thursday 24 September; 52-week low 59.0bp. ⚠ A cross-page derivation gives 110.5bp and is the documented mixed-page artefact — both published, neither averaged. V025 is ~25bp clear of its 80bp stop |
| Gilt 10y · 30y · 2y · 5y | 5.3593 · 5.85 · 4.69 · 4.91% | −1.1bp (2-session) | — | ⭐ THE NAMED PRIORITY IS RESOLVED — every UK tenor carries a live Sep/25 stamp. Last week's wholesale rejection does not repeat: no stale stamps, no 2024 dates, no implied rally into a global selloff. Gilts were the G10 outperformer, essentially unchanged across two sessions of a global rout. The change shown is against the verified 23 Sep 5.37% |
| ACGB 2y · 3y | 5.01 · 5.00% | −7 · −4bp | — | ⭐⭐ THE THREE-WAY 3-YEAR CONFLICT IS RESOLVED AND THE HOUSE VIEW CAN BE MARKED TO A POINT FOR THE FIRST TIME IN FOUR EDITIONS. It was a date problem, not a vendor problem: two independent routes agree at 5.00% for 25 September, and the page's own prose states a 0.04pp fall, implying a 5.04% Thursday — which matches the previously-conflicting dedicated read exactly. The 5.09% figure was the stale outlier. The front end rallied into the RBA |
| ACGB 10y · 5y · 7y | 5.39 · 5.05 · 5.18% | unch to +1.7bp | — | 10y corroborated on two sources. ⚠ The daily change is genuinely unresolved: the vendor's own prior implies +1.7bp, the carried Thursday implies 0 to −1.8bp — opposite signs, both published. The bull-steepening conclusion is robust to either |
| ACGB 3s10s | 39bp | ≈+4bp | from 43bp entry | ⚠ Now markable to a point — and the honest mark is less flattering than the range published Friday. 39bp against a 43bp entry is 4bp in the money, not the 7–13bp the 30–36bp range implied. The curve steepened INTO the meeting, which is the market marking the hike as near-terminal |
| Canada 10y · Swiss 10y | 3.93 · 0.6488% | disputed · +1bp | — | ⚠ Canada's daily change is a directional conflict — the vendor's own prior implies −7bp (matching its oil-driven narrative), the carried 3.881% implies +4.9bp. The 12bp gap fits the one-session-stale pattern proven on CNY and MXN this week; neither asserted. The Swiss 10y is +28bp in a month — the market is pricing the SNB off the zero floor |
| Credit — ⭐ all three series posted together at 24 September, and they widened in beta order. No 25 September observation exists for any of them | ||||
| US IG OAS | 79bp | +2bp | +2bp | Path 77 / 77 / 77 / 77 / 79 across 18–24 Sep. The investment-grade end barely moved |
| US HY OAS | 280bp | +7bp | +12bp | Path 268 / 266 / 268 / 273 / 280. ⭐ Independently matched tick-for-tick on all five observations by a second source — eleven consecutive matches. At 280bp it still sits in the richest decile against a long-run median near 450bp |
| US CCC OAS | 1,112bp | +19bp | +29bp | ⭐⭐ THE CLEANEST RISK SIGNAL IN THIS NOTE. Path 1,083 / 1,077 / 1,075 / 1,093 / 1,112. CCC widened eighteen times as much as IG. Beta-ordered widening is credit risk repricing, not a rates pass-through. V017 is now 38bp from the 1,150bp level that CONFIRMS it, having been 25bp from the level that closes it — the view is working and Friday's conviction cut is reversed |
| SoftBank $11.1bn BB+ — the secondary | not obtainable | — | — | ⚠ The literal question this desk named — did the tranches hold reoffer — cannot be answered: five sources carry final coupons only, with no secondary levels, no break price and no new-issue concession. But the index answers it indirectly: the deal priced 23–24 September and over exactly those two sessions HY widened 12bp and CCC 37bp. Pricing through talk bought the issuer a good outcome; the secondary market paid for it. Derived inference, not reported colour. ⭐ Book revised up to "over $30bn of dollar orders" from the carried $20bn+ — headline-level only |
| FX | Fri close | 1d | WTD | Note |
|---|---|---|---|---|
| ⭐ Every pair below is taken from a DATED historical table with open-matches-prior-close verified across at least three rows. All twelve pairs passed. The newest row on every table is Friday 25 September — a session that has closed | ||||
| DXY | 100.97 | −0.31% | +0.75% | ⭐ The dollar-reconciliation test passes: weight-summing the legs gives a weekly +0.73% against the table's +0.75% — a 2bp residual. On the single Friday session the sum gives −0.21% against −0.31%, and the 0.10pp residual is consistent with a snapshot-time offset in the yen leg, which traded a 203-pip range. Flagged rather than hidden; the weekly test is the stronger one |
| EUR/USD | 1.1391 | +0.09% | −0.81% | Series 1.1449 / 1.1382 / 1.1381 / 1.1391. The move was concentrated on Wednesday 23 September, not Thursday — that was the repricing day across the whole complex |
| USD/JPY | 157.28 | −0.99% | +0.25% | ⭐⭐ A 203-pip reversal from a 158.97 high, and it breaks the carried "dovish hike" read. The yen gained 0.99% on the same session JGB 2-year yields rose 3.1bp — both legs hawkish, the opposite configuration to Thursday. Attributed to intervention caution near 160 plus short-carry unwinding. Finance Minister Katayama, 25 Sep: the post-coordinated-intervention "foreign exchange principles remain in effect." 160 is behaving as a policy level, not a technical one |
| GBP/USD | 1.3253 | +0.26% | −1.06% | Two exact open-matches-prior-close links. Friday's low of 1.3207 was a three-month low before the recovery |
| AUD/USD | 0.7024 | +0.17% | −1.33% | ⭐ The series validates itself: its 18 September close prints 0.7119, matching this desk's independently verified figure exactly. Path 0.7115 / 0.7039 / 0.7012 / 0.7024. The worst G10 performer on the week, and 70.2 US cents was independently corroborated. A sell-side year-end target was cut from 0.73 to 0.72 with sub-0.700 flagged as the risk |
| NZD/USD | 0.5658 | −0.07% | −1.14% | Path 0.5728 / 0.5674 / 0.5662 / 0.5658. ⚠ The carried Thursday 0.5689 was 27 pips too high — the error that corrupted last week's AUD/NZD mark |
| USD/CAD · USD/CHF | 1.4144 · 0.8285 | +0.05% · +0.08% | +1.10% · +0.77% | USD/CAD has three consecutive exact open-matches-prior-close links — the strongest chain evidence in the set |
| USD/CNY · PBoC fix | 6.7204 · 6.7489 (24 Sep) | +0.09% | +0.34% | ⚠ The dated table's Friday CNY row was REJECTED — it served open = high = low = close with a 0.00% change, a degenerate stale carry-forward on a day the onshore market traded. A second vendor's 6.7204 was substituted and its implied Thursday sits within 11 pips of the rejected table's Thursday. No Friday fix exists — the mainland was shut. The 24 Sep fix at 6.7489 sits ~285 pips ABOVE spot, i.e. fixing the yuan weaker than the market: a lean against appreciation. Today's fix, the first since the holiday, is the cleanest read on whether the PBoC wants stability through Golden Week |
| USD/MXN · USD/INR | 17.6844 · 95.818 | −0.23% · −0.15% | +2.65% · −0.12% | ⚠ The carried Thursday MXN mark was 182 pips off and was provably Wednesday's close. The peso was the week's worst EM performer at −2.65%, consistent with a Banxico that has just pre-committed to decoupling from a hiking Fed |
| USD/KRW | 1,354.24 | −0.91% | −2.30% | The standout EM performer, +2.30% on the week on a Monday–Tuesday rally, despite Seoul being shut for the last two sessions. Three chain links within 0.06 won |
| AUD/NZD | 1.2414 | +0.24% | −0.20% | ⭐ Cross-computed from two legs of the same dated table; no cross quote page was fetched. Path 1.2422 / 1.2406 / 1.2384 / 1.2414. ⚠ The carried ≈1.2329 Thursday mark was wrong purely because the NZD leg was wrong — the correct Thursday cross is 1.2384. V023 is +0.80% from its 1.2315 entry, which is better than the roughly-flat figure restated on Friday and far below the +1.05–1.67% originally published |
| AUD/JPY · EUR/JPY | 110.47 · 179.16 | −0.83% · −0.91% | −1.09% · −0.56% | Both cross-computed from same-table legs. AUD/JPY is now inside 50 pips of the 109–110 tripwire flagged in prior editions |
| Commodities & digital | Last | 1d | WTD | Note |
|---|---|---|---|---|
| Brent (Nov-26, LCOX6) | $104.32 | −2.14% | +0.43% | ⚠⚠ THE CARRIED THURSDAY FIGURE WAS WRONG BY $3.77 AND THE CLUSTER THIS DESK REJECTED WAS THE RIGHT ONE. The dated table and an independent vendor's stated previous close both put 24 September at $106.60, not the $102.83 published. Triple-verified for Friday; the table is internally consistent with no phantom Sunday row. Brent is roughly flat on the week while WTI fell 3.8% — a Gulf-located, not US-located, risk premium. ⚠ The vendor's own weekly (">1%") does not reconcile with the derived +0.43% |
| WTI (Nov-26) | $92.41 | −2.33% | −3.82% | Exact match on two sources. ⚠ Thursday restates to $94.61 from the $93.66 carried. Brent–WTI widened to ≈$11.91 |
| Henry Hub · TTF | $3.196 · €71.89 | −3.06% · −4.07% | — | Henry Hub gave back Thursday's 8.17% spike. TTF fell on Hormuz reopening talk; EU storage ~70%, below the 80% target and below seasonal norm — the pre-winter vulnerability is intact and is now a demand-side risk as much as a price one |
| Gold (spot) | $4,285 | +0.24% | ≈−1% | ⭐ Two-source verification via the implied cross: a timestamped $4,284.20 with a $4,254.40–4,316.60 range, and an independent CAD print of C$6,069.69 implying USDCAD 1.4167 against a verified 1.4144 close. A THIRD consecutive close below $4,300 — and it traded up to $4,316.60 intraday and failed there, which is a worse signal than a clean close below. ⚠ A syndicated feed's $4,321.20 is rejected. V014's stop continues to look right |
| Silver · Platinum | $64.26 · $1,777.60 | +1.01% · +1.43% | ≈−3% | Silver corroborated to 0.1% via the same CAD cross. Still the week's worst major metal. The cited driver is rising Treasury yields and expectations of further Fed hikes — note that framing cuts against precious metals |
| Copper — LME cash · 3M · Comex | $14,740 · $14,647/t · $6.70/lb | −0.17% · +0.05% · +0.22% | +1.45% · +0.91% | ⭐ The primary reproduced all three carried observations exactly. Cash−3M path: −$5 (16 Sep) → −$8.50 → +$14 → +$58 → +$37 → +$77 → +$125 → +$93. Eight sessions, five direction changes — do not extrapolate from any single reading. But the trend is unambiguous: backwardation has widened 6.6× in five sessions while LME stocks fell 3,600t. Tightening physical, choppy execution. ⚠⚠ The carried "Comex stocks 254,250t" was mislabelled — that is the LME figure. This desk holds no Comex copper stock number |
| Aluminium · Zinc · Nickel | $3,281.50 · $3,905.90 · $16,359.75 | +0.79% · +0.63% | — | Nickel −0.36%. ⚠ The vendor's own change column disagreed with its own levels on zinc (stated −1.78%) and gave two different nickel figures; all recomputed |
| Iron ore | $97.06/t | −0.08% | — | A FOURTEENTH consecutive sub-$100 reading, in a four-session band of $97.06–97.30. Extraordinarily inert. China is shut 1–7 October, so no fresh demand signal is coming for a week. The retirement of the fade-above-$100 view on Friday looks straightforwardly right |
| Lithium carbonate · Uranium | CNY 133,150/t · $89.50/lb | −1.52% | −15.2% m/m | Lithium is at its lowest level year-to-date on higher stockpile estimates and increased Australian output — the only metal in outright downtrend, and the reason ASX lithium names fell Friday |
| Bitcoin | ≈$84,728 | +0.53% | — | Level ~20:00 UTC Sunday 27 Sep; sign and percentage from an independent source stamped 00:05 UTC. Three sources span $84,339–84,728, a 0.46% spread. No sign inversion this edition and no duplicate-percentage failure |
| Ether · Solana | $2,692.41 · $122.98 | +0.25% · disputed | — | ETH essentially unchanged since Thursday. ⚠ SOL's 24-hour sign is unresolved — a −0.28% print against +4.92% since Thursday, on precisely the asset where the sign-inversion failure has recurred. The multi-day move is the reliable part: SOL is the weekend outperformer |
| XRP · BNB · total cap · dominance | $1.53 · $778.61 · $2.971trn · 57.3% | flat · −0.06% | — | ⭐ BNB is publishable this week — cross-checked against the carried prior after last week's 20%-low bad fetch. Dominance unchanged |
| Spot ETF flows | BTC +$134.5m · ETH +$87.0m (25 Sep) | decelerating | — | ⭐⭐ THE PLACEHOLDER RULE IS NOW VINDICATED WITH A MEASURED MAGNITUDE: ETH's 23 September row, carried last week as "incomplete at $2.5m", has settled at $104.5m — 42 times higher. Publishing $2.5m as a near-zero day would have been wrong by two orders of magnitude and would have supported the opposite narrative. BTC path +999.0 → +714.7 → +346.9 → +190.7 → +134.5: five straight inflow sessions, but −87% across the week. Positive direction, fading impulse |
| Futures open interest · funding | BTC $27.5bn · ETH $18.3bn | +0.98% · +2.14% | −4.2% vs Thu | Per-asset scope, not market-wide. BTC open interest is $1.2bn below Thursday — the deleveraging continued then stabilised. Funding mildly positive on both (BTC +0.0028%, ETH +0.0080%): no crowded-long stress. 24h liquidations $106.2m of which only 19% was BTC+ETH — quiet at the core, messy in the tail |
Conventions: 1d = change on Friday 25 September; WTD = against the verified Friday 18 September close where one exists, otherwise marked. "≈" marks a derived value. US Treasury levels are Treasury's OFFICIAL daily par (CMT) series for both 24 and 25 September — the primary, not a vendor read; par runs 1–2bp from on-the-run quotes. US cash index closes are AP's tabulation, verified by arithmetic reconciliation against two independent anchors, with settled 16:00 ET ETF closes as the cross-check. No CFD or futures quote was used for any cash index. Every FX pair is from a dated historical table with open-matches-prior-close verified; crosses are computed from two legs of that same table and no cross quote page was used. Brent and WTI are settles with the contract named. Gold is spot. LME copper is the official settlement. Every vendor change column was recomputed from levels against a verified prior. Mainland Chinese, Korean and Taiwanese levels are the last actual close and the date is stated — no price is quoted for a market that was shut.
What is driving markets
1. The energy shock's paper leg de-escalated and its physical leg did not
Friday was a clean de-escalation trade: Brent −2.14%, WTI −2.33%, TTF −4.07%, the US 2-year 6bp lower, equities up across the board and gold's safe-haven bid absent. It priced one thing — Araghchi's seven-day roadmap, with Hormuz reopening at the end of day seven. That premise was publicly destroyed on Saturday when Trump rejected the offer, and the same day the Houthis put drones over Riyadh and a ballistic missile at Khamis Mushait. Meanwhile the physical picture on Friday was not the picture the paper market traded. Tracked tanker transits through Hormuz were in the low single digits per day — a seven-day tanker average of 1.3 on one provider, 13 vessels of all types on another, with LNG at zero on the first — and VLCC rates are at a record $1.27m/day, which is a freight market pricing a closed strait, not a reopening one. The Saudi East–West pipeline did restart on 22 September, which settles a two-week argument in favour of the "within days" camp on the restart and the three-to-five-week camp on full throughput; but the Red Sea terminus at Yanbu has now become the exposed node, so the workaround has acquired the vulnerability it was built to avoid. Read the divergence this way: the paper market is trading the diplomacy and the physical market is trading the tankers, and only one of them has moved.
2. The long end is the stress, and the Treasury is losing the argument with it
This is the running theme and it acquired its cleanest test this week. Treasury doubled its long-end liquidity buybacks from $2bn to at least $4bn per operation in the 10–20y and 20–30y buckets, effective 9 September through 4 November, and the Secretary has said publicly he is ready to increase them further to push down long-term yields, with reports he could tap the Treasury General Account to fund it. Three weeks into that programme the 30-year printed a new cycle high of 5.49% and the 20-year a new high of 5.54% — on a session when the 2-year rallied 6bp and every other tenor fell. 5s30s went 41bp to 51bp in two sessions and 2s30s widened 8bp in one. The same repricing is global and it is no longer confined: the Bund 10-year reached its highest since June 2009 and finally joined the rout after being the conspicuous holdout, the JGB 10-year touched a fresh thirty-year high of 3.11% intraday, and OAT–Bund set a new 52-week wide of 109.9bp on Thursday. The one exception was gilts, essentially unchanged across two sessions — and that was into a Bailey who had just turned hawkish. The policy question is now binary: either the buyback sizing escalates materially at the 4 November quarterly refunding, or the market concludes the tool cannot influence term premium.
3. Credit is repricing beta-first, and it is the cleanest risk signal in the note
All three spread series posted together at 24 September and they widened in strict beta order: IG +2bp to 79, HY +12bp to 280, CCC +37bp to 1,112. CCC widened eighteen times as much as IG. That ordering is the whole point — a rates move passes through the whole stack roughly proportionally, whereas widening concentrated at the low-quality end is credit risk being repriced. The timing makes it sharper: SoftBank's $11.1bn BB+ — the largest high-yield corporate bond sale on record globally, beating a 2014 deal, and the largest ever by an Asia-Pacific issuer — priced on 23–24 September through talk on every dollar tranche, into a book since revised up to "over $30bn of dollar orders". Over exactly those two pricing sessions the index cheapened 12bp at HY and 37bp at CCC. The issuer got a good outcome; the secondary market paid for it. The analyst read is split in a way that matters: Fitch was "positively surprised by the market appetite", noting AI-driven debt issuance "has reached the high-yield market at scale", while CreditSights called risks for the credit "material and increased as concentration has increased and cash flow has come under material strain." Equity rewarded the funding — SoftBank shares +7% — and credit indices did not.
4. Japan is now a hawkish-repricing story, not a dovish-hike one
This theme reverses. The carried interpretation — that the 18 September hike was read as dovish — rested on Thursday's signature of JGB yields at a thirty-year high with a weaker yen. Friday produced the opposite configuration on both legs and a specific cause. A BoJ official indicated the Bank expects to raise rates roughly once every three months, potentially lifting them to 2% by around June next year — which from the September move implies December, March and June. The curve priced it in exactly the right shape: 2-year +3.1bp to 1.94% and 5-year +2.2bp, while the 10-year eased 1.2bp and the 20s, 30s and 40s all fell. A front-end-led bear flattener with the long end bid is "more hikes sooner, inflation credibility improving", not a term-premium event. The yen gained 0.99% and the megabanks rose about 4% on a +1.30% index — banks up four per cent on a 1.3% day is a rate-path trade. At 1.94% the 2-year embeds roughly 69bp over the 1.25% policy rate, about 2.75 further hikes, which is materially more than the ~30% for October this desk has been carrying. The missing leg is fiscal: the Takaichi government is preparing a record budget while the Bank normalises, and that combination is what took the 10-year to a thirty-year high in the first place.
5. Three mechanical bids withdraw at once, into internals that are already rotting
The index is 0.93% from its record and the participation underneath it is not. Only 49.1% of S&P 500 constituents sit above their 200-day moving average — through 50% for the first time in the carried series, and 11 to 13 points below each of the last two year-ends. On Friday, a session the index rallied 40 points, there were 308 new 52-week lows against 46 new highs, a 6.7:1 ratio, with the McClellan Oscillator at −22.59 and three of four Hindenburg conditions met. The qualifier matters and is stated rather than buried: the condition that fails is the binding one — new highs at 0.97% are far below the 2.2% threshold — so this is deteriorating internals, not an imminent signal. Against that backdrop three sources of mechanical demand step back simultaneously. The corporate bid goes from 10% of index weight in blackout on 18 September to 61% by 30 September and does not reopen until 1 November — now confirmed from a dated, attributed primary source rather than carried. CTAs are asymmetric 21:1: one house estimates they could sell $84bn of global equities in a down market while buying only $4bn in an up market, with their US equity z-score already down from +2.4 to +1.1. And vol-control funds sit at ~86% equity exposure, the highest since March — little room to add, meaningful room to shed. The one offset is that dealer gamma flipped positive on Friday, from −$20.26bn to +$32.56bn in a single session, which damps moves while it lasts.
Central bank watch
Fed funds pricing — implied probabilities by meeting
| Bank | Policy rate | Last move / vote | Next decision (Sydney) | Market pricing | Bias |
|---|---|---|---|---|---|
| Fed | 3.75–4.00% | +25bp 16 Sep, 12–0. Chair Kevin Warsh since May 2026. Named roster confirmed structurally unpublished — see §13 | Wed 28 Oct · 05:00 Thu 29th AEDT · no SEP | ⭐ Oct hike 66.6% on a SETTLED read, down from a 71.2% pre-close. Dec modal two hikes 51.6% — a coin flip. Polymarket event page 65%, agreeing within 1.6pp | Hawkish |
| BoJ | 1.25% | +25bp 18 Sep, 7–2. Effective 24 Sep — highest in 31 years | Meeting 29–30 Oct, decision Fri 30 Oct (the Bank's own schedule, tenth consecutive edition) | ⚠ ~30% for 1.50% in October is CARRIED AND UNREFRESHED — treat as unverified. The 2-year at 1.94% embeds ~69bp, about 2.75 hikes. An official signalled hikes ~quarterly to 2% by June 2027 | Hiking, now guided |
| RBA | 4.35% | Hold since 11 Aug; all four majors call +25bp to 4.60% | TUE 29 SEP 14:30 AEST · presser 15:30 (board sits 28–29 Sep) | ⭐ 88–90%, DOWN from 96.5% — and for the first time the tracker and the bonds agree. Tracker 90.0%, sell-side 88%, ACGB 3y at 5.00% embeds +65bp and the 2y +66bp. Market prices a further ~40bp beyond tomorrow; CBA says 4.60% is the peak | Hiking tomorrow |
| ECB | DFR 2.50% | +25bp 10 Sep | Thu 29 Oct · 00:15 Fri 30th AEDT | ⭐ The 29%–51.6% span is RESOLVED: ~48% for October. The 29% figure is dated 10 September and predates the 23 Sep PMI — discard it. ⚠ Lane's 24 Sep remarks argue for December over October, so the live figure is plausibly in the low 40s | Hike coming, Oct a coin flip |
| BoE | 3.75% | Held 17 Sep, 6–3 (Bailey in the majority) | Thu 5 Nov · 23:00 AEDT · with a Monetary Policy Report | ~81% November (carried, not re-verified). ⭐ Bailey turned hawkish Friday: "it's going to get harder to maintain that stance the longer we have high energy prices." This validates the pricing rather than fighting it. Ramsden on QT today 20:00 AEST | Hawkish hold |
| Norges | 4.50% | +25bp 24 Sep with an MPR, all members supported | Thu 5 Nov | ⭐ THE NAMED GAP IS CLOSED. Path: the rate "remains close to the current level for a period ahead before declining somewhat" and stays elevated "somewhat longer than the June forecast". Bache: prepared to raise again if the outlook warrants. One house reads ~40% for another hike within six months. ⭐ EUR/NOK 10.79 → 10.75 on the announcement | Hawkish hold |
| Riksbank | 1.75% | Held 24 Sep (meeting in Gothenburg 23 Sep) | Wed 4 Nov | ⭐ STRONGER THAN CARRIED: a 2026 hike is now in the BASELINE, not merely likelier. Verbatim: "the Riksbank assesses that the increases in the policy rate will begin this year." Hawkish risk clause on a faster pace. ⚠ Vote still not obtained; the MPR records no reservations | Hiking this year |
| SNB | 0.00% | Held 24 Sep, fifth consecutive | Thu 10 Dec | Inflation forecast raised across the whole horizon — 2026 0.7%, 2027 0.8%, 2028 0.8% — on energy. The Swiss 10y at 0.6488% is +28bp in a month: the market is pricing the SNB off the zero floor | Extended hold, hawkish tilt |
| BoC · RBNZ | 2.25% · 2.75% | Held 2 Sep · +25bp 2 Sep | Wed 28 Oct (both) | BoC neutral. RBNZ October ~31% (carried, not re-verified). ⚠ The RBNZ date remains unconfirmed against the issuer for a second edition | Neutral · Tightening |
| Banxico | 6.50% | Held 24 Sep, unanimous 5–0, third consecutive | Not obtained | ⭐ The guidance gap is filled and it is dovish at the margin: Banxico softened its language to state that policy would NOT mechanically follow Fed decisions, and that future decisions will consider the ongoing disinflation process. Consistent with the peso's −2.65% week | Decoupling |
| PBoC | LPR 3.00% / 3.50% | Unchanged 21 Sep, 16th month (carried) | Mon 19–20 Oct | No Friday fix — the mainland was shut. The 24 Sep fix of 6.7489 sits ~285 pips ABOVE spot, fixing the yuan weaker than the market. Today's fix is the first since the holiday and the cleanest read on Golden Week intentions | Easing bias |
| Emerging markets — where a next date is "not obtained" it was not confirmable against an issuer this run | |||||
| Korea · Taiwan | 3.00% · 2.00% | +25bp 27 Aug · held 17 Sep, highest since 2008 | Thu 22 Oct · 17 Dec | Korea consensus 3.25% by October. Taiwan's date is now issuer-confirmed | Tightening · Hold |
| India · Brazil | 5.25% · 13.75% | Held Aug · eased Sep | Wed 7 Oct · October | India consensus 5.50% — a hike priced. The domestic 10y eased 1.1bp to 7.100% on Friday | Tightening · Easing |
| South Africa · Indonesia | 7.25% · 5.75% | +25bp 23 Sep unanimous · held, third consecutive month | Not obtained | Indonesia is the watch item: BI on hold three months while the IDX fell 3.09% on the week, the region's worst, with the rupiah exposed to a 5.17% UST and Brent above $104. Least room of any regional bank | Hiking on oil · Hold |
| Czech · Philippines · Hungary · Turkey | 3.75% · 5.00% · 6.50% · 37.00% | Held 17 Sep unanimous, 2nd pause · +25bp 27 Aug, 3rd hike · held · held | 5 Nov · 22 Oct · — · — | Czech and Philippine rates and dates are issuer-confirmed this edition. ⚠ Hungary and Turkey are carried rather than re-verified | Mixed |
Fed detail. Sixteen of nineteen FOMC members projected at least one further hike before year-end in the September dot plot. Warsh on 16 September: "This summer's inflation readings do not tell me that underlying trends have meaningfully improved. Inflation is too high and has been for too long." The statement's closing line — "The Committee will deliver price stability" — is notably assertive. Williams on 24 September called another hike by year-end "reasonable" and said repeated supply shocks "cannot be ignored", and Hammack, a 2026 voter, spent Friday warning about the formation of an "inflationary mindset" and explicitly declining to treat 5%-plus yields as a policy problem, saying current yields reflect growth, US debt and the rate path. A methodological note that has cost this desk coverage: the Board's speeches page lists Governors only — Reserve Bank presidents, including voters like Hammack, never appear there, so a clean negative on that page is not a clean negative on Fed speak. The next blackout begins Saturday 17 October. The live tail risk remains Governor Cook: the Supreme Court ruled 5–4 in her favour on 29 June, holding that "for cause" removal requires notice and an opportunity to respond, and on 7 August the White House notified her it was "considering" removal on mortgage-fraud allegations with a three-week response window — i.e. running the very process the Court mandated. She continues to serve and to vote. What happened after that deadline could not be established and is a standing daily check.
RBA detail, and the guidance question that decides two house views. The logistics are issuer-confirmed from a page last updated 23 September: board 28–29 September, decision 14:30 AEST Tuesday, media conference 15:30. The speeches list proves a negative — no RBA speech between 23 and 28 September, so the Bank has been silent for five sessions into the decision — and it also corrects a date this note has carried: Bullock's CEDA fireside chat is dated 22 September by the RBA, not 21 September. The split on guidance is real. CBA has 4.60% as the stopping point, calling it "firmly restrictive" and expecting the Bank to leave the door open without further hikes as base case, with a Q3 trimmed mean at or above 1.0% q/q the trigger for 4.85%; it still forecasts 50bp of cuts in 2027 with the first pushed to August. The sell-side alternative has the statement emphasising that "inflation risks remain tilted to the upside and that further vigilance is required" and the market pricing a further ~40bp over nine months. The evidence tilts toward the CBA read for six reasons set out in §07, of which the strongest is new: a voting Monetary Policy Board member publicly rejected the wage-price-spiral thesis on 22 September, citing a WPI that rose more slowly than inflation post-pandemic, only ~10% of enterprise-agreement-covered employees renegotiating annually, union density down from above 50% to 13%, and the RBA's own forecast that labour cost growth moderates as spare capacity emerges. Watch the 15:30 presser more than the 14:30 statement — the 4.5–5% band question is a press-conference question. And note the sequencing: the monthly CPI indicator prints at 11:30 Wednesday, twenty-one hours after the decision, so the Board writes tomorrow's statement blind to it.
ECB detail. The contested October range is resolved as a date-quality artefact rather than a genuine disagreement: the 29% reading traces to a piece dated 10 September, the day of the hike itself and thirteen days before the PMI beat. The current figure is ~48%, up from ~45% before the 23 September PMI (composite 53.1 against 51.7 expected, services 53.0, manufacturing 52.7). The speaker slate then cut the other way and is worth reading as a whole: Lane — the Chief Economist — argued on 24 September in a way that strengthens the case for December rather than October, having said a day earlier that future decisions "may be less straightforward" than September's; Kocher preferred reaching 2% without further hikes; Radev wanted vigilance over patience but "not a call for automatic tightening"; Vujčić was hawkish on diesel and refining capacity; Sleijpen called growth "considerably more resilient than expected". Net: a hike is coming, October is a coin flip skewed to December. Institutionally, Schnabel leaves the Executive Board on 4 January 2027 to become IMF Financial Counsellor.
Regional briefs
United States
A risk-positive tape with a defensive-hawkish composition. The first winning week in three, the S&P 0.7% off its record, the Dow up 0.93% on Friday — and none of it came from growth or the Fed. It came from one exogenous input, oil coming off on Iran diplomacy that was dead by Saturday. Underneath: the 30-year and 20-year both set new cycle highs, 2s30s widened 10bp on the week, core capital goods orders rose 1.6% validating the FOMC's "capital investment is robust" language, and a sitting voter spent Friday warning about an inflationary mindset. Against that, Michigan sentiment finished September at 48.1, a four-month low and 15% below January, with one-year inflation expectations at 4.6% against 3.4% before the Iran conflict. Settled Fed pricing has October at 66.6% and December's two-hike path at a coin-flip 51.6%. Government funding is not a near-term risk — a clean CR passed the Senate 90–6 and runs to 11 December, and the debt ceiling does not bind until 2027. This week compresses into Wednesday: core PCE lands in the same minute as personal spending, with the 2y, 5y and 7y notes all settling that day and no coupon supply to absorb a surprise until 6 October. Q3 earnings have barely begun — nine S&P companies reported, banks start 13 October.
Euro area
The best equity week since early August, on the worst consumer print of the month. The Stoxx 600 rose 0.50% and the Euro Stoxx 50 1.07%, snapping three consecutive ~−3% weeks, with banks +1.3% and energy −1.3% — a pure lower-oil rotation, large-cap financials rather than breadth. But German October consumer climate fell to −30.6 against −27.4 expected, a near-two-year low driven specifically by a significant drop in income expectations while economic expectations were virtually unchanged. That is an energy-cost real-income shock, not a growth scare, and it sits directly against the 23 September PMI beat. It also makes Friday's euro-area flash CPI consensus of 3.7% headline, up from 3.2%, look demand-destructive rather than merely inflationary. Bunds finally joined the global rout, the 10-year at its highest since June 2009 with ~100bp of ECB hikes priced by late 2027. France is the fiscal story and its catalyst is this week (see §11): €54bn of consolidation, a 5.0% deficit target against a −5.4% no-measures baseline, and the arithmetic problem that €54bn of announced effort buys only ~0.4pp. Italy is the counterpoint — committed to a 2026 deficit around 2.8% despite weaker growth — and BTPs now trade ~17.7bp inside OATs, which is the cleanest fiscal-divergence expression in Europe.
United Kingdom
The most hawkish Bailey of the cycle, and gilts that did not care. At Oxford on Friday the Governor said "it's going to get harder to maintain that stance the longer we have high energy prices" and that the Bank "could not afford to wait for the full evidence" on energy pass-through into expectations, while calling the existing evidence "quite subdued". He was in the 6–3 majority to hold on 17 September, and he plus three deputy governors have signalled openness to increases. Gilts were nonetheless the G10 outperformer, essentially unchanged across two sessions of a global selloff — the 10-year at 5.3593%, 30-year 5.85%, and every tenor carrying a live timestamp, which resolves the wholesale rejection of last week. The Budget is 28 October and it carries the OBR's first official headroom reassessment in eleven months, the government having legislated that the rules are assessed only annually at Budgets. The last official figure is £22bn from November 2025; credible third-party estimates run £8–11bn, with the £8bn figure explicitly embedding the economic impact of the Iran conflict. Two scheduled events this week are larger than they look: Ramsden on quantitative tightening today, and Alan Taylor — the committee's most dovish dissenter — delivering a named lecture on inflation persistence on Tuesday. A dedicated QT speech in the same week as a 10-year gilt auction and a month before that Budget is a gilt event, not a technical aside.
Japan
The best major equity week anywhere, on a hawkish repricing rather than despite one. The Nikkei rose 1.30% to 66,364.20 and 2.07% on the week; the TOPIX matched it at +1.31%, so this was a market-wide re-rating rather than a heavyweight artefact. The cause is a BoJ official signalling hikes roughly once every three months, potentially to 2% by around June 2027, and the curve took the right shape for it: 2-year +3.1bp to 1.94%, long end bid, megabanks up about 4%. The 10-year touched 3.11% intraday, a fresh thirty-year high, before closing lower at 3.07%. The yen gained 0.99% to 157.28 after a 203-pip reversal from 158.97, with Finance Minister Katayama invoking the post-coordinated-intervention "foreign exchange principles" — 160 is a policy line. The missing leg is fiscal: the Takaichi government is preparing a record budget while the Bank normalises, which is precisely the combination that produced the thirty-year yield high. This week: industrial production and retail sales Tuesday, the Summary of Opinions and the Q3 Tankan on 1 October JST (large manufacturers consensus +25 to +26 against +22 prior, a sixth consecutive improvement), and Tokyo CPI with consensus jumping 1.9% to 2.3%. The Summary is the document that confirms or breaks the new front-end pricing.
China & Hong Kong
Three sessions, then the lights go out for a week. The mainland was shut 25–27 September for Mid-Autumn and reopens today into a window of exactly three sessions before Golden Week closes it 1–7 October. Its last print was weak — CSI 300 −1.73%, Shanghai −1.22%, Shenzhen −2.34% on 24 September, attributed to uncertainty over the truce extension. Hong Kong, which stays open throughout, fell 1.01% to 24,510.09 on surging global yields with technology and financials leading. The event to flag is Wednesday's NBS manufacturing PMI: consensus 50.1 against a 49.8 prior, i.e. the market expects the factory sector to cross back above the expansion line — and it prints on the last session before a seven-day closure, with non-manufacturing still expected to be contracting at 49.3. A miss leaves the mainland shut for a week with no way to trade it, which makes Hong Kong the only expression vehicle — that asymmetry is the tradeable structure of the week. Also Wednesday: a trade balance consensus of $79bn against $119.1bn prior, a $40bn expected narrowing. On the truce, the two-month extension to 10 January 2027 is confirmed across five-plus outlets, but the rare-earths gap is now filled and the answer is that nothing was agreed: tariffs, purchases, technology restrictions and rare-earth supply all remain unresolved, Beijing maintains controls on dysprosium, terbium and yttrium, and magnet shipments to the US fell to 512 tonnes in August, −20% m/m and −13% y/y. The flow is deteriorating while the truce is extended. Note the collision: US DFARS magnet restrictions enter a new phase on 1 January 2027, nine days before the truce expires.
Emerging Asia
Korea reopens today into a backlog of exactly the conditions its own analysts flag as the downside trigger. The KOSPI last closed at 7,080.92 on Wednesday, shut Thursday and Friday for Chuseok, so it missed Friday's regional risk-on. The Sunday domestic read names the bull case (7,500–7,700 on semiconductor momentum plus summit expectations, with foreign investors buying a net ₩400bn on 18 September, the first net buying in eight sessions) and the bear case explicitly: oil above US$100 and US Treasury yields above 5%, which previously drove the index to the 6,600 region. Both are live. Expect a gap-and-fade rather than a clean catch-up to Tokyo. Taiwan is shut again today and reopens Tuesday carrying the largest information backlog in the region, having ended a two-session record run at 48,024.60. India rebounded on Friday but the Nifty still logged a seventh consecutive weekly loss — down 5.8% over seven weeks, the longest streak since the Covid crash — with a domestic political overhang building toward threatened nationwide protests from 2 October. Indonesia was the region's worst on the week at −3.09% while Bank Indonesia held for a third month: with the rupiah exposed to a 5.17% ten-year and Brent above $104, it is the regional market under the most visible pressure and the central bank with the least room.
Australia & New Zealand
The RBA: the hike is not the risk, the guidance is
The board sits today and tomorrow; the statement lands 14:30 AEST Tuesday with a press conference at 15:30, both confirmed from the RBA's own forward diary on a page last updated 23 September. All four majors call +25bp to 4.60% from a cash rate held at 4.35% since 11 August. What changed over the weekend is conviction about what comes after: pricing came back from 96.5% to 88–90%, and this time the derivatives and the physical curve moved together — the ACGB 2-year rallied 7bp to 5.01% and the 3-year 4bp to 5.00%, taking the embedded premium from ~73bp to ~66bp, and 3s10s steepened about 4bp into the meeting. That is a market marking this as near-terminal.
Six strands support the "last hike" case, and one of them is new to this note. First, Bullock's own stated test is already met: at CEDA on 22 September she said unemployment "between 4.5 and 5" would "probably take enough heat out of the labour market", and August printed 4.6%. Second — and this has not appeared here before — a voting Monetary Policy Board member publicly dismantled the wage-price-spiral premise five sessions before the decision. Iain Ross, 22 September: "there is no evidence of the emergence of a wage-price spiral in the present circumstances and recent data suggest such an outcome is unlikely." His evidence: the wage price index rose more slowly than inflation post-pandemic; only about 10% of enterprise-agreement-covered employees renegotiate annually, so three-year agreements act as a shock absorber; union density has fallen from above 50% to 13%; comparative wage justice and quarterly indexation are both long abandoned; and disputation is sharply lower. He spoke citing unemployment at 4.5% — before the 4.6% print, so the slack he describes has since increased. Third, the ABS itself caveated the print that drove the repricing, flagging potential methodological impacts from Labour Force Modernisation changes on August data reliability and an unusually high share of people moving straight from outside the labour force into unemployment. Fourth, the domestic transmission is already running ahead of the Bank: 18 lenders raised fixed rates at least once in September, with CBA, NAB, ANZ and Westpac all lifting by up to 0.48pp between 15 and 22 September and Macquarie moving twice — nearly two standard hikes delivered without the cash rate moving. Fifth, demand-side data is rolling over (below). Sixth, the curve itself steepened into the meeting.
Against that, the market prices a further ~40bp beyond tomorrow and one house preview expects the statement to emphasise that "inflation risks remain tilted to the upside and that further vigilance is required." CBA's view is the opposite: 4.60% is "firmly restrictive" and the stopping point, with a Q3 trimmed mean at or above 1.0% q/q the trigger for 4.85%. Trimmed mean is running ~3.6% y/y. If the statement drops the vigilance framing or leans on Bullock's own band, that 40bp is vulnerable and the front end rallies hard. Watch the presser more than the statement. ⚠ Only two previews were obtainable, which is thinner than ideal for a decision resolving two views.
The sequencing is the trap. The monthly CPI indicator for August prints 11:30 Wednesday — twenty-one hours after the decision — with consensus +0.5% m/m and 4.1% y/y headline against a 3.5% prior, and trimmed mean +0.3% m/m. The Board writes the statement blind to it, and Wednesday's number can immediately reprice the November path. A hawkish statement paired with a soft Wednesday CPI is the cleanest front-end-receiving setup of the week. Also note the Household Spending Indicator at 11:30 Tuesday gives three hours of tradeable information before the 14:30 decision, and the Financial Stability Review lands Thursday 11:30.
Markets: a three-month low, and a curve that disagrees with the pricing
The ASX 200 fell 0.43% to 8,665.00 on Friday — its lowest close since 12 June and a fourth consecutive weekly decline, −0.76% on the week. Verified on four sources. The session shape is worth noting: it opened at Thursday's close of 8,702.00, which was also the day's high, and ground down to 8,639.90 before a small recovery. Volume was 589.9m units, roughly 19% below Thursday, because Friday was a Melbourne public holiday for the AFL Grand Final — the wrap explicitly attributes "light corporate newsflow" to it, so Friday's internals are thin and should not be over-read. Only two of eleven sectors rose: Consumer Staples +0.73% and Financials +0.27%, and financials on roughly a third of index weight is the only reason the fall was not worse. The losers were rate-sensitive and cyclical: Information Technology −1.66%, Consumer Discretionary −1.37%, Utilities −1.25%, Industrials −0.87%, Materials −0.84%. Breadth 336 advancers to 719 decliners. The A-VIX fell 1.74% to 11.53 — which ties exactly to Thursday's 11.73 — and volatility falling on a down day into a central-bank decision is complacent by any standard.
The wrap's own commentary carries more information than the index: Tech, Real Estate and Consumer Discretionary have all entered correction territory, the average ASX 200 constituent sits 22.9% below its 52-week high, the median tech name is 43% off its high, and 36% of software names are down more than 50%. Movers: Netwealth −8.42% to $16.97, a two-and-a-half-year low, and Hub24 −4.35% to $64.49 — one theme, a First Guardian class action against Netwealth subsidiaries, hitting both wealth-platform names. AP Eagers −4.43%; EQ Resources −10.5%. Lithium names fell as Chinese lithium carbonate futures dropped 4.8%, with the metal at its lowest level year-to-date. Gainers: Channel Infrastructure NZ +10.3%, Electro Optic Systems +6.4%, Healius +5.5%. ⚠ Dollar turnover was not obtainable for a fourth edition, and the SPI remains unquotable for an eleventh — the contract is rolled to December and no clean read with the contract named exists. That line should either get a terminal source or be retired.
Rates: the front end rallied into the meeting while the long end held. ACGB 2-year −7bp to 5.01%, 3-year −4bp to 5.00%, 10-year unchanged at 5.39%, with 5s and 7s a touch higher. ⭐ The three-way conflict in the 3-year that has blocked a house-view mark for three editions is resolved, and it was a date problem rather than a vendor problem: two independent routes agree at 5.00% for Friday, and the page's own prose stating a 0.04pp fall implies a 5.04% Thursday, which matches the previously-conflicting dedicated read exactly. The 5.09% figure was the stale outlier. That puts 3s10s at 39bp against a 43bp entry — markable to a point for the first time in four editions, and 4bp in the money rather than the 7–13bp the published range implied. ⚠ The 10-year's daily change is genuinely unresolved, with the vendor's own prior implying +1.7bp and the carried Thursday implying flat-to-lower; the bull-steepening conclusion holds either way. AUD/USD closed 0.7024, +0.17% on the day but −1.33% on the week — the worst G10 performer — and 70.2 US cents was independently corroborated. One house cut its year-end target from 0.73 to 0.72 and flags sub-0.700 as the risk; with the hike ~90% priced, AUD upside needs hawkish guidance, while dovish guidance puts 0.6990 in reach quickly.
The household: five months down and the transmission already tightening
Cotality's August Home Value Index fell 0.9% nationally — a fifth consecutive monthly decline — with the national median at $912,885, down 3.6% from the March peak, and 93% of capital-city suburbs recording a fall through winter. Sydney was worst at −1.4%, then Melbourne and Canberra −1.1%, Brisbane −1.0%, Adelaide and Perth −0.8%, Hobart −0.2%; Darwin was the only riser at +0.6%. The September print is expected around 1 October — a sixth consecutive fall would land two days after the hike. Saturday's auctions are consistent: recomputed from the city rows rather than taking a headline, national clearance was 523 sold of 1,076 reported = 48.6%, with Sydney 50.9%, Melbourne 52.0%, Canberra 47.3%, Adelaide 36.8% and Brisbane 36.1%. Every city row reconciled against its own stated rate this week. Two derived reads matter more than the headline: the withdrawal rate was 297 of 1,076 = 27.6%, and 35.3% in Sydney alone — vendors are pulling stock rather than meeting the market, which is a leading indicator that the decline extends. And Melbourne scheduled only 291 auctions against Sydney's 1,029, a roughly 70% collapse on the long weekend in what is normally the country's largest auction market, so treat its 52% as a thin, unrepresentative sample and not as improvement.
The consumer and business surveys are both rolling over. Westpac–Melbourne Institute consumer sentiment fell to 84.40 in September from 88.90 — a 5.9% fall — attributed to the strong July CPI raising rate-hike odds plus fuel prices; next print 5 October. NAB business confidence fell to −8 from −6, the lowest since May, with conditions at a six-year low and declines in sales, profitability and employment alongside softening forward orders. And there is a live petrol channel into Wednesday's CPI: pump prices are around $2.39/litre, with one chief economist warning unleaded could exceed $2.70 if the Iran crisis is unresolved. That is the mechanism by which the energy shock reaches the Australian inflation print — and the weekend's news made it more, not less, likely.
New Zealand
⭐ The recurring gap is closed with a proper close rather than an intraday read: the NZX 50 eased 0.10% to 13,811.11 on Friday, and rose about 0.5% on the week — a second consecutive weekly gain, making New Zealand one of the few regional markets up on the week. Decliners were Delegat Group −4.1%, Infratil −2.1% and Briscoe Group −1.8%, with weakness concentrated in financials, real estate, utilities and communication services — the same rate-sensitive complex that led Australia lower, and attributed to the global bond selloff. The RBNZ is at 2.75% after a 25bp hike on 2 September, with the next decision 28 October and October priced around 31% — both carried rather than re-verified this edition, and the date remains unconfirmed against the issuer for a second edition. NZD/USD closed 0.5658, −1.14% on the week, and the carried Thursday mark of 0.5689 was 27 pips too high — the error that corrupted last week's AUD/NZD figure. Recomputed from two legs of the same dated table, AUD/NZD closed 1.2414, which puts V023 +0.80% from its 1.2315 entry rather than the roughly-flat figure restated on Friday.
| Australia — key data trail | Latest | Prior | Next release (AEST) |
|---|---|---|---|
| Cash rate | 4.35% | 4.35% since 11 Aug | TUE 29 SEP 14:30 · presser 15:30 · then Tue 3 Nov (with a Statement on Monetary Policy) |
| Employment · full-time · part-time (Aug) | +39,000 · −6,000 · +46,000 | +20k consensus | Sep: mid-October |
| Unemployment rate (Aug) | 4.6% — a five-year high | 4.5% | ⚠ ABS flagged Labour Force Modernisation effects on August reliability |
| Participation · underemployment · hours | 67.1% (+0.2pp) · 6.2% (−0.1pp) · +1.7% y/y | 66.9% · 6.3% | Underemployment FELL and hours rose — the internals are less soft than the headline rate |
| Monthly CPI indicator (Aug ref) | 3.5% y/y (Jul) | trimmed mean ~3.6% | WED 30 SEP 11:30 · consensus +0.5% m/m, 4.1% y/y; trimmed mean +0.3% m/m |
| Household spending (Jul) | +1.1% m/m | — | Tue 29 Sep 11:30 · consensus +0.4% — three hours before the decision |
| Building approvals · private credit | −3.6% · +0.6% m/m | — | Wed 30 Sep 11:30 · consensus −0.9% · +0.5% |
| Cotality home values (Aug) | −0.9% m/m, fifth consecutive fall | −3.6% from the March peak; median $912,885 | Sep: ~1 Oct (vendor, not confirmed on an official calendar) |
| Weekend auction clearance (26 Sep) | 48.6% recomputed from city rows | withdrawal rate 27.6% | ⚠ Melbourne volumes −70% on the long weekend — treat its 52% as unrepresentative |
| Consumer sentiment · business confidence | 84.40 (−5.9%) · −8 | 88.90 · −6 | 5 Oct · 13 Oct. Conditions at a six-year low |
| Fixed mortgage repricing | 18 lenders in September; all four majors up to +0.48pp | Macquarie twice: 1yr 6.49%, 2–3yr 6.59% | Nearly two standard hikes delivered without the RBA moving |
| Iron ore · petrol | $97.06/t · ~$2.39/litre | 14th consecutive sub-$100 | China shut 1–7 Oct. One forecast has unleaded above $2.70 if Iran is unresolved |
| ACGB 3y · 10y · 3s10s | 5.00% · 5.39% · 39bp | 5.04% · 5.39% | ⭐ The 3-year conflict is resolved; V004 is markable to a point for the first time in four editions |
House views & tactical framework
One view opens, one conviction cut made on Friday is reversed, one view becomes markable to a point for the first time in four editions, and a view closed on Thursday has to be revisited because the spread retraced ten basis points back through its own stop. The book goes from six to seven.
| Asset | Bias | Conv. | Horizon | Rationale | What changes the view |
|---|---|---|---|---|---|
| Closed — revisited, because the primary has moved against the close | |||||
| US 5s30s (V003) | CLOSED — WRONG | Med | opened 7 Sep, closed 24 Sep | ⚠⚠ THIS IS THE LEAST COMFORTABLE NUMBER IN THE NOTE AND IT IS PUBLISHED RATHER THAN LEFT OUT. The view was closed WRONG on a vendor mark of 39bp, and Treasury's primary confirmed 41bp against a published 45bp stop. The official path since is 41bp (23 Sep) → 44bp (24th) → 51bp (25th). The spread has retraced ten basis points back through its own stop in two sessions and now sits 6bp below the 57bp entry rather than 16bp below it. The close was correct by its own published rules and it fired at a local extreme. The diagnosis is unchanged — a bet on fiscal term premium met a policy repricing, which flattens — but the honest addition is that a single-print stop on a spread this volatile can fire at the low, which is precisely the evidence for the ledger's 21 September convention that triggers be written as conditions rather than single prints. Not re-opened — see the declined candidate below | Closed and settled. Re-entry is explicitly declined, with the reason stated below rather than left implicit |
| Rates | |||||
| ACGB 3s10s (V004) | Flattener | Med | 1–2 mo | ⭐ MARKABLE TO A POINT FOR THE FIRST TIME IN FOUR EDITIONS — and the honest mark is less flattering than the range published on Friday. The three-way 3-year conflict resolved as a date problem: two independent routes agree at 5.00% for 25 September and the page's own prose implies a 5.04% Thursday, matching the previously-conflicting read; the 5.09% figure was the stale outlier. Against a 10-year of 5.39% that puts 3s10s at 39bp versus a 43bp entry — 4bp in the money, not the 7–13bp the 30–36bp range implied. ⚠ And the direction of travel is against the view: the curve STEEPENED about 4bp into the meeting as the front end rallied 7bp, which is the market marking the hike as near-terminal — exactly the scenario this view needs not to happen | Tomorrow's 14:30 statement is the live test and the presser matters more than the release. Guidance that frames 4.60% as terminal steepens this and should close it. Also a China stimulus impulse steepening the long end. The sourcing blocker is cleared, so there is no longer an excuse for carrying this unmarked |
| OAT–Bund (V025) | Widener | Low | 1–3 mo | Still the best view in the book, and it set a fresh 52-week wide during the week. The dedicated same-page series — now twice-validated, since its own legs reconcile to its headline and it reproduces the carried 23 September 101.7bp to the decimal — gives 105.4bp at the 25 September close against a ≈94bp entry, roughly 11bp of widening with ~25bp of room to the 80bp stop. A new 52-week high of 109.9bp was set on Thursday 24 September (52-week low 59.0bp). ⚠ A cross-page derivation gives 110.5bp and is the documented mixed-page artefact; both published, neither averaged. Conviction stays Low only because the catalyst has not happened — and it happens this week | A compression inside 80bp. Also a credible French consolidation that survives censure. The catalyst is dated: the PLF 2027 goes to the Conseil des ministres on 1 October (a 30 September variant is flagged), the Assembly deposit deadline is 6 October, and a French 10-year OAT auction lands Thursday 1 October — but per the V003 post-mortem, strong cover is evidence about clearing, not direction |
| Equities | |||||
| S&P 500 (V028) NEW | Own downside convexity, not delta | Low | 2–4 wk | NEW TODAY. Reference 7,743.41. The mechanism is mechanical rather than narrative, which is what makes the instrument honest: three sources of demand withdraw simultaneously — the corporate bid goes from 10% to 61% of index weight in blackout by 30 September and does not reopen until 1 November (now confirmed from a dated primary source); CTAs are asymmetric 21:1, able to sell $84bn in a down market against buying $4bn in an up one, with their z-score already down from +2.4 to +1.1; and vol-control sits at ~86% exposure, the highest since March. Underneath, only 49.1% of the index is above its 200-day — through 50% for the first time in the carried series — with 308 new lows against 46 new highs on a session the index rallied, and the index 0.93% from its record. The level is 7,680, where the 200-day (7,679.54) and the dealer gamma flip (7,680) coincide: a close below breaks the trend support AND re-enters negative gamma at the same price, so support and the damping mechanism fail together. This is why the expression is convexity rather than delta — and why Russell puts are preferred to S&P puts, the damage being in the tail and the Russell the only major US index lower on the week | Written as a condition, per the ledger's own rule: TWO CONSECUTIVE CLOSES ABOVE 7,800 — the dealer call wall — closes this view. A CLOSE BELOW 7,680 confirms it and conviction rises to Med. ⚠ Cost caveat that is part of the view, not a footnote: SKEW's level trend is DOWN 6.2% over two weeks, not up as this desk previously published, and equity put/call has risen from 0.47 to 0.55 — price the protection, do not assume it is cheap |
| ASX 200 (V006) | Underweight | Med | 2–4 wk | Working, and the week extended it: 8,665.00 against a 9,005.9 entry, −3.78%, with Friday the lowest close since 12 June and a fourth consecutive weekly decline. The internals remain worse than the index: breadth 336/719, Tech, Real Estate and Consumer Discretionary all in correction, the average constituent 22.9% below its 52-week high and the median tech name 43% off. The household channel keeps tightening without the Bank: 18 lenders repriced fixed rates in September and all four majors moved up to 0.48pp, house values are down five consecutive months, and Saturday's auctions cleared a recomputed 48.6% with a 27.6% withdrawal rate. ⚠ Friday was a Melbourne public holiday with volume ~19% light — do not over-read one session's internals. ⚠ And note the honest risk: an A-VIX at 11.53 falling into the decision means the market is not paying for the event | An RBA statement tomorrow that frames 4.60% as terminal; iron ore reclaiming $100 (China shut 1–7 October, so not this week); banks stabilising as a trend rather than a session. The concentration risk named in §11 item 1 stands: this and V004 both settle on the same 14:30 statement |
| FX | |||||
| AUD/NZD (V023) | Long | Low | 1–2 mo | Better than Friday's restatement, and this time the mark is built properly. Cross-computed from two legs of the same dated table with open-matches-prior-close verified — no cross quote page was fetched — the cross closed 1.2414, which is +0.80% from the 1.2315 entry. Path 1.2422 / 1.2406 / 1.2384 / 1.2414. ⚠ The carried Thursday 1.2329 was wrong purely because the NZD leg was 27 pips off; the correct Thursday cross is 1.2384. That is the second consecutive edition in which this view's mark has been restated, and both errors were in a leg rather than in the thesis. The policy-gap thesis is intact and is the cleanest trans-Tasman divergence of the cycle; the P&L has simply never been what was first claimed | An RBA hold tomorrow; a hawkish RBNZ on 28 October; a China shock hitting Australia harder than New Zealand. ⚠ Do not mark this off spec positioning this week: AUD is the single largest disagreement in the COT report — leveraged funds net long 58,726 against legacy non-commercials net short 46,814, a 105,540-contract spread. One of the two cuts is the wrong map |
| Commodities, credit and digital | |||||
| Brent (V024) | Residual call spread only | Low | 1–3 mo | ⚠⚠ THE MARK IS RESTATED AND THE CLUSTER THIS DESK REJECTED WAS THE RIGHT ONE. Thursday settles at $106.60, not the $102.83 published — a $3.77 error — and Friday at $104.32, +3.7% above the $100.60 reference. Roughly flat on the week while WTI fell 3.8%, so the premium is Gulf-located. The weekend is the whole story: Friday's −2.14% priced a Hormuz roadmap that Trump rejected on Saturday, with Houthi drones over Riyadh the same day. Physical unimproved — tanker transits in the low single digits, LNG at zero on one provider, VLCC rates at a record $1.27m/day — and the pipeline restart of 22 September has moved the exposure to the Yanbu terminus. ⭐ The "reject the high cluster" heuristic is retired, not reinforced | Rewritten as a condition, discharging an item outstanding since 24 September: two consecutive settles in the $92–95 band re-owns this outright. Also a confirmed physical restart at scale, or a Hormuz reopening Oman confirms rather than Tehran asserting. ⚠⚠ And the expression rationale has inverted: crude is net LONG 141,106 on NYMEX, not net short as this desk has been publishing, so escalation has less squeeze fuel and de-escalation carries long-liquidation risk |
| US credit (V017) | UW HY/CCC; prefer 3–5y IG | Med — RESTORED from Low | 1–3 mo | ⭐⭐ THE VIEW IS WORKING AND FRIDAY'S CONVICTION CUT IS REVERSED. All three series posted at 24 September and widened in strict beta order: IG +2bp to 79, HY +12bp to 280, CCC +37bp to 1,112 — CCC widened eighteen times as much as IG. Beta-ordered widening is credit risk repricing, not a rates pass-through. The named live test resolves against the primary market and for the view: SoftBank's $11.1bn BB+ priced through talk on 23–24 September and over exactly those two sessions the index cheapened 12bp at HY and 37bp at CCC. The issuer got a good outcome; the secondary paid for it. The view is now 38bp from the level that CONFIRMS it, having been 25bp from the level that closes it. ⚠ The literal question — did the tranches hold reoffer — is unanswerable from published sources; the index inference is derived, not reported colour | Unchanged: CCC inside 1,050bp with IG flat or tighter closes it; through 1,150bp with IG flat confirms it and conviction goes to High. The process lesson is named rather than buried: Friday's cut was made on two sessions of loud primary-market evidence while the index was silent, and the index has now spoken the other way. Cutting conviction on the loudest available evidence rather than the most relevant was the error |
The candidate declined, and it is the one this note's own themes most obviously imply. Theme 2 argues that Treasury has doubled its long-end buybacks and the long end has gone up anyway — which is a term-premium thesis wanting a curve steepener, and 5s30s has just steepened 10bp in two sessions. It is declined on the ledger's own test: can this instrument only pay if the named mechanism is what moves it, or can a different and more powerful force dominate the same price? A 5s30s steepener can be dominated by a front-end policy repricing in either direction, which is exactly what stopped V003 out at 41bp eleven sessions after it was opened — and the fact that the spread then retraced straight back through the stop is evidence about volatility, not about the thesis being right. Re-entering the same instrument days after being stopped out of it, on the same reasoning, is the single most recognisable way to convert one loss into two. The exposure this desk wants to that theme is the 8 October 30-year reopening as an observation, not a position. That is the fourth time this month a candidate has been named in advance and declined on evidence rather than taken on enthusiasm.
Portfolio-level read. Seven views, and the shape is better than last week in one respect and worse in another. Better: the book now has an equity expression (V028) built on a mechanically self-reinforcing level rather than a narrative, and the credit underweight has index confirmation rather than primary-market contradiction. Worse: the Australian concentration is unchanged and it resolves in about thirty-one hours. V006 and V004 both settle on the same 14:30 statement, and V023's short leg is the same currency — three of seven views are exposed to one press conference, and the A-VIX at 11.53 says the local market is not paying for it. That was named a week ago as more single-event concentration than this book should carry, and it has not been reduced, which is worth stating plainly rather than restating the observation as though it were an action. Elsewhere the structure holds: a fiscal-consequence trade in Europe with a dated catalyst inside a week (V025), optionality rather than delta in oil with the expression rationale now corrected (V024), and a credit underweight at restored conviction. On hedging, the SKEW correction matters more than it looks: this desk published that the level trend was up, and it is down 6.2% over two weeks. A hedge recommendation premised on steepening skew was wrong in the wrong direction. Protection should be priced, and with equity put/call up from 0.47 to 0.55 it is marginally less neglected than the carried framing claimed.
These are analytical framings for a professional reader, expressed in the vernacular of a macro desk; they are not personalised investment advice and carry no position sizing. The "what changes the view" column is the accountability mechanism — each view is logged and scored in the project's views ledger, and every close this month has fired on language written in a prior edition.
Positioning, flows & sentiment
| Indicator | Latest | Change / context | Read |
|---|---|---|---|
| CFTC — positions as of Tue 22 Sep, released Fri 25 Sep 15:30 ET. ⭐ 26 of 27 contract-sides reconstructed EXACTLY to printed open interest. Next release Fri 2 Oct | |||
| UST — leveraged funds, six CBOT contracts | −6,517,822 | +56,214 WoW (a 0.9% cover) | 2y −1,350,740 · 5y −1,858,062 · 10y −1,926,947 · ultra-10y −395,678 · bond −162,052 · ultra-bond −824,343. Composition rotated: shorts covered at the 5-year (+128,866) and the classic bond (+49,683), added at the 10-year and 2-year. Corroborated qualitatively by two banks independently — one calls CTAs "heavily short across the curve", the other flags "stretched UST shorts" |
| ⚠⚠ UST — the same book, legacy cut | −3,301,181 | +130,824 WoW | The two cuts agree in sign on all six contracts but differ by 3,216,641 contracts — the leveraged-fund short is 1.97× the legacy non-commercial short. Any sentence of the form "speculators are short X of Treasuries" is wrong by a factor of two depending on which series it silently used. Name the series, every time |
| ⭐ The library correction, settled | 6,574,036 for 15 Sep | the library's 6,863,118 is wrong by 289,082 (4.4%) | A FOURTH independent rebuild, and the first by a different arithmetic path — backing the change columns out of the new 22 September levels reproduces all six contracts and the total to the single contract. Flagged in three prior addenda and still not fixed in the library body. Fixed today in the source library |
| E-mini S&P 500 — LF net | −375,574 | −82,431 WoW | The largest single directional position in the equity complex and the biggest weekly equity-positioning move in the report. Legacy non-commercials are short only −133,228 — a 2.8× series discrepancy again. ⚠ 22 September is the first snapshot after the 18 September quarterly expiration: the spreading columns (LF −115,135) are the roll, not a view |
| ⚠ NEW CONFLICT — Nasdaq-100 | legacy +52,346 LONG · LF −35,670 SHORT | +19,777 · −22,618 WoW | The two cuts moved in OPPOSITE directions on the week. Add Nasdaq to the standing conflict list. ⚠ Publish Nasdaq figures with a ±1 contract note: the Consolidated contract's printed short side is 1 off my sum, symmetrically on both sides — a rounding residual from aggregating E-mini and Micro at different multipliers, not a transcription error |
| Russell 2000 — LF · legacy | −107,982 · −75,783 | −10,779 · −3,433 | Short and getting shorter in both cuts — consistent with the Russell being the only major US index lower on the week |
| FX majors — both cuts, and three carried conflicts | see read | ⭐ NZD RESOLVED BY THE DATA: legacy non-commercials flipped from +10,518 long to −11,380 short (−21,898, the largest proportional swing in the complex) while leveraged funds were already short and got shorter. Both cuts are now net short the kiwi. ⚠ GBP UNRESOLVED and wider — legacy −82,568 against LF +13,239. ⚠⚠ AUD UNRESOLVED and it is the single largest disagreement in the book: legacy −46,814 SHORT against leveraged funds +58,726 LONG, a 105,540-contract spread on 306,488 open interest. Agreeing in sign: EUR short both, JPY long both (legacy +71,982 / LF +7,423), CAD short both, CHF short both, MXN long both | |
| ⚠ Why the AUD conflict matters this week | RBA tomorrow 14:30 | Whichever way the statement breaks, one of the two cuts is the wrong map, so the positioning-driven follow-through is genuinely unforecastable. The mechanical explanation is visible in the raw columns — legacy bundles the leveraged-fund long with other reportables and index traders. Do not mark an AUD view off "spec positioning" this week | |
| Dollar index — legacy only | +10,330 long | −263 WoW, essentially unchanged | ⚠ Two cautions. There is no leveraged-fund cut for DXY at all — it is an ICE contract absent from the TFF page, so any "leveraged funds are long the dollar" claim is unsourceable from COT. And open interest of 46,328 is 0.7% of the 5-year note's — a thin, easily-distorted signal that should not carry weight against the majors' own cuts |
| ⭐ MSCI EAFE · MSCI EM — legacy | +30,873 · +52,171 long | +10,009 · +24,960 WoW | Developed-ex-US and EM shorts were covered aggressively into 22 September — the only clearly risk-positive positioning signal in the whole report, and it corroborates the fund survey's EM net 38% overweight |
| COMEX gold · silver · copper | +225,853 · +25,444 · +90,522 | −4,485 · +118 · +15,388 | ⭐ Gold is the most crowded long in the entire report — a net long equal to 54.7% of open interest, with non-commercial shorts of just 28,129 against commercial shorts of 320,361. It shed only 4,485 on the week, so it has not begun to unwind: the position with the least room on the other side. Copper was the week's biggest builder (+20.5% on the net) with longs added and shorts covered — one-sided, and consistent with the physical tightening in §03 |
| ⚠⚠ NYMEX WTI — THE CARRIED CLAIM WAS WRONG | +141,106 NET LONG | +5,201 (more long) | ⚠⚠ This desk has been publishing that crude is net short on both cuts. It is not. NYMEX WTI legacy non-commercials are net long 141,106 on 1,841,811 open interest. The only net-short crude contract is the far smaller ICE Futures Europe listing at −4,834 on a third of the OI — the carried claim took the first crude block on the disaggregated page and attributed it to NYMEX. The trade implication reverses: an escalation has far less short-squeeze fuel than assumed, and a de-escalation carries real long-liquidation risk. ⚠ NYMEX WTI managed-money was not retrieved |
| NYMEX gas · ULSD · RBOB | −216,530 · +10,631 · +96,038 (MM) | +5,057 · +1,678 · +12,867 | RBOB managed money is the crowded long in the energy complex and built hard on the week. Gas remains heavily net short |
| Flows — ⚠⚠ three different universes and three different weeks. Labelled individually and never netted | |||
| ⭐ BofA/EPFR Flow Show — the named gap is closed | published Fri 25 Sep | Bull & Bear Indicator 9.3 | An ACTIVE contrarian SELL signal (threshold 8.0), down from 9.7 on 28 August but still deep in sell territory. Global equities −$10.2bn of which US equities −$21.1bn — so large ex-US inflows offset. Bonds +$17.3bn, the 74th consecutive week of inflows. ⚠ Single source behind a paywalled preview; cash, gold, crypto and regional lines not retrieved. Week is the EPFR convention, to ~23 September |
| ICI — US-registered funds and ETFs | week to 16 Sep: total −$10,098m | equity −$13,297m · bond +$3,244m | Four straight weeks of domestic equity outflows (−12.1, −13.3, −4.7, −18.1bn) while bonds took in money every week. The 16 September week is the first in four where the bond bid was too small to offset it |
| ICI money market — week to 23 Sep | $7.94trn, +$15.00bn | prime +$10.95bn · government +$0.39bn | The entire week's cash build was institutional (+$15.26bn against retail −$0.26bn) and it went into PRIME, not government — a yield-reach inside cash, not a flight to safety |
| ⚠ The three universes do not agree | ICI equity −$13.3bn vs BofA global −$10.2bn | bond +$3.2bn vs +$17.3bn — a 5.3× gap | Different weeks, different universes. Report side by side with the universe named; never sum or average. ⚠ Lipper NOT RETRIEVED for a fourth consecutive edition — both carried routes are now broken (one 404, one robots-disallowed). Recommend reclassifying it as a structural sourcing failure alongside the % above the 50-day, or finding a new route |
| ⭐ Systematic — the number that matters | CTAs could sell $84bn vs buy $4bn | a 21:1 asymmetry | One house, 25 September. The single most actionable systematic-flow figure this week, and its UST-short leg independently corroborates the 6.52m-contract leveraged-fund total above. Separately: vol-control funds at ~86% equity exposure, the highest since March — little room to add, meaningful room to shed — and CTA US equity z-score down from +2.4 at end-August to +1.1 on 15 September, i.e. de-risking was already underway before this COT week |
| Spot ETF flows — BTC · ETH | +$134.5m · +$87.0m (25 Sep) | BTC −87% across the week | BTC path +999.0 → +714.7 → +346.9 → +190.7 → +134.5; ETH +270.0 → +162.2 → +104.5 → +66.1 → +87.0. Five straight inflow sessions in both, with a hard deceleration in BTC and a stabilisation in ETH. ⭐⭐ THE PLACEHOLDER RULE IS VINDICATED WITH A MEASURED MAGNITUDE: ETH's 23 September row, carried as "incomplete at $2.5m", settled at $104.5m — 42×. Publishing $2.5m would have been wrong by two orders of magnitude and supported the opposite narrative |
| Sentiment | |||
| AAII — week ending 23 Sep | Bulls 32.7% · Bears 48.1% | spread −15.4, from −24.5 | Bears have run above 44% in three of the last five weeks and bulls have not seen 40% once. Sentiment improved off the 16 September low (bears 53.3 → 48.1) but the spread is still deeply negative. ⚠ The page carries no long-run averages and none are quoted. Next print 30 September. Note the tension with the Bull & Bear at 9.3: surveys measure stated mood, B&B measures money already committed — different instruments, not a contradiction |
| BofA Fund Manager Survey | cash 3.9% · equities net +49% OW | ⚠ survey 4–10 Sep — 17 to 23 days old | ⚠ IT PREDATES THIS ENTIRE WEEK, the 18 September expiration and the 22 September COT snapshot. Say so wherever it is cited. Every carried reading re-confirmed independently: cash below the 4.0% Cash Rule threshold = SELL signal active; bonds net 48% UW, the largest since May 2022 and a 17th consecutive month below neutral; biggest tail risk a disorderly rise in bond yields at 33%, which DISPLACED AI-bubble concerns; policy-too-stimulative net 25%, the highest since September 2022; most-crowded trade global semiconductors 53%; recession split no-landing 55% / soft 38% / hard 2%; stagflation scenario 50%; EM net 38% OW, eurozone net 5% UW, UK net 35% UW, US net 25% OW, Japan net 1% UW. ⚠ Publication date is 18 September per this source, not the 15 September carried; n and AUM still unconfirmed after four attempts |
| ⭐ University of Michigan — final September | 48.1 | lowest in four months, −15% from January | Year-ahead inflation expectations 4.6%, the highest since June and up from 3.4% "before the Iran conflict." ⭐ This is the sharpest divergence in the note: a household index at 48.1 against institutional investors net 49% overweight equities and 55% expecting no landing. And the survey names the Iran conflict as the inflation driver — the same escalation channel the crude positioning correction bears on. ⚠ Single source; sub-indices not retrieved |
| Options, vol and hedging cost | |||
| ⭐ VIX · VIX3M | 14.87 · 17.93 | −0.80 on the day | ⭐ THE THREE-EDITION GAP IS CLOSED. Path 14.81 (18 Sep) → 14.87 → 14.21 → 15.18 → 15.67 → 14.87. Tie-checked on three independent points. IVTS 0.8293 — contango, day 119. Backwardation has occurred on ~7.6% of trading days since 2010 across 103 episodes averaging 3.1 days; the last was a single day on 7 April. A 119-day contango run is a deeply entrenched calm regime: carry is against anyone long vol and the curve gives no warning. ⚠ FRED's official series still stops at 22 September, four editions running — the dated table tie-checked against FRED's last value is the method that works |
| ⚠⚠ CBOE SKEW — the basis is resolved and a carried claim is WRONG | 144.9 | 54th pctile (1y) · 93rd all-time | ⭐ The three-way mess resolves: the percentile basis, quoted verbatim, is a 1-year lookback AND an all-history one stated simultaneously. The carried 144.8 / 53rd / 93rd reading was on exactly this basis and was the correct one; the 146.2/67th and 142.2/31st readings are different observation dates whose 1-year percentile moved as the trailing window rolled — the all-time figure barely moves (93/94/91) while the 1-year swings 31st to 67th, the signature of a fast-rolling window. ⚠⚠ AND THE CARRIED TREND CLAIM IS WRONG: this desk published that the LEVEL trend is up. It is DOWN, monotonically: 154.5 → 152.1 → 146.6 → 144.9, −6.2% over two weeks. Skew is FLATTENING. A hedge recommendation premised on rising skew is wrong in the wrong direction |
| Put/call — 24 Sep close | equity 0.55 · index 0.87 · total 0.80 | 14th percentile | ⚠ Update to the carried read: equity put/call has RISEN from 0.47 to 0.55, +17%, so some protection buying returned. The "almost nobody is buying it" framing overstates the case. ⚠⚠ And a percentile caveat that materially weakens the figure: the page's free history begins in 2025, so "14th percentile" is computed over roughly 20 months, not a market cycle — it is not comparable to the all-time bases used for SKEW and should not sit beside them as though it were. No 25 September reading was posted |
| ⭐⭐ Dealer gamma — a regime inversion in one session | flip 7,680 · net GEX +$32.56bn | from flip 7,709 / −$20.26bn | Same vendor on both days, so these are a legitimate series and are chained deliberately. Net GEX swung +$52.82bn, the flip level FELL 29 points while spot rose 75, and the index went from 40 points below the flip to 64 above it — a 104-point traverse. Mechanism: post-expiration gamma reset plus a rally through a falling flip. Positive gamma means dealers fade pushes and buy dips — moves DAMPED, with a call wall at 7,800, put wall 7,700, max pain 7,710. ⚠ Do NOT chain to the other vendor, whose last published flip remains a 10 September 7,600; nothing newer exists there |
| ⚠ Citadel/Rubner denominators — confirmed, and one revised | 18 Sep expiry ~$7trn / ~25% | superseding $6.2trn / 23% | Confirmed from the primary rather than carried: $9.6trn/~35% is the CUMULATIVE through 18 September and $6.2trn/23% was the single-day estimate — then Rubner revised the single-day figure UP to ~$7trn / ~25%, second largest on record, as open interest built. Cite ~$7trn as the figure of record and present $6.2trn as superseded, not as a contradiction. ⚠ Both notes are now historical — the expiration has passed and no October estimate was obtainable |
| Breadth, technicals, valuation | |||
| ⭐ % above the 200-day | 49.10% | from 52.9%, through 50% | Fewer than half the index is in a 200-day uptrend while the index sits within 1% of its record. Reference points on the same page: 62.10% at end-2025 and 60.30% at end-2024, so participation is 11 to 13 points below each of the last two year-ends. Per the standing instruction the % above the 50-day is not carried and not reported as a gap — that is a closed structural sourcing failure |
| New highs / lows · McClellan · Hindenburg | 46 / 308 · −22.59 · 3 of 4 | re-fetched, not carried | Lows outnumber highs 6.7:1 on a session the index rallied 40 points to within 0.93% of its record. Raw new lows are down 18% from the 23rd but new highs FELL from 56 to 46 and the ratio is flat. ⚠⚠ The condition that FAILS is the binding one — new highs at 0.97% are far below the 2.2% threshold — so a Hindenburg trigger is not currently possible. Report "3 of 4" only with that qualifier: this is deteriorating internals, not an imminent signal. Cluster inactive (1 trigger in 30 days, last 17 August; 19 in 2026). ⚠ Universe 4,733 US common stocks, not NYSE-only, and the eligible count itself drifts |
| ⚠ S&P technicals — two problems with the source | 200d 7,679.54 · 50d 7,699.99 · RSI 59.1 | stamped after Friday's close | ⚠ PROBLEM 1: the pivot block is NOT a daily timeframe and is not published as one. Its R1−S1 of 10.50 points implies an impossible daily range on the S&P; it is computed on a short intraday interval. The carried 7,712.63 "daily pivot" and this fetch's 7,738.95 are not comparable and neither is a daily pivot. ⚠ PROBLEM 2: a 50-day SMA cannot move 32.87 points in one session, so the carried 7,667.12 and the current 7,699.99 are not the same series. Independent corroboration via SPY's 50-day at $761.57 favours the current fetch. The "overbought fully unwound" narrative built on RSI 47.3 is stale either way — RSI is back to 59.1, mid-range. Also reported as printed: the 100-day (7,666.87) sits BELOW the 200-day |
| ⭐ FactSet Earnings Insight — the named gap is closed | Q3 est. +29.1% · fwd P/E 19.2× | revised up from +28.9% | 25 September edition confirmed live via the 302 redirect. 9 companies reported (7 EPS beats); positive guidance 72 of 116 = 62% against a 41% five-year average — a 21-point beat on a real sample rather than the 9 reporters. CY2026 +32.0%, CY2027 +15.4%; bottom-up target 9,275.04. ⚠⚠ A CARRIED FRAMING NEEDS QUALIFYING: at 19.2× the index is below the 5-year average of 19.8× but ABOVE the 10-year average of 19.0×. Quoting only the 5-year — which is contaminated by the 2021 and 2024–25 multiple peaks — makes the index look cheaper than it is |
| ⭐ Buyback blackout — CONFIRMED, not carried | 10% → 61% by 30 Sep | reopens 1 November | ⭐ All three legs now come from one dated, attributed primary source — the gap is closed. The mechanical corporate bid withdraws from 10% to 61% of index weight over the next three days and does not return for five weeks. Overlay the 21:1 CTA asymmetry and vol-control at 86%: the three largest sources of mechanical demand are all either withdrawing or out of room, simultaneously |
| ⚠⚠ Sell-side targets — the dateline trap fired again | HSBC 8,100 (13 Sep) | the only fresh, fully attributable target | ⚠⚠ THREE TARGETS WERE CAUGHT AND QUARANTINED: a high-ranking article carrying BofA 7,100, Citi 8,100 and US Bank 8,040 is dated 9 JUNE 2026. None is published as current. And it creates a live conflict that is disclosed rather than resolved: the carried BofA figure is 7,400 year-end / 7,800 12-month, vintage 14 September and unverified today, against this article's 7,100 from 9 June. Neither is presented as confirmed-current. The one clean target is HSBC's 8,100, raised from 7,650, published 13 September, on CY2026 EPS $360 and a 22.5× multiple — ⚠ which is 3.3 turns above FactSet's actual 19.2×, so the call requires multiple expansion, not just the earnings it cites |
The organising fact, and it makes six separate observations one story. A Fed that has already hiked to 3.75–4.00% with a settled 66.6% chance of hiking again in October sits alongside: leveraged funds short 6.52m Treasury contracts with CTAs "heavily short across the curve"; a fund survey where net 25% say policy is too stimulative, the highest since September 2022, the biggest tail risk is a disorderly rise in bond yields at 33% having displaced AI-bubble concerns, and bonds are net 48% underweight, the most since May 2022; and a household sector at 48.1 with one-year inflation expectations of 4.6%. That is a stagflation-hedging book, which is precisely what the survey's own 50% stagflation scenario says. The equity complex — net 49% overweight, a Bull & Bear sell signal at 9.3, vol-control at 86% — is positioned as though the other 50% is certain. The reflexivity is the risk rather than the level: a short of 6.5m contracts means a yield spike is self-reinforcing until it forces a cover, and then violently the other way. Positioning extremes are flagged by magnitude and by ratio, never by percentile: no historical percentile was sourceable for the Treasury or E-mini shorts, so none is published, and gold's 54.7%-of-open-interest net long is stated as a ratio rather than dressed up as an extreme reading.
The week ahead
| Day | AEST | ET | Event | Cons. | Prior | Imp. |
|---|---|---|---|---|---|---|
| Monday 28 September — Taiwan closed · Korea reopens after Chuseok · mainland China reopens for three sessions | ||||||
| Mon | 20:00 | 06:00 | UK — BoE RAMSDEN SPEECH ON QUANTITATIVE TIGHTENING, Bank of England (11:00 London) | H | ||
| Mon | 11:30 · day | Sun 21:30 | CN — industrial profits YTD y/y · PBoC fix, the first since the holiday (~11:15 AEST) | +18.0% | +17.6% | M |
| Mon | 22:15 · 23:30 | 08:15 · 09:30 | US — Fed Bowman · EA — ECB Lagarde · JP — BoJ Minutes | M | ||
| Mon | Tue 03:25 · 03:30 | 13:25 · 13:30 | US — Fed Cook · Fed Barkin · UST 13-week and 26-week bill auctions | L | ||
| Tuesday 29 September — THE MAIN EVENT FOR THIS READER | ||||||
| Tue | 11:30 | Mon 21:30 | AU — HOUSEHOLD SPENDING INDICATOR (Aug) — three hours of tradeable information before the decision | +0.4% | +1.1% | M |
| Tue | 14:30 | 00:30 | ⭐⭐ RBA CASH RATE DECISION + RATE STATEMENT — issuer-confirmed | 4.60% (+25bp), 88–90% priced | 4.35% | H |
| Tue | 15:30 | 01:30 | ⭐⭐ RBA PRESS CONFERENCE — watch this more than the statement. The 4.5–5% band question is a presser question, and it resolves V006 and V004 | H | ||
| Tue | 17:00 | 03:00 | ES — flash CPI y/y (the first euro-area CPI leg) · CH — KOF barometer | 4.7% · 106.0 | 4.3% · 106.7 | M |
| Tue | Tentative | — | IT 10y auction (prior 4.10%, 1.6×) · UK 10y GILT auction (prior 5.16%, 3.6×) — the gilt auction lands the same week as the QT speech | M | ||
| Tue | 18:30 · 20:00 · 21:00 | 04:30 · 06:00 · 07:00 | UK — M4, mortgage approvals, net lending · DE — Bundesbank Nagel · EA — ECB Lagarde | 57K · £6.2bn | 56K · £6.3bn | L |
| Tue | 22:30 · 23:00 | 08:30 · 09:00 | CA — GDP m/m · US — FHFA HPI, S&P/CS 20-city y/y | +0.1% · 2.2% | +0.3% · 2.1% | M |
| Tue | Wed 00:00 | 10:00 | US — CB CONSUMER CONFIDENCE · JOLTS OPENINGS — the confidence print is the test of §09's household-vs-institution gap | 90.1 · 7.23M | 89.4 · 7.27M | M |
| Tue | Wed 01:30 | 11:30 | ⭐ UK — MPC ALAN TAYLOR, NIESR Dow Lecture: "Expectations and the risk of inflation persistence" — the committee's most dovish dissenter, three days after Bailey turned hawkish | H | ||
| Tue | Wed 01:00 · 03:00 · 04:00 | 11:00 · 13:00 · 14:00 | UK — MPC Mann · US — Fed Goolsbee · Fed WILLIAMS · also Barr, Musalem, BoC Gravelle | M | ||
| Tue | Wed 05:00 | 15:00 | US — Fed WALLER · Micron fiscal Q4 (consensus ~$50.8bn revenue, $31.45 EPS) — the test of the most-crowded trade | H | ||
| Tue | Wed 09:50 | 19:50 | JP — preliminary industrial production m/m · retail sales y/y | +1.2% · +3.2% | −0.2% · 3.7% | M |
| Wednesday 30 September — the US print of the month, the Australian CPI, and China's last session before Golden Week | ||||||
| Wed | 11:30 | Tue 21:30 | ⭐⭐ AU — MONTHLY CPI INDICATOR (Aug ref) — issuer-confirmed. Lands 21 hours AFTER the RBA decision, so the statement is written blind to it | +0.5% m/m · 4.1% y/y; trimmed mean +0.3% m/m | 1.0% · 3.5% · 0.5% | H |
| Wed | 11:30 | Tue 21:30 | AU — building approvals m/m · private sector credit m/m · RBA Chart Pack | −0.9% · +0.5% | −3.6% · 0.6% | M |
| Wed | 11:30 | Tue 21:30 | ⭐ CN — OFFICIAL NBS MANUFACTURING PMI · non-manufacturing PMI · balance of trade — the LAST China read before a seven-day closure | 50.1 · 49.3 · $79bn | 49.8 · 49.0 · $119.1bn | H |
| Wed | All day | — | ⭐ DE — PRELIMINARY SEPTEMBER CPI m/m — the largest single euro-area CPI leg · FR prelim CPI 16:45 · IT prelim CPI 19:00 | +0.5% · −0.5% · 0.0% | +0.2% · 0.7% · 0.5% | H |
| Wed | 16:00 · 19:30 | 02:00 · 05:30 | UK — current account, final Q2 GDP, business investment · UK — FPC STATEMENT AND MINUTES | −£25.6bn · +0.4% | −£22.1bn · 0.4% | M |
| Wed | Tentative | — | DE 10y BUND auction (prior 3.39%, 1.5×) — into a Bund at its highest since June 2009 | M | ||
| Wed | 22:15 | 08:15 | US — ADP employment change | +70K | +38K | M |
| Wed | 22:30 | 08:30 | ⭐⭐ US — CORE PCE PRICE INDEX m/m — THE MOST IMPORTANT US PRINT OF THE MONTH | +0.3%; ~3.2% y/y core, ~3.6% headline | +0.2% | H |
| Wed | 22:30 | 08:30 | ⭐ US — PERSONAL SPENDING m/m · personal income · goods trade balance — the REAL second event of the minute | +1.0% · +0.5% · −$113.8bn | +0.2% · 0.4% · −$118.9bn | H |
| Wed | 22:30 | 08:30 | US — FINAL Q2 GDP q/q and GDP price index · BEA annual revisions · wholesale inventories | +1.5% · +6.4% | 1.5% · 6.4% | M |
| Wed | 23:45 · Thu 00:30 | 09:45 · 10:30 | US — Chicago PMI · crude oil inventories (EIA, week to 25 Sep) · Fed Barkin, Cook, Goolsbee, Kashkari | 51.3 | 47.1 · +3.0M | M |
| Wed | All day | — | Quarter end · the buyback blackout peaks at 61% of index weight · the 2y, 5y and 7y notes all SETTLE — a month-end cash drain the same day as core PCE, with no coupon supply until 6 October · Canada closed | H | ||
| Thursday 1 October — the French catalyst, the Tankan, and China goes dark for a week | ||||||
| Thu | ~day | — | ⭐⭐ FRANCE — PLF 2027 TO THE CONSEIL DES MINISTRES. €54bn consolidation, a 5.0% deficit target against a −5.4% no-measures baseline and 120% debt. THE LIVE V025 CATALYST. ⚠ Date conflict: 1 Oct (freshest source) against 30 Sep — not asserted | H | ||
| Thu | Tentative | — | ⭐ FR 10y OAT AUCTION (prior 4.23%, 2.3×) — the cleanest scheduled read on French issuance appetite in the PLF week · ES 10y auction | H | ||
| Thu | 09:50 | Wed 19:50 | ⭐ JP — BoJ SUMMARY OF OPINIONS (the 17–18 Sep meeting, 7–2) · Q3 TANKAN large manufacturing / non-manufacturing. The Summary is what confirms or breaks the new front-end pricing | 25–26 / 36 | 22 / 37 | H |
| Thu | 11:30 | Wed 21:30 | AU — RBA FINANCIAL STABILITY REVIEW — issuer-confirmed · AU goods trade balance | A$2.10bn | 1.92bn | M |
| Thu | ~day | — | AU — Cotality Home Value Index (September). ⚠ A private vendor, absent from official calendars — date unconfirmed. A sixth consecutive fall would land two days after the hike | −0.9% m/m | M | |
| Thu | 18:00 | 04:00 | ⭐ UK — BoE GOVERNOR BAILEY, LSE/BoE "Future of Money" conference — his first remarks since Friday's hawkish Oxford speech | H | ||
| Thu | 16:30 · 17:15–18:00 | 02:30 · 03:15–04:00 | CH — CPI m/m · EA — FINAL MANUFACTURING PMIs (euro area 52.7, Germany 53.8, France 50.3, Italy 49.9, Spain 50.2) | 0.0% | 0.4% | M |
| Thu | 22:30 · 23:05 | 08:30 · 09:05 | US — initial claims · Fed Barkin, Collins, Schmid · Challenger job cuts 19:30 | 199K | 197K | M |
| Thu | Fri 00:00 | 10:00 | ⭐ US — ISM MANUFACTURING PMI · ISM PRICES PAID — prices paid is the tariff-plus-energy pass-through read and deserves as much attention as the headline · Fed WALLER | 55.0 · 72.0 | 54.6 · 71.1 | H |
| Thu | All day | — | ⭐ CN — GOLDEN WEEK BEGINS, mainland shut 1–7 OCTOBER · US Treasury: 4-week and 8-week bills; 3y/10y/30y announcements | M | ||
| Thu | Fri 09:30 | 19:30 | ⭐ JP — TOKYO CORE CPI y/y — a 50bp jump in consensus and the cleanest leading indicator for the 30 October BoJ · unemployment rate | 2.4% · 2.4% | 1.8% · 2.4% | H |
| Friday 2 October — payrolls and euro-area CPI, four hours apart | ||||||
| Fri | 19:00 | 05:00 | ⭐⭐ EURO-AREA FLASH CPI — headline y/y · core y/y. THE MOST IMPORTANT EUROPEAN PRINT OF THE WEEK | 3.7% · 2.5% | 3.2% · 2.4% | H |
| Fri | 22:30 | 08:30 | ⭐⭐ US SEPTEMBER PAYROLLS — issuer-confirmed. Non-farm payrolls · unemployment rate · average hourly earnings m/m | +98K · 4.1% · +0.3% | +162K · 4.1% · +0.3% | H |
| Fri | Sat 00:00 · day | 10:00 · 15:30 | US — factory orders m/m · Fed Logan · next CFTC COT release 15:30 ET · China shut | −0.1% | +0.9% | M |
| The sessions after — ⚠ Sydney moves to AEDT (UTC+11) on SUNDAY 4 OCTOBER, so every entry below is ET+15. US DST ends 1 November; from 2 November it is ET+16 | ||||||
| Tue 6 Oct | — | — | ⭐⭐ FRANCE — STATUTORY DEADLINE to deposit the PLF 2027 with the National Assembly. THE CENSURE CLOCK STARTS. LFI, the Greens and the PCF are committed to a censure motion on reading, the PS is leaning that way, and the RN's 122 votes are decisive — its published price is no new taxes. Also UST 3-year auction | H | ||
| Wed 7 · Thu 8 Oct | — | — | UST 10-YEAR auction · ⭐⭐ UST 30-YEAR auction — the term-premium test, into a 30-year at a new cycle high with Treasury's doubled buybacks visibly failing | H | ||
| Tue 13 Oct · mid-Oct | 11:30 | — | AU — RBA minutes · US banks start Q3 earnings 13 Oct · US September CPI — ⚠ DATE NOT CONFIRMED, a top-tier event with no date in this note | H | ||
| Sat 17 Oct | — | — | Fed blackout begins ahead of the October meeting | M | ||
| WED 28 OCT | — | — | ⭐⭐ A FOUR-WAY CLUSTER: FOMC DECISION (no SEP) · BANK OF CANADA · RBNZ · THE UK BUDGET, carrying the OBR's first official headroom reassessment since November 2025 | Fed hike 66.6% | 3.75–4.00% | H |
| Thu 29 · Fri 30 Oct | — | — | ECB DECISION (~48% priced for October, skewed to December) · BANK OF JAPAN DECISION — the Bank's own schedule confirms the statement is released Friday 30 October; third-party calendars quoting the 29th are giving the meeting start | H | ||
| TUE 3 NOV | 14:30 AEDT | — | ⚠ RBA DECISION AND STATEMENT ON MONETARY POLICY, presser 15:30 — the RBA's next meeting is 3 NOVEMBER, not 28 October. This was omitted from a prior forward list and is corrected here | H | ||
| Wed 4 · Thu 5 Nov | — | — | Riksbank (a 2026 hike now in its baseline) · Norges Bank · BANK OF ENGLAND WITH A MONETARY POLICY REPORT (~81% priced) · Treasury's quarterly refunding — the next buyback sizing decision · US October payrolls Fri 6 Nov | H | ||
| 27 Nov · 11 Dec | — | — | German final 2027 budget vote (carried, unverified) · US government funding expires 11 December — the next real fiscal cliff, with the debt ceiling not binding until 2027 | H | ||
| 8–18 Dec | — | — | RBA (8th) · RBNZ, BoC and FOMC WITH A SEP (9th) · SNB (10th) · Riksbank (16th) · Norges, ECB, BoE (17th) · BoJ (18th) | H | ||
| 1 · 10 JAN 2027 | — | — | ⚠⚠ A GENUINE CALENDAR COLLISION: US DFARS magnet restrictions enter a new phase on 1 January, NINE DAYS BEFORE the US–China truce expires on 10 January — and the extension agreed nothing on rare earths | H | ||
⚠ Every H/M/L rating in this table is this desk's own judgement and is presented as such: the calendar aggregator's impact column failed to render for a third logged time, returning no impact data at all. The event list itself rendered completely for all five days and was cross-checked against the RBA, BoJ, Fed, BoE, BLS, ABS, Census, Treasury and BoE-events issuer pages without a single conflict; consensus and prior figures are the aggregator's unless an issuer is named, and are unverified on that basis. AEST = ET + 14 until Saturday 3 October. Sydney moves to AEDT (UTC+11) on Sunday 4 October, so from then AEDT = ET + 15; US DST ends 1 November, so from 2 November AEDT = ET + 16. A prior edition's framing of "core PCE alongside Q2 GDP" is corrected here: that GDP is the final estimate of a quarter that ended three months ago with consensus unchanged, and the genuinely high-information companion in the same minute is personal spending at a +1.0% consensus against a +0.2% prior.
Risk radar
| # | Risk | Trigger / timing | Probability | Hedge / expression |
|---|---|---|---|---|
| 1 | Crude gaps on a weekend premise that no longer exists | Sydney open today; Iran talks "as early as Monday" via Qatari mediators | — | Friday's −2.14% priced a roadmap Trump rejected on Saturday, with Houthi drones over Riyadh the same day. ⚠⚠ And the correction in §01 item 2 reverses the asymmetry: NYMEX WTI non-commercials are net LONG 141,106 and adding, so escalation has far less squeeze fuel than this note assumed and de-escalation carries long-liquidation risk. Own optionality, not futures — and size it for a two-sided gap rather than a one-way one |
| 2 | The RBA statement resolves two house views at the same instant, and a third is the same currency | Tue 29 Sep 14:30 AEST, presser 15:30 | Hike 88–90% | V006 and V004 both settle on the guidance and V023's short leg is the AUD — three of seven views on one press conference. Named a week ago as more concentration than this book should carry and not reduced since, which is worth stating rather than restating. The hike is not the risk; "terminal" versus "first of several" is. An A-VIX at 11.53, falling into the event, says the local market is not paying for it |
| 3 | Core PCE and personal spending land in the same minute as a month-end cash drain | Wed 30 Sep 22:30 AEST | Oct hike 66.6% · Dec two hikes 51.6% | Consensus core +0.3% m/m (~3.2% y/y) and personal spending +1.0% against a +0.2% prior — that pair is a nominal-demand signal the front end has to respect. The 2y, 5y and 7y all settle the same day and there is no coupon supply to absorb a surprise until 6 October. An upside print pushes December's two-hike path through 60%. Front-end payers; flat duration remains a position |
| 4 | The long end keeps rising while Treasury buys it — the tool visibly failing | Ongoing; the 8 Oct 30-year reopening, then the 4 Nov refunding sizing decision | — | Doubled buybacks ($2bn → $4bn+) have run since 9 September and the 30-year printed a new cycle high of 5.49% with the 20-year at 5.54%. Last week's 7-year auction is the warning: headline "in line", but indirects 57.2% against a 64.6% average. Either sizing escalates materially or the market concludes term premium cannot be bought down. Own convexity, not a curve position — see §08 for why the obvious steepener is declined |
| 5 | Three mechanical bids withdraw simultaneously into rotting internals | Blackout peaks at 61% of index weight on 30 Sep; no reopening until 1 Nov | CTA asymmetry 21:1 | CTAs could sell $84bn in a down market against buying $4bn in an up one; vol-control at ~86% exposure, highest since March; only 49.1% of the index above its 200-day, 308 new lows to 46 new highs. The level is 7,680, where the 200-day and the gamma flip coincide — a close below breaks trend support AND re-enters negative gamma at the same price. This is V028. Prefer Russell puts to S&P puts |
| 6 | Credit's beta-ordered widening extends | Any observation after 24 Sep | CCC 1,112bp · 38bp from V017's confirm | IG +2bp, HY +12bp, CCC +37bp — CCC widened 18× IG. The widening is accelerating (+2, +5, +7, +19bp on successive days) with high beta leading. This is the view working, not the risk to it — the risk is that it extends far enough to matter for equity multiples, which is the channel §09's stagflation-hedging book is not positioned for |
| 7 | The French budget arrives with a censure threat and a decisive far-right bloc | Conseil des ministres ~1 Oct; Assembly deposit deadline 6 Oct; OAT auction Thu 1 Oct | — | €54bn sought against a target that buys only ~0.4pp of deficit. LFI, the Greens and the PCF are committed to censure on reading, the PS is leaning that way, and the RN's 122 votes are decisive with a published price: no new taxes. Both branches are OAT-negative — satisfy the RN and €54bn has to come from spending alone into a censure-committed left; raise taxes and the budget falls. V025 is the expression and its conviction stays Low only until this happens |
| 8 | Japan's quarterly-hike path is confirmed or broken by the Bank's own record | BoJ Summary of Opinions, 1 Oct JST; Tokyo CPI the same day | 2y embeds ~69bp ≈ 2.75 hikes | An official signalled hikes roughly every three months to 2% by June 2027 and Friday's bear flattener priced it. The Summary covers the 7–2 September meeting and will show whether the dissents were dovish or hawkish. Tokyo CPI consensus jumps 1.9% to 2.3% on energy. ⚠ And 160 in USD/JPY is a policy line, not a technical one — the Finance Minister invoked the post-intervention principles on Friday with the pair having traded 158.97 |
| 9 | China's PMI prints on the last session before the lights go out for a week | Wed 30 Sep 11:30 AEST; mainland shut 1–7 Oct | Consensus 50.1 vs 49.8 | The market expects the factory sector to cross back above 50 for the first time — and a miss leaves the mainland shut for seven days with no way to trade it, which makes Hong Kong the only expression vehicle. That asymmetry is the tradeable structure of the week. Expect position-squaring into Wednesday's close and gaps rather than drifts on any holiday headline |
| 10 | The household-versus-institution gap is the widest in the note | CB consumer confidence Wed 00:00 AEST; Challenger job cuts Thu | Michigan 48.1 · 1y inflation exp. 4.6% | A consumer index at a four-month low, 15% below January, with inflation expectations up from 3.4% "before the Iran conflict" — against institutions net 49% overweight equities and 55% expecting no landing. Germany's −30.6 consumer climate says the same thing on the other side of the Atlantic, and it is an income-expectations shock rather than a growth scare. This is the demand leg of the regime tag finally showing up in hard data |
| 11 | Payrolls, where both tails are adverse for a net-long book | Fri 2 Oct 22:30 AEST | Consensus +98K | A weak print against a hiking Fed is the stagflation confirmation that 50% of surveyed managers already expect; a hot print compounds risk 3. Both tails are unfriendly to a net 49% overweight, and the reflexivity is in the positioning: 6.52m contracts of leveraged-fund Treasury shorts means a yield move is self-reinforcing until it forces a cover |
| 12 | The crowded semiconductor trade meets an earnings test | Micron fiscal Q4, Wed 05:00 AEST | Most-crowded trade 53% | Consensus ~$50.8bn revenue and $31.45 EPS. Context cuts both ways: semi one-month implied vol has already fallen from 77.2 to 46.0, leveraged-ETF AUM is down about 50% from the June peak and retail semi notional 46% below it — the leverage has already left, which limits the unwind and also the support |
| 13 | Gold is the most crowded long in the entire COT report | No catalyst identified | Net long = 54.7% of open interest | +225,853 net, trimmed only 4,485 on the week, with non-commercial shorts of just 28,129 against commercial shorts of 320,361. Flagged as a ratio, not a percentile — none was sourceable. Meanwhile spot has now closed below $4,300 three sessions running and failed at $4,316.60 intraday. The position with the least room on the other side |
| 14 | Governor Cook and Fed independence | Response window closed ~28 August; status unknown | — | The Supreme Court ruled 5–4 in her favour on 29 June, requiring notice and an opportunity to respond before removal; the White House then began exactly that process on 7 August. She continues to serve and vote, and what happened after the deadline could not be established. This is the one channel through which a hawkish Fed becomes dollar-negative, and it is a standing daily check rather than a resolved item |
Key levels
| Instrument | Last | Support | Resistance | Comment |
|---|---|---|---|---|
| S&P 500 | 7,743.41 | 7,710 (max pain) · 7,700 (put wall / 50d) · 7,680 (gamma flip / 200d) | 7,800 (call wall) · 7,816.70 (record) | ⭐ A genuine four-way confluence at 7,680–7,710, and it is why V028 exists: the put wall (7,700), the 50-day (7,699.99), the gamma flip (7,680) and the 200-day (7,679.54) stack into a 20-point band. Above it, positive dealer gamma damps moves and the 7,800 call wall caps. A close below 7,680 breaks the 200-day AND re-enters negative gamma simultaneously — the two supports fail together and moves go from damped to amplified. RSI 59.1, mid-range. 0.93% from the record on 49.1% participation |
| Nasdaq Comp · Russell 2000 | 27,068.72 · 2,837.55 | 26,500 · 2,800 | 27,244 · 2,890 | The Composite was the week's best US index (+2.06%); the Russell was the only one lower (−0.80%) and is where the breadth damage sits — 308 new lows against 46 highs. Prefer Russell expressions |
| UST 2y · 5y | 4.81 · 4.98% | 4.75 · 4.90 | 4.87 (24 Sep high) · 5.00 | Official par series. The 5-year has still never closed above 5% — the break remains intraday-only across the whole move |
| UST 10y · 30y | 5.17 · 5.49% | 5.05 · 5.40 | 5.18 (2007 high) · 5.49 (cycle high, made Friday) | The ten-year's high was Thursday's 5.18%, the highest since July 2007. Friday's new high was the LONG BOND, and the 20-year at 5.54% likewise — on a session the rest of the curve rallied |
| 5s30s · 2s10s | 51 · 36bp | 44 · 31 | 57 (V003 entry) · 40 | ⚠⚠ Official path 48 → 46 → 46 → 41 → 44 → 51. V003 was stopped at 41bp and the spread has retraced ten basis points back through its 45bp stop in two sessions. Disclosed in full in §08 |
| JGB 2y · 10y | 1.94 · 3.07% | 1.85 · 2.99 | 2.00 · 3.11 (intraday, 30-yr high) | The 2-year is the instrument to watch, not the 10-year — it embeds ~69bp over policy, about 2.75 hikes. Friday was a bear flattener with the long end bid |
| Bund 10y · OAT–Bund | 3.621% · 105.4bp | 3.50 · 80 (V025 closes) | 3.70 · 109.9 (52-wk wide, 24 Sep) | Bunds finally joined the rout at their highest since June 2009. OAT–Bund set a fresh 52-week wide on Thursday; ~25bp of room to the stop. BTP–Bund 92.7bp — Italy trades 17.7bp inside France |
| Gilt 10y · 30y | 5.3593 · 5.85% | 5.30 · 5.75 | 5.45 · 5.95 | ⭐ Every tenor live-stamped, resolving last week's wholesale rejection. The G10 outperformer — essentially unchanged across two sessions of a global rout, into a Governor who has just turned hawkish |
| ACGB 3y · 10y · 3s10s | 5.00 · 5.39% · 39bp | 4.90 · 5.30 · 35 | 5.10 · 5.50 · 43 (V004 entry) | ⭐ Markable to a point for the first time in four editions. 39bp against a 43bp entry — 4bp in the money, and the curve STEEPENED into the meeting, which is the market marking the hike as near-terminal. The 3y embeds +65bp over a 4.35% cash rate |
| DXY · EUR/USD | 100.97 · 1.1391 | 100.6 · 1.1350 | 101.40 (Thu high) · 1.1450 | The dollar-reconciliation test passes on the week to 2bp. Note the Friday divergence: US 10y roughly flat while the dollar fell 0.31% and the yen gained 0.99% — rising US yields with a falling dollar is not a carry story |
| USD/JPY | 157.28 | 156.94 (Fri low) · 155.00 | 158.97 (Fri high) · 160 (policy line) | A 203-pip reversal off 158.97. 160 is behaving as a policy level, not a technical one — the Finance Minister invoked the post-coordinated-intervention principles on Friday |
| AUD/USD · AUD/NZD | 0.7024 · 1.2414 | 0.7012 (Thu low) · 0.6990 · 1.2384 | 0.7115 · 1.2450 | The week's worst G10 at −1.33%. With the hike ~90% priced, upside needs hawkish guidance; dovish guidance puts 0.6990 in reach quickly. V023 is +0.80% from entry |
| Brent · WTI | $104.32 · $92.41 | 100 · 92–95 (V024 re-own zone) | 106.60 (Thu) · 96 | ⚠ Thursday's Brent restates to $106.60 from the $102.83 published. WTI is already inside the re-own band, now written as a condition: two consecutive settles in $92–95. Brent–WTI ≈$11.91 — the premium is Gulf-located |
| Gold · Silver | $4,285 · $64.26 | 4,254 (Fri low) · 4,200 | 4,300 (V014 stop) · 4,316.60 | A THIRD consecutive close below the stop — and it traded up to $4,316.60 intraday and failed there, which is a worse signal than a clean close below. Silver remains the week's worst major metal |
| Copper · Iron ore | $14,740/t · $97.06/t | 14,500 · 95 | 14,900 · 100 | Cash−3M backwardation +$93/t, having widened 6.6× in five sessions while LME stocks fell 3,600t — but eight sessions and five direction changes, so do not extrapolate. Iron ore's fourteenth sub-$100 print, in a four-session $97.06–97.30 band |
| Bitcoin · Ether | ≈$84,728 · $2,692 | 83,000 · 2,600 | 86,200 · 2,800 | Flat price against ETF inflows down 87% across the week and futures open interest $1.2bn below Thursday — a market held up by fewer marginal buyers. Funding mildly positive, no crowded-long stress |
| ASX 200 | 8,665.00 | 8,639.90 (Fri low) · 8,600 | 8,702 · 9,005.9 (V006 entry) | Lowest close since 12 June; fourth straight weekly fall; −3.78% from the V006 entry. A-VIX 11.53 and falling into the decision. Breadth 336/719 |
| Nikkei 225 · TOPIX | 66,364.20 · 4,128.59 | 65,639 (Fri low) · 65,000 | 66,410.27 (Fri high) | The best major index on the week at +2.07%, and broad — TOPIX +1.31% against Nikkei +1.30%. ⚠ Whether Friday's high was a record could not be verified and is not asserted |
| IG / HY / CCC OAS | 79 / 280 / 1,112bp | CCC 1,050 (V017 closes) | CCC 1,150 (confirms) | 38bp from the level that confirms the view, having been 25bp from the level that closes it. HY at 280bp still sits in the richest decile against a long-run median near 450bp |
| VIX · SKEW | 14.87 · 144.9 | 14.00 · 140 | 16.57 (24 Sep high) · 150 | ⭐ Gap closed, tie-checked on three points. IVTS 0.8293, contango day 119. ⚠⚠ SKEW's level trend is DOWN 6.2% over two weeks, not up as previously published — 54th percentile on a 1-year basis, 93rd all-time. Price protection, do not assume it |
Data notes & sources
How this edition's data was built — and why it is the cleanest set in several weeks
The Monday advantage is real and it paid off on all three of the structural problems that have degraded Tuesday-to-Friday editions. Because Friday's US session closed roughly 48 hours before filing rather than ten minutes, AP's tabulation of the major US indexes posted and was obtained on the first attempt — after failing at filing time on four consecutive prior sessions — and it reconciles arithmetically to two independently verified anchors: 7,743.41 − 92.91 = 7,650.50, the verified 18 September S&P close to the cent, and 27,068.72 − 129.34 = 26,939.38 against Thursday's verified Nasdaq Composite of 26,939.37. US cash closes are therefore VERIFIED this edition rather than derived from ETF ratios, and the settled-ETF route was retained as a cross-check (SPY implies 7,745.9, a 3bp tracking difference; IWM implies 2,838.7). Treasury's official par curve had posted BOTH the 24 and 25 September rows, so the entire US curve is the primary series rather than a vendor read, and the 23 September row this desk published last week is confirmed unchanged. And the Fed Rate Monitor gave a genuinely settled read for the first time in several editions, stamped verbatim "Sep 26, 2026 12:35AM EDT" — after Friday's 17:00 ET futures close — so the 66.6% is quoted without the pre-open/pre-close caveat that has attached to it since 15 September. A methodological recommendation follows for the framework document: Monday editions should lead with AP rather than the two-route ETF derivation.
Corrections to prior editions — twenty, of which eight are material and two reverse a published conclusion
(1) MATERIAL, and it reverses a call this desk made explicitly: Brent's 24 September settle was $106.60, not the $102.83 published — an error of $3.77, or 3.7%. Both the dated series and an independent vendor's stated previous close agree. Last week this desk rejected a high cluster at $104.61–107.34 in favour of a ~$102.8–103.1 cluster and said so in terms; the rejected cluster was the correct one and the "reject the high cluster" heuristic is retired rather than reinforced. WTI's 24 September also restates, to $94.61 from $93.66. (2) MATERIAL, and it reverses a trade implication: crude is net LONG, not net short. NYMEX WTI legacy non-commercials are +141,106 and adding; the only net-short crude contract is the far smaller ICE Futures Europe listing at −4,834. The carried claim took the first crude block on the disaggregated page and attributed it to NYMEX. The published inference — that escalation "cuts the other way" via a squeeze — is withdrawn. (3) MATERIAL: the Nikkei's 24 September close was 65,513.99, BELOW the entire 65,647–65,828 range published. The exchange's own archive shows an intraday spike to 66,249.11 reversing to close barely above the low — both carried vendor reads were intraday snapshots of a day that reversed hard, not closes — and the TOPIX independently fell 0.39% the same day, which is irreconcilable with a Nikkei up 1%. The "reopened and held 65,000" characterisation stands; the "+1.0%" does not. (4) MATERIAL: the FTSE 100 fell on Thursday and Europe had no gainer at all. Thursday's close was 10,679.99, −0.24%, not 10,728.00, +0.21% — a 48-point error with the sign inverted — so the published claim that the FTSE was "Europe's only gainer" was false and every major European index fell that day. (5) MATERIAL: Treasury's official 24 September row replaces Friday's vendor curve — 2y 4.87% against a published ≈4.92%, 5y 5.03 against ≈5.06, 10y 5.18 against ≈5.19, 30y 5.47 against ≈5.48, giving 5s30s 44bp against an implied ≈42bp. (6) MATERIAL, and the least comfortable item in the note: V003's stop fired at a local extreme. The official 5s30s path is 41bp (23 Sep) → 44bp (24th) → 51bp (25th), so the spread has retraced ten basis points back through its own 45bp stop in two sessions. The close was correct by its published rules; the outcome is disclosed in §08 rather than omitted. (7) MATERIAL: SKEW's level trend is DOWN, not up as published — 154.5 → 152.1 → 146.6 → 144.9, −6.2% over two weeks. Skew is flattening, so a hedge recommendation premised on rising skew was wrong in the wrong direction. (8) MATERIAL: the ETH ETF row carried as "incomplete at $2.5m" for 23 September has settled at $104.5m — 42 times higher. Publishing $2.5m as a near-zero day would have been wrong by two orders of magnitude and would have supported the opposite narrative; the placeholder rule is vindicated with a measured magnitude and this is the number to cite in its defence. (9) CAC 40 Thursday corrects to 8,081.43 (−0.52%) from 8,100.10 (−0.29%). (10) FTSE MIB Thursday corrects to ≈51,542 (−0.85%) from ≈51,629 (−0.7%). (11) Dow Thursday corrects to 51,349.98 from the ≈51,378 unsettled figure. (12) Five of twelve carried Thursday FX marks were 17 to 182 pips wrong, and for USD/CNY and USD/MXN the carried "Thursday" marks were provably Wednesday's closes — CNY's 6.7114 matches Wednesday to the pip. (13) AUD/NZD's Thursday cross was 1.2384, not ≈1.2329; the error was entirely in the NZD leg (0.5662 actual against 0.5689 published). (14) The carried "Comex copper stocks 254,250t on 22 September" is mislabelled — that is the LME figure, matching the LME stock column exactly. This desk holds no Comex copper stock number and should stop implying one. (15) The gold figure of $4,321.20 seen on one syndicated feed is rejected; three independent routes agree at $4,284–4,285, including a CAD cross-check whose implied USDCAD of 1.4167 sits within 0.16% of the verified close. (16) Against the SOURCE LIBRARY for a fourth consecutive edition, now settled definitively: the leveraged-fund Treasury total of 6,863,118 is wrong by 289,082 contracts (4.4%). A fourth independent rebuild — the first by a different arithmetic path, backing the change columns out of the new 22 September levels — reproduces 6,574,036 on all six contracts and the total to the single contract. This has now been flagged in three addenda; it is corrected in the library today. (17) The 50-day SMA of 7,667.12 carried from the prior edition is not reconcilable with the current 7,699.99 — a 50-day average cannot move 32.87 points in one session, so they are not the same series; independent SPY corroboration favours the current fetch, and the "overbought fully unwound" narrative built on an RSI of 47.3 is stale either way (RSI is 59.1). (18) Bullock's CEDA fireside chat is dated 22 September by the RBA's own speeches page, not 21 September as carried. (19) The BofA Fund Manager Survey published 18 September per an independent source, not the 15 September carried. (20) The buyback-blackout figures previously carried unverified (10% of index weight rising to 61% by 30 September, reopening ~1 November) are now confirmed from a dated, attributed primary source — all three legs.
Conflicts and how they were handled
Four named priorities from Friday were resolved and are worth recording as resolutions rather than as new data. The ACGB 3-year three-way conflict (5.09 / 5.04 / 5.027) is resolved and it was a date problem, not a vendor problem: two independent routes agree at 5.00% for 25 September and the page's own prose implies a 5.04% Thursday, which matches the previously-conflicting dedicated read exactly — the 5.09% figure was the stale outlier, and a house view is markable to a point for the first time in four editions. ⚠ A residual defect is disclosed: the page's stamp reads Sep/26, a Saturday and therefore a non-session, while its prose explicitly names 25 September; the level is accepted on the prose with the stamp defect flagged. The gilt curve is resolved — every UK tenor carries a live Sep/25 stamp, so last week's wholesale rejection does not repeat. The ECB's 29%–51.6% October span is resolved as a date-quality artefact: the 29% figure is dated 10 September, the day of the hike itself and thirteen days before the PMI beat, and is discarded; ~48% is the live figure, with Lane's 24 September remarks arguing it is now somewhat lower. SKEW's three-way percentile mess is resolved — the basis, quoted verbatim, is a 1-year lookback and an all-history one stated simultaneously, and the carried 144.8 / 53rd / 93rd reading was on exactly that basis and was the correct one. Conflicts left open, with both sides published and neither averaged: the ACGB 10-year daily change (the vendor's own prior implies +1.7bp, the carried Thursday implies 0 to −1.8bp — opposite signs; the bull-steepening conclusion is robust to either); the Canada 10-year daily change (−7bp on the vendor's own prior, +4.9bp on the carried figure, with the 12bp gap fitting the one-session-stale pattern proven on CNY and MXN); OAT–Bund at 105.4bp on the dedicated same-page series against 110.5bp on a cross-page derivation, the latter being the documented mixed-page artefact — 105.4bp is used because that series reproduces the carried 23 September value to the decimal and its own legs reconcile to its headline; the Bund 10-year change magnitude (+6.7bp against the carried prior, +1.1bp against the vendor's own; direction agrees either way); Brent's weekly change (+0.43% derived against the vendor's stated ">1%", with the derived figure used); SOL's 24-hour sign (a −0.28% print against +4.92% since Thursday, on precisely the asset where the sign-inversion failure has recurred — the multi-day move is published as the reliable part); the French cabinet date (1 October on the freshest and most specific source against 30 September on another, with a French cabinet conventionally sitting Wednesday — published as 1 October with the variant flagged, and named as the first verification target of the next session); the Q3 Tankan consensus (+25 against +26, published as a range); and the live BofA S&P target conflict (a carried 7,400 year-end / 7,800 12-month of 14 September vintage, unverified today, against 7,100 from a 9 June article — neither presented as confirmed-current).
Rejected outright
A Russell 2000 close of 2,844.32 (+0.31%) published identically by two mastheads — byte-identical content, so one syndicated feed rather than two sources — refuted by IWM's settled +0.11% and by AP's internally consistent weekly. The same feed printed a stale 10-year yield and its sector data is therefore treated as unverified. A gold print of $4,321.20 from that feed, against three convergent routes at $4,284–4,285. The dated FX table's Friday USD/CNY row, which served open = high = low = close with a 0.00% change — a degenerate stale carry-forward on a day the onshore market traded; a second vendor was substituted and its implied Thursday sits within 11 pips. A vendor's BTP prose ("eased to 4.50%, −0.05pp") contradicting its own headline level of 4.5481%; the level was used. The intraday pivot block on the S&P technical page, whose R1−S1 of 10.50 points implies an impossible daily range — it is a short intraday interval, so neither it nor the carried 7,712.63 is published as a daily pivot. Trading Economics' change columns, for a sixth consecutive edition, on the ACGB curve, the German curve (2y/5y/30y printing changes of +0.0003/+0.0001/+0.0002 alongside a 30-year 8bp above the carried prior), zinc, nickel and Brent's weekly. And Trump's claim, confirmed by a US defence official, that "over 22 million barrels" transited Hormuz on Friday night: that requires roughly eleven VLCC-equivalents in one night against tracked single-digit daily transits of all vessel types, and a freight market clearing at a record $1.27m/day is pricing a closed strait rather than a record one. It is published as a claim, not a figure. This is the third consecutive US-government volume claim to fail against vessel tracking, after "above 10 million bpd" and "40 ships a day under American protection" — the desk should stop treating them as data.
Not published
Weekly bond changes outside the US and Australia — no verified 18 September levels were obtainable for the Bund, OAT, BTP, gilt, JGB, Canadian or Swiss curves, so only daily changes are given. Weekly changes for Henry Hub, gold (beyond a vendor's ">1% lower"), silver, platinum, copper, aluminium, zinc, nickel, iron ore, lithium and uranium. The Hang Seng Tech close (the issuer's page returned metadata only). ASX dollar turnover, for a fourth edition, and the day range in dollar terms. The SPI, for an eleventh consecutive edition — the contract is rolled to December and no clean read with the contract named exists; this line should either get a terminal source or be formally retired. Japanese single-stock closing prices, for a sixteenth consecutive edition, and therefore no Japanese relative value. SoftBank's secondary-market levels, its break price and any reported new-issue concession — five sources carry final coupons only. Any 25 September observation for IG, HY or CCC OAS. NYMEX WTI managed-money positioning. Lipper weekly flows, for a fourth consecutive edition, both carried routes now broken rather than merely slow. The fund survey's n and total AUM, after four attempts across editions. The Riksbank's vote and numeric path revision. Banxico's verbatim old-versus-new guidance text. The German 2027 budget arithmetic, not re-verified this run and carried — and its two-bases problem restated verbatim: "over €200bn including special funds" and "€118.73bn core federal" are different numbers measuring different things, so no headline borrowing figure should be quoted without naming its basis. The German Ifo actual for 24 September (the beat is carried, no number recovered). The US September CPI release date — a top-tier October event with no date in this note. Russia–Ukraine overnight drone and intercept tallies, deliberately not forced. Any Hormuz closure probability. The named FOMC voting roster — see below. Next-meeting dates for Banxico, South Africa, Indonesia, Hungary and Turkey.
Cleared this edition
⭐ AP's tabulation, after four consecutive failures — US cash closes are verified rather than derived. ⭐ Treasury's 24 and 25 September par rows, discharging the ledger's highest-priority standing item and replacing a vendor curve. ⭐ A genuinely settled Fed-pricing read. ⭐ The VIX, ending a three-edition structural gap, with the series tie-checked on three independent points. ⭐ The Stoxx 600 and Euro Stoxx 50, so a European weekly change exists again. ⭐ The ACGB 3-year, unblocking a house-view mark after three editions. ⭐ The gilt curve, fully live-stamped. ⭐ A verified Nikkei point close from the exchange archive, which also exposed the carried error. ⭐ The TOPIX, after four editions. ⭐ The NZX 50 at a proper close. ⭐ The A-VIX, arithmetic-tied. ⭐ The full ASX sector table, and the evening wrap rehabilitated with a genuine in-body dateline after last week's total failure. ⭐ The BofA Flow Show and the FactSet edition that 404'd. ⭐ The buyback reopening date, now primary-sourced. ⭐ SKEW's percentile basis. ⭐ Norges Bank's rate path and the NOK reaction, and the Riksbank's path — which is stronger than carried, with a 2026 hike now in its baseline. ⭐ Banxico's guidance substance. ⭐ ECB October pricing, resolved. ⭐ The rare-earths gap, and the answer is that the truce extension agreed nothing on them, with magnet shipments to the US down 20% month-on-month. ⭐ The NZD COT sign conflict, resolved by the data. ⭐ The "protection is cheap" provenance, traced and left withdrawn. ⭐ The FOMC voting roster is resolved as structurally unavailable: both the statement HTML and the statement PDF carry only "approved by a 12–0 vote" with no named paragraph, so this is not a fetch failure and no PDF variant will produce it. Recommend retiring it as a statement-sourced item and re-tasking it to the 16 September minutes, due around 6–7 October.
Traps caught
Two syndicated mastheads presented as two sources — byte-identical content, one feed, and it had the Russell wrong. A degenerate FX row with zero range and zero change on a session that traded. An intraday pivot block that would have been published as a daily one, caught by checking that R1−S1 equals the daily range. A 50-day moving average that moved 32.87 points in one session, which is arithmetically impossible and therefore a different series. Three stale sell-side targets from a 9 June article surfacing on a September query — the dateline trap firing for a third consecutive edition, all three quarantined. A fabricated front row that was transient: the index table that served a 25 September row on the 24th has been replaced by a value that reconciles, which is why the Stoxx 600 is publishable today — the lesson being to re-test a rejected source rather than blacklist it. A weekly sector table misread as daily — an intraday Australian live blog showed "Utilities −4.7%" against a post-close −1.25%, a 3.5pp difference, because the live figures were weekly. A third-party source placing the RBA board meeting on "30 September–1 October" against the Bank's own diary of 28–29 September; the issuer wins. A vendor date-fill fabricating a 23 September TOPIX row on a day Tokyo was closed — it does not contaminate the 24th and 25th, which chain correctly, but it confirms that table cannot be trusted on holiday boundaries. Maysan (Iraq) and Mayun (Yemen) held apart for an eighth consecutive edition, and noted this week: a source this library recommends served four-month-old content, so it is downgraded from "most reliable geopolitical fetch" to "verify the dateline every time". And the holiday check held — no Friday price is quoted for mainland China, Korea or Taiwan, each date is stated, and Hong Kong's Friday session is correctly treated as open.
Tomorrow's first verification targets
The RBA statement and press conference at 14:30 and 15:30, which resolve V004 and V006 and are the reason three of seven views sit on one event. Then: the French Conseil des ministres date, where a 1 October and a 30 September source conflict and V025's catalyst hangs on it; any IG, HY or CCC observation after 24 September, with V017 38bp from its confirm trigger; Treasury's 28 September par row and whether 5s30s extends past 51bp; the Australian monthly CPI consensus, which could not be obtained; today's PBoC fix, the first since the holiday; the Hang Seng Tech close; the US September CPI release date; the Cotality HVI date; SoftBank secondary levels if any dealer run becomes available; Lipper weekly flows via a new route, the existing two being broken; and Governor Cook's post-deadline status, which is a standing daily check rather than a one-off.
United States — markets, data, Fed, fiscal
- US Treasury — official daily par yield curve, September 2026
- AP — How major US stock indexes fared, Friday 25 September
- Google Finance — Nasdaq Composite settled close
- stockanalysis — SPY · IWM
- Investing.com — Fed Rate Monitor (settled read)
- Polymarket — Fed decision in October (event page)
- Census — advance durable goods, August · M3 release schedule
- Cleveland Fed — inflation nowcasting
- Federal Reserve — 2026 speeches · 16 September statement · FOMC calendar
- FXStreet — Hammack, 25 September
- US Treasury — expanded long-end buybacks · tentative auction schedule
- Helious — 7-year auction detail · MarketScreener — 33-year high auction yield
- SCOTUSblog — Trump v. Cook · the August removal notice
- CRFB — fiscal deadlines (⚠ last updated 11 June 2026) · House Appropriations — clean CR to 11 December
- FactSet — Earnings Insight, 25 September
Rates, FX & central banks
- Dated FX histories: AUD/USD · NZD/USD · EUR/USD · USD/JPY · GBP/USD · USD/CAD · USD/CHF · DXY
- ideal-investisseur — OAT–Bund spread series
- TE — gilts · Bunds · JGBs · ACGBs · ACGB 3-year · OATs · BTPs
- RBA — forward diary · speeches list (proves a negative) · Iain Ross — "A Wage-price Spiral: What are the Chances?", 22 September
- CBA — RBA preview (4.60% as the peak) · FXStreet — RBA preview
- centralbank.watch — RBA pricing
- Bank of Japan — meeting schedule (the only authority for the decision date)
- Norges Bank — September meeting and rate path · Reuters via Investing.com — the NOK reaction
- Riksbank — September Monetary Policy Report
- Reuters via Investing.com — Banxico guidance
- Econostream — ECB speaker slate · investingLive — ECB October pricing post-PMI
- Bank of England — upcoming events (Ramsden, Taylor, Mann, Bailey) · MPC dates · Reuters via Yahoo — Bailey at Oxford
- TE — global policy rates · cbrates — meeting calendar
Australia, New Zealand & Asia
- Investing.com — ASX 200 dated history · Market Index — evening wrap, 25 September 16:28 AEST · ABC — markets live, 25 September
- Investing.com — ASX close, breadth and A-VIX
- ABS — August labour force · forward release calendar
- Cotality/CoreLogic — Home Value Index tracker · Domain — auction results (recomputed from city rows)
- Australian Broker — 18 lenders repriced in September
- TE — Westpac consumer sentiment · NAB business confidence
- Nikkei — the exchange's own dated archive with OHLC · TOPIX dated history · TE — Japan equities and the BoJ path signal
- JPX — trading calendar
- TE — Hang Seng · mainland China · China calendar
- Korea JoongAng Daily — the KOSPI post-Chuseok, 27 September
- BBN Times — Sensex and a seventh weekly Nifty loss
- TE — NZX 50 · Rio Times — Asia intelligence brief, 26 September
- MarketScreener — Taiwan closed Friday and Monday
Europe & geopolitics
- Stoxx 600 dated history · Euro Stoxx 50 · marketscreener — CAC 40 · finanzen.at — SMI · hl.co.uk — FTSE 100 · IBEX 35 · SoldiOnline — FTSE MIB
- Reuters via Investing.com — European weekly wrap and sectors
- NIM — German consumer climate, October (primary) · Newsquawk — French consumer confidence
- francebudget.fr — PLF 2027 statutory timeline · L'Échiquier Social — cabinet date (27 September) · Euronews — the €54bn effort · Journal du Net — the censure motion
- House of Lords Library — UK fiscal outlook and headroom
- Al Jazeera — Iran's seven-day roadmap · Trump rejects it, 26 September · Axios — Trump on further talks, 27 September
- gCaptain — the Riyadh strikes · EnergyConnects
- Tankermap — Hormuz transits · Windward · OilPrice — transits, VLCC rates, pipeline status
- Al Jazeera — drones launched from Maysan, Iraq · Euronews — Mayun Island, Yemen · Dryad — the pipeline restart and throughput
- Rare Earth Exchanges — what the truce did not agree · CNBC — the two-month extension · EUNews — the EU Chamber on rare earths
Commodities, credit & digital assets
- Brent dated settle series · TE — Brent · WTI dated series · Buckhead Energy
- Kitco — gold spot, timestamped · TE — gold · Gold Stock Canada — the CAD cross-check
- Westmetall — LME copper official settlements and stocks · TE — iron ore
- FRED — IG OAS · HY OAS · CCC OAS · Convex — HY corroboration
- Business Recorder — SoftBank $11.1bn · Traders Agency — tranches and analyst colour
- CoinGecko · CoinDesk · Farside — BTC ETF flows · ETH ETF flows · Coinalyze — derivatives
- Polymarket — CLARITY Act (event page) · CoinGecko — Counterparty/XCP
- EIA — weekly release schedule · Rigzone — EIA weekly inventories
Positioning, flows & sentiment
- CFTC — Traders in Financial Futures (leveraged funds) · CBOT legacy · CME legacy · COMEX · NYMEX · petroleum disaggregated · ICE Futures US (dollar index) · release schedule
- AAII — dated sentiment results
- BofA Flow Show via Finvaulta, 25 September · the September Fund Manager Survey detail
- ICI — combined flows · money market assets
- Citadel Securities GMI — "2H September: Getting Closer" (buyback blackout, vol-control, CTA z-score) · "September Setup" (the expiry denominators)
- VIX dated history · VIX term structure · CBOE SKEW with percentile basis · put/call · McClellan, highs/lows, Hindenburg conditions
- zerogex — SPX gamma levels, flip and walls · % above the 200-day · S&P 500 technicals
- University of Michigan — final September sentiment · HSBC — S&P target, 13 September
Calendars
- ForexFactory — week of 28 September (⚠ impact column did not render for a third time; all H/M/L ratings are this desk's own) · BLS — Employment Situation schedule · BEA · TE — Japan calendar