Warsh Reprices Rates; Quality Splits Earnings

2. Complete macro-event table

Rank	Event and time (EDT)	Actual vs consensus / prior	Surprise and immediate reaction	Why it mattered
1	Fed Chair Warsh, Jackson Hole — 10:00	No rate decision; 2% target “firm, fixed,” employment at full employment, inflation still too broad	September hike odds 35.4%→55.7%; 2Y +12.8 bp; DXY +0.61%; RUT −1.4%	Reset the burden of proof toward tightening while reducing forward guidance. (Fed)
2	Hormuz operational reopening — overnight/session	Mines cleared from recognized lanes; weekly transits +30%+; no signed normalization deal	Brent $89.31, −0.43%; WTI $83.40, −0.16%; weekly crude losses >4%	Reduced near-term scarcity without removing attack, insurance or diplomatic risk. (Reuters)
3	Chicago PMI — 9:45	47.1 vs 58.0 median; prior 57.6, unrevised	10.9-point miss; orders −15.4, production −8.8, prices paid +3.8	Large regional contraction with an adverse stagflationary mix; reaction confounded by 10:00 releases. (ISM Chicago)
4	Japan FX-intervention disclosure — 6:00	¥15.3993tn / $96.5bn during Jul. 30–Aug. 26, a record	No clean disclosure-window move; USD/JPY later 160.11 as Warsh dominated	Quantified the scale of yen defense and the limits of intervention against carry/fiscal pressure. (MOF)
5	Canada Q2 GDP — 8:30	+3.3% annualized vs +3.4%; prior revised +0.3% from −0.1%; June +0.3% vs +0.2%; July flash 0.0%	Near-consensus rebound; CAD/rates response modest and quickly confounded	Broad domestic-demand recovery, but July stall and new tariffs made it backward-looking for BoC policy. (Statistics Canada)
6	France GDP/CPI/fiscal bundle — 2:45	Q2 GDP 0.0% vs preliminary/consensus +0.2%; Q1 −0.2% revised; Aug CPI 2.4%, HICP 2.7%	OATs already under fiscal pressure; no clean one-print shock	Stall-speed growth plus energy inflation made the 5% deficit target harder and kept OAT political risk elevated. (INSEE GDP, INSEE CPI)
7	Tokyo CPI — 7:30 p.m. Aug. 27	Headline 1.9%; core ex-fresh food 1.8% vs Reuters 1.7%; core-core 2.0%; July rebased to 1.8/1.7/1.8	USD/JPY initially near flat; JGB weakness had oil, Himino and auction confounds	Services/rents supported a September BOJ hike even as subsidies restrained energy. (Statistics Bureau)
8	BLS preliminary CES benchmark — 10:00	Nonfarm −79k (−0.1%); private −178k; government +99k	Much smaller than last year’s preliminary revision; no official monthly history changed today	Headline measurement relief concealed softer private employment. (BLS)
9	Spain flash CPI — 3:00	National 4.3% vs Reuters 4.2%, prior 3.6%; HICP 4.5% vs 4.6%, prior 3.9%; core 2.9% from 3.0%	Headline upside but underlying easing; no isolatable EUR/rates move	Energy and fuel-tax normalization lifted headline inflation without proving a demand-driven core breakout. (INE)
10	Michigan final sentiment — 10:00	Headline 51.7 vs 51.0 flash/consensus, prior 55.2; 1Y inflation 4.0% vs 4.3% flash; 5Y 3.3% unchanged	Better than flash, still weak; Warsh/BLS made reaction unidentifiable	Purchasing-power stress persisted; shorter-run inflation relief did not re-anchor long-run expectations. (Michigan)

3. Detailed macro events and opinion clusters

3.1 Warsh at Jackson Hole: September became live, not certain

Facts. Warsh’s official text made price stability the predominant current focus, called the 2% PCE objective fixed, described labor markets as consistent with full employment and rejected routine forward guidance. He said he was committed to a discipline, not a decision. The front end, dollar, gold and bitcoin supplied the clean event signature; closing equity/commodity moves also contained company-specific and Hormuz news. (Fed text, Reuters)

Cluster A — “The hike clock restarted” (high evidence). Nathan Shetty/SEI, Christopher Hodge/Natixis, Gary Schlossberg/Wells Fargo Investment Institute, Jon Faust/Johns Hopkins and Aditya Bhave/Bank of America read the combination of sticky inflation, resilient demand and non-restrictive conditions as a genuine near-term hike setup. Bhave’s sharp test was that firm August data should compel a September move. Implication: higher front-end yields, a flatter curve and pressure on small caps/liquidity assets over 0–3 months. Risk: soft payrolls or CPI. (Reuters analyst roundup)

Cluster B — “A compass, not a GPS” (medium-high evidence). Ellen Hazen/F.L. Putnam, Robert Pavlik/Dakota Wealth, Brian Storey/Brinker and Sam Stovall/CFRA saw principles and optionality rather than an imminent commitment. Implication: unstable meeting odds and more event volatility, not a one-way hike trade.

Cluster C — “Talk is cheap” (medium evidence). Michael Strain/AEI, Eugene Epstein/Moneycorp, Cyrus Amini/Hyphen and Bhave focused on the rhetoric/action gap and the absence of a transparent reaction function. A hold despite firm data would reverse part of the front-end/USD move and damage credibility. (AP economist reaction)

Consensus / disagreement / resolver. Consensus: hawkish versus positioning. Sharpest disagreement: September hike versus restored optionality. August payrolls on Sept. 4, CPI on Sept. 11, and the Sept. 16 FOMC decision resolve it. Confidence: high on the repricing, medium on the meeting outcome.

3.2 Hormuz: partial operational reopening, diplomatic normalization unverified

Facts. CENTCOM declared recognized lanes mine-free; Lloyd’s List Intelligence counted 114 transits in Aug. 17–23, more than 30% higher week over week; Goldman estimated Gulf exports at 15–16 mb/d, still 7–8 mb/d below prewar levels. Oman, Iran and Qatar were discussing a phased corridor, but no security/sanctions agreement was signed, vessels were still being attacked, and the IMO said up to 400 ships remained unable to depart safely. (Oman, USNI/Lloyd’s List, IMO)

Cluster A — flow recovery is bearish near term (medium-high). Janiv Shah/Rystad argued that extra barrels, corridor prospects and mine clearance surprised the market. Implication: a softer prompt barrel over days/weeks. Cluster B — dangerous workaround, not normality (high). Richard Meade/Lloyd’s List emphasized shuttle tankers, ship-to-ship transfers and a narrow risk-tolerant owner/crew pool. Supply can improve faster than freight and war-risk premia. Cluster C — rumor risk (medium). Phil Flynn/Price Futures and John Evans/PVM treated weekend-deal talk as premature and noted Russian refinery strikes supporting products. (Reuters oil report)

Consensus / disagreement / resolver. Export availability improved; “open” versus “normal” is the disagreement. Require sustained two-way transits, discharged Gulf volumes, falling VLCC freight/insurance, no attacks and published corridor terms over 1–3 weeks. Confidence: high that operational risk eased; low that diplomacy normalized.

3.3 Chicago PMI: an adverse regional growth/inflation mix

Facts. The 47.1 print was 10.9 points below the eight-economist WSJ median. New orders fell 15.4 points, backlogs 12.1, production 8.8 and inventories 14.0; employment rose 4.3 into slight expansion, while prices paid rose 3.8 to its highest since February 2022. (ISM Chicago, Trading Economics)

Cluster A — demand air pocket (high on event, medium nationally). The orders/production/backlog stack says more than delivery noise and is negative for manufacturing cyclicals over 1–3 months. Cluster B — stagflationary, not a duration-bull miss (medium-high). Rising prices paid offsets the clean easing interpretation. Cluster C — wait for ISM (high). Chicago is volatile and narrow; national ISM orders, production and prices must confirm it. Same-day named public economist commentary was sparse, so this cluster relies mainly on the survey and reputable market coverage rather than invented sell-side views.

Consensus / disagreement / resolver. Consensus: exceptionally large downside surprise. Disagreement: national signal versus regional volatility. National ISM and August payroll/manufacturing-hours data resolve it. The 9:45–10:00 event window was contaminated by Michigan, BLS and Warsh; no clean asset move is attributed to Chicago alone.

3.4 Japan’s record FX intervention: forceful defense, incomplete regime change

Facts. MOF disclosed ¥15.3993tn ($96.5bn) of yen buying for Jul. 30–Aug. 26, covering the Jul. 30 intervention and the Jul. 31 U.S.–Japan coordinated action; daily allocation is not available until the quarterly data. MOF had previously confirmed coordination and possible future use of the Fed’s FIMA repo facility. (MOF monthly release index, MOF coordinated-action statement, Reuters)

Cluster A — credible two-way risk (high). The record amount and U.S. backing defended 164 and proved a reusable official threat. Cluster B — carry still wins without monetary/fiscal help (high). FX strategists emphasized that most of the 163-to-155.2 gain had retraced; intervention cannot erase the rate differential, fiscal unease or energy-import drag. Cluster C — bridge to BOJ action (medium). A September BOJ hike would turn intervention from a level defense into policy alignment.

Consensus / disagreement / resolver. Consensus: intervention changed tail risk, not the structural carry. Success can be scored as staying below 164; failure as retracing most of the move. November daily MOF data, USD/JPY behavior at 160–164, BOJ action and any FIMA disclosure resolve it. No clean 6:00 a.m. EDT asset reaction was found; Warsh drove the later U.S. rates/FX move.

3.5 Canada Q2 GDP: broad rebound, but the July stall matters more now

Facts. Real GDP grew 3.3% annualized, narrowly below 3.4% consensus; Q1 was revised to +0.3% from −0.1%. Household consumption, housing and business investment broadened the rebound. June GDP rose 0.3% versus 0.2% expected; July’s advance estimate was flat. (Statistics Canada, Reuters)

Cluster A — stronger baseline (medium-high). RBC and National Bank highlighted broad private demand and a smaller output gap, limiting room for near-term BoC easing. Cluster B — “old news” (high). Desjardins said the near-consensus Q2 print predated the July stall and fresh tariff shock; policy should focus on H2. (Desjardins) Cluster C — no recession, moderate trend (medium). Revisions remove a technical contraction without proving a strong ongoing expansion.

Consensus / disagreement / resolver. Consensus: Q2 was solid and broad. Disagreement: smaller output gap versus backward-looking bounce. BoC guidance, official July GDP, jobs, core CPI and tariff-exposed manufacturing resolve it. Confidence: high on Q2; medium-low on H2 persistence.

3.6 France: stall-speed GDP plus an energy/fiscal constraint

Facts. INSEE revised Q2 growth to 0.0% from +0.2%; Q1 was −0.2%. Purchasing power per consumption unit fell 0.6%, household savings remained high at 17.2%, and corporate margins were 31.5%. August CPI accelerated to 2.4% and HICP to 2.7%, led by energy. The reported deficit remained near 5.1% of GDP against a 5.0% target, and the 10-year OAT traded above 4.1%. (INSEE GDP, INSEE CPI, Le Monde)

Cluster A — fiscal/political OAT risk (high). Growth arithmetic makes the deficit target harder and pushes the adjustment burden toward politically difficult savings. Cluster B — energy headline, not domestic overheating (medium-high). Danske and same-day European coverage treated the CPI acceleration as mostly energy; underlying demand remained weak. Cluster C — longer ECB tail risk (medium). If the energy shock broadens into services/wages, a one-and-done ECB hike becomes inadequate.

Consensus / disagreement / resolver. France is at stall speed and fiscal risk matters more than a single CPI print. Disagreement: temporary energy versus second-round inflation; manageable 5% deficit versus 5.3%–5.4% slippage. Final CPI Sept. 15, the budget update, Q3 GDP and autumn rating reviews resolve it.

3.7 Tokyo CPI: rebasing softened the history, services kept BOJ tightening alive

Facts. The core print of 1.8% beat Reuters’ 1.7% but matched Bloomberg’s 1.8%; July headline/core/core-core were all revised down 0.2 point under the 2025-base update. Rents and services were firm; subsidies and rice disinflation restrained headline pressure. USD/JPY barely moved on release; JGB weakness also reflected oil, Deputy Governor Himino and a poor auction. (Statistics Bureau, Reuters)

Cluster A — September hike (medium-high). Ko Nakamura/Okasan and several same-day economists said broad cost pass-through and sticky services remove the reason to wait. Cluster B — subsidies delay, not eliminate, inflation (high). Energy support masks some pressure, while services/rents are more relevant for BOJ persistence. Cluster C — rebase caution (medium). The “beat” depends on the consensus source and a revised prior, so the signal is less dramatic than the headline.

Consensus / disagreement / resolver. BOJ normalization remains live. Disagreement: underlying persistence versus a temporary energy realization. Nationwide CPI, wages, services and September BOJ guidance resolve it. Confidence: medium-high on hike compatibility, medium on September timing.

3.8 BLS benchmark: a small headline revision with a softer private core

Facts. The preliminary March 2026 benchmark was −79,000 for total nonfarm employment and −178,000 private, offset by +99,000 government. Trade/transport/utilities was −98,000, manufacturing −67,000, professional/business services −76,000 and education/health −96,000; construction, information and finance were positive. The revision will not alter official monthly history until the final benchmark in February 2027. (BLS, Reuters)

Cluster A — measurement relief (high). The −0.1% total revision is small relative to last year’s much larger preliminary shortfall and weakens the “payroll data are badly overstated” claim. Cluster B — private weakness matters (medium-high). The government offset masks softer market-sector employment, relevant for profits and demand. Cluster C — sector reallocation (medium). Construction/information/finance gains alongside wholesale/retail/manufacturing losses may reflect logistics/capital reallocation rather than uniform recession.

Consensus / disagreement / resolver. Small total revision, less comfortable private mix. Final February benchmark, QCEW detail and subsequent payroll/revision patterns resolve it. Same-day named commentary was sparse; the 10:00 event window was dominated by Warsh.

3.9 Spain CPI: an energy/fiscal spike, not yet a core breakout

Facts. National CPI accelerated to 4.3% and HICP to 4.5%, while national core eased to 2.9%. Energy rose 16.7% year over year; fuel and tax/base mechanics were central. (INE, Reuters)

Cluster A — mostly energy (high). Same-day economists and the Economy Ministry emphasized fuel/base effects and easing core pressure. Cluster B — second-round risk (medium). A temporary shock can still enter food, services, wages and expectations, especially if diesel-tax relief changes relative prices. Public, named sell-side commentary was thin; no synthetic broker consensus is supplied.

Consensus / disagreement / resolver. Headline inflation rose materially; the underlying series did not show a new demand breakout. Euro-area flash HICP, Spain’s Sept. 15 final components, autumn services/wages and post-tax fuel prices resolve the disagreement.

3.10 Michigan sentiment: better than flash, still poor

Facts. Final sentiment was revised to 51.7 from 51.0 but stayed below July’s 55.2. Current conditions were 51.9, expectations 51.5, one-year inflation 4.0% and five-year inflation 3.3%. (Michigan, Reuters close)

Cluster A — less bad, no recovery (high). Joanne Hsu/Michigan stressed that consumers remained constrained despite the upward revision. Cluster B — short-run relief, incomplete anchoring (high). One-year expectations eased; five-year stayed high. Cluster C — survey noise (medium). Political polarization and high-frequency headlines mean spending/payroll confirmation is necessary.

Consensus / disagreement / resolver. Purchasing-power stress is the durable signal. Disagreement: genuine consumption warning versus noisy attitudes. Sept. 4 payrolls, Sept. 11 CPI/preliminary Michigan and Sept. 16 retail sales resolve it. Reaction was inseparable from simultaneous Warsh/BLS headlines.

4. Complete earnings and call table

Timing note: “Pending” means the release crossed after that listing’s close. Friday’s move is shown for context but is not mislabeled as a reaction.

Rank	Company	Release / call (EDT)	Result versus credible benchmark	Guidance / decisive issue	Price reaction
1	Victory Giant (300476.SZ/2476.HK)	Aug. 27 9:42 p.m.; call record Aug. 28	Q2 revenue RMB6.11bn +29.5%; recurring profit RMB1.16bn −5.3%; GM 32.1%, −673 bp; profit proxy miss 26.8%	AI-PCB orders extend into 2027, but front-loaded labor/depreciation and RMB18bn capex ceiling challenge returns	−8.15% in the fully post-print A-share session. (filing)
2	MINISO (MNSO/9896.HK)	4:34; call 5:00–6:19	Q2 revenue RMB5.81bn, +0.4% vs FactSet; adjusted EPS RMB1.76 vs 2.24	FY margin decline widened to 3–4 points; overseas direct-store investment and distributor cleanup	ADR −4.44%; HK close preceded release. (release)
3	BYD (1211.HK/002594.SZ)	6:36; no call found	Q2 revenue RMB194.6bn, 7%–10% miss; profit RMB8.24bn, +30% YoY but estimate-dependent	Export/premium mix restored margin; domestic volume, FX/finance cost and 5m-unit path are the test	Both closes pre-release; Aug. 31 pending. (filing)
4	Frontline (FRO)	1:28; call 9:00	Record net $659.2m/$2.96; adjusted $580.2m/$2.61; reported beat, adjusted comparison mixed	Q3 substantially booked at exceptional rates; $2.61 regular + prospective $0.80 special dividend	Oslo +2.30%; NYSE +1.01%. (release)
5	China Construction Bank (0939.HK/601939.SS)	4:52; call 5:30	Profit RMB169.56bn, 10% below Visible Alpha; NIM 4 bp above	NII recovery versus +20.8% impairment and worsening retail NPL ratios	Both closes pre-release; Aug. 31 pending. (filing)
6	Hafnia (HAFN/HAFNI)	1:30; call 8:30	EPS $0.56 vs ~$0.55; TCE $372.9m; clean EBITDA/profit comparison slightly below S&P	Q3 80% covered at $30,716/day, below Q2; maximum 90% payout	NYSE +2.17%; Oslo +1.30%. (release)
7	CHAGEE (CHA)	7:00; call 8:00–8:52	Revenue RMB3.415bn +2.5%, miss; adjusted EPS modest beat; adjusted profit −22%	July comps low-single-digit negative; August expected positive; no numeric H2 guide	+4.35%, confounded by same-day Macquarie upgrade. (release)
8	Sungrow (300274.SZ)	6:19; no call	Derived Q2 revenue RMB15.35bn, ~37% miss to public proxy; profit RMB2.97bn; GM 38.6%	Inverter margin strong; storage margin −749 bp; H2 must accelerate sharply	Close pre-release; Aug. 31 pending. (filing)
9	Zurich Airport (FHZN.SW)	European morning; call same day	H1 revenue CHF673.6m, EBITDA CHF374.2m, net CHF163.7m; each narrowly below AWP	FY EBITDA roughly flat; Noida startup, heavy capex and dividend capacity dominate	Shares fell despite higher YoY results. (H1 report)
10	Codelco (unlisted)	~12:19 report; no public Q&A found	H1 pre-tax $1.97bn vs $429m; production −11% to 564kt; cash cost +7%	Copper price rescued earnings; El Teniente disruption makes prior output target hard	Copper gained about 1.5%; no listed Codelco equity. (Reuters)
11	BOC Hong Kong (2388.HK)	4:31; briefing 5:00	Profit HK$23.74bn, 3.5% beat; credit cost 27 bp; adjusted NIM 1.57%	HK$10.5bn minimum extra returns below Goldman expectation; special yield below UBS expectation	Close +2.27%, but pre-release; Aug. 31 pending. (results)
12	China Merchants Bank (3968.HK/600036.SS)	5:59; call Aug. 31	Q2 revenue proxy +3%; EPS proxy −2%; H1 profit +2.0%	Wealth fees recover; cards/consumer NPL formation worsen	Both closes pre-release; Aug. 31 call/reaction pending. (filing)
13	Montage Technology (688008.SS/6809.HK)	6:59–7:52; briefing Sept. 8	Q2 revenue RMB1.875bn, +5.9% vs proxy; recurring H1 profit +21%, headline +72% on marks	DDR5 monetizing; newer MRDIMM/PCIe/CXL revenue roughly flat sequentially	Both closes pre-release; Aug. 31 pending. (filing)
14	Wuliangye (000858.SZ)	7:04 server time; no call	Q2 revenue −13.2%; H1 profit near bottom of preannouncement; negative operating cash flow	Flagship volume/share improve, promotions and weak cash conversion challenge quality	Close pre-release; Aug. 31 pending. (filing)
15	BOE Technology (000725.SZ)	6:49; no call	H1 headline profit within preannouncement; recurring profit 2.3% below floor; Q2 GM down sequentially	OLED/MLED mix versus panel pricing and fair-value-gain dependence	Close pre-release; Aug. 31 pending. (filing)
16	COSCO Shipping Holdings (1919.HK/601919.SS)	Post-close	H1 profit RMB13.4bn range; Q2 volumes recovered while margins/rates remained below year ago	Port strength and Hormuz scarcity versus fleet-cycle/cost normalization	Both closes pre-release; Aug. 31 pending. (HKEX)
17	Yankuang Energy (1171.HK/600188.SS)	7:32–7:48; call Aug. 31	CAS profit +47.8%, recurring only +4.0%; price/mix and chemicals improved	One-off Xintai gain, rising costs and H2 volume/capex bridge	Both closes pre-release; Aug. 31 pending. (HKEX)
18	China Shenhua (1088.HK/601088.SS)	7:13–8:40; briefing Sept. 4	H1 revenue +7.9%, IFRS profit +1.9%; near top of company range	Purchased coal inflated volume/revenue; core coal and power PBT fell; 74% payout	Both closes pre-release; Aug. 31 pending. (HKEX)
19	Weichai Power (2338.HK/000338.SZ)	Post-close	H1 headline growth aided by financial/one-off items; engine/AI-data-center exposure offset softer trucks	KION/Europe execution and core cash conversion	Listed closes precede full filing; clean reaction pending. (HKEX issuer feed)
20	Three-Circle (300408.SZ/6951.HK)	Post-close	H1 profit near upper half of guide; margin expanded; cash conversion lagged	MLCC pricing/utilization and AI optical demand versus working capital/valuation	Mixed positioning; Aug. 31 gives cleaner verdict. (HKEX)
21	CITIC Limited (0267.HK)	Post-close; Q&A Sept. 4	H1 profit RMB33.76bn +8.1%, dividend RMB0.21	Securities/materials drove growth; impairments/property keep recovery narrow	+0.70%, but subsidiaries already disclosed; not a clean surprise test. (results)
22	ICBC	4:30; call 5:00	Profit RMB173.68bn +3.3%, 3.1% miss; NIM 1.29%	Impairment +22%, mortgage/property NPL pressure	Pending Aug. 31. (filing)
23	Bank of China	4:32; call 5:00	Common profit RMB115.33bn, ~2% beat; NIM +5 bp vs Visible Alpha	Overseas franchise helps; mortgage NPL 0.96% vs 0.60%	Pending Aug. 31. (filing)
24	Agricultural Bank of China	filing 5:13; call 5:00	Profit ~5.1% below Visible Alpha; NIM 7 bp above	Scale/NII versus weaker fees, higher provisions and lower CET1	Pending Aug. 31. (official)
25	Bank of Communications (3328.HK/601328.SS)	4:30; call 5:00	Q2 profit 1.8% above S&P; NIM +2 bp YoY	Funding relief versus personal-loan/card deterioration beneath the headline NPL ratio	Both closes pre-release; Aug. 31 pending. (filing)

5. Detailed company sections and opinion clusters

5.1 Victory Giant: AI demand met a capacity-return veto

Facts. The Aug. 28 session was fully post-print: the filing crossed the prior evening. Revenue and headline profit grew, but Q2 recurring profit fell 5.3%, recurring margin fell seven points and gross margin dropped to 32.1%. H1 operating cash improved, yet RMB7.08bn of capex left simple free cash flow deeply negative. Management blamed new-capacity labor/depreciation and said orders extend into 2027. (filing, call record)

Bull cluster (medium): AI PCB demand, long order visibility and cash generation support eventual operating leverage. Bear cluster (high): the 8.15% selloff, 26.8% profit-proxy miss, recurring decline and RMB18bn capex ceiling say the market now prices return on capital, not revenue. Consensus/disagreement/resolver: demand is real; economics are disputed. Q3 recurring margin, utilization, free cash flow and capex conversion resolve it. This is the day’s strongest AI-supply-chain warning.

5.2 MINISO: a real sales win, an equally real margin reset

Facts. Q2 revenue beat FactSet narrowly, but adjusted EPS missed by about 21%. H1 China same-store sales grew mid-single digits; overseas same-store GMV fell low single digits, distributor revenue declined and overseas inventory days rose. Management now expects FY adjusted operating profit ex-FX to decline high single digits and margin by 3–4 points, double the previous margin-reset range. (release, call summary)

Bull cluster (medium): Goldman’s Michelle Cheng, CICC’s Runbo Yang and management focused on China large-format productivity, IP and North America stockouts; proprietary IP is ahead of plan. Bear cluster (high): UBS’s Samuel Wang and the tape focused on direct-store losses, distributor cleanup, TOP TOY slowing and structural mix dilution. Consensus/disagreement/resolver: MINISO is taking China category share; whether FY26 investment creates a 2027 margin turn is open. September in-stocks, overseas inventory/sell-through and direct-store losses resolve it. ADR reaction: −4.44%.

5.3 BYD: export-margin inflection, domestic/top-line fragility

Facts. Q2 gross margin recovered 261 bp and profit grew 30%, but revenue missed Visible Alpha by 10.1% and a second public proxy by 7.4%. Overseas revenue rose 34% and exceeded domestic revenue; H1 NEV sales fell 16% while exports surged 68%. Finance expense swung adversely by RMB8.34bn and inventory days lengthened. (filing, Reuters)

Bull cluster (medium-high): Jiahui Huang/Dow Jones and Phate Zhang/CnEVPost see the first quarterly profit growth in five quarters, overseas margin and premium brands as an inflection. Bear cluster (medium-high): Yale Zhang/Automotive Foresight and Zavier Wong/eToro say domestic competition, tariffs and brand-building costs limit the offset. Resolver: 500k-plus monthly sales, overseas realized margin, Blade Battery capacity and finance-cost normalization. Consensus: mix improved; disagreement: capacity bottleneck versus weak demand. Listed reaction awaits Aug. 31.

5.4 Frontline: record scarcity earnings, already discounted

Facts. Q2 spot TCEs reached $152.7k/day for VLCCs, $111.5k Suezmax and $92.4k LR2/Aframax. Q3 is mostly booked at similarly extraordinary rates, though load-to-discharge accounting and ballast days mean full-quarter realized rates will be lower. Management elected to distribute cash and sell ships rather than deleverage or buy expensive assets. (release, call)

Bull cluster (high near term): route inefficiency, waiting, dark fleets and inventory refill keep effective supply tight. Bear cluster (medium): Oda Hånes Lund/SB1 and Bendik Folden Nyttingnes/Danske see normalization, weaker U.S. exports and the VLCC orderbook capping duration. Consensus/disagreement/resolver: Q3 cash is largely locked; the dispute is 2027–30. Q4 fixtures, China/OECD refill, Hormuz/Red Sea routing and scrapping versus deliveries resolve it. The modest +1.01% NYSE move implies records were expected.

5.5 Chinese mega-banks: the NIM floor is real; household recovery is not

Shared fact. Deposit costs fell faster than asset yields at ICBC, BOC, AgBank, CCB and BoCom, stabilizing NIM and lifting NII. Profit grew 3%–5%. Yet impairment rose 18%–22% at several banks; mortgages, cards or unsecured consumer NPL ratios worsened; CET1 generally fell. (Reuters sector synthesis)

Constructive cluster (high): management teams and pre-print S&P/Wanlian work see maturing high-cost deposits creating a defensible NIM floor and supporting 31% interim payouts. Cautious cluster (high): Fitch’s Elaine Xu sees little additional margin room under low rates/weak demand; Moody’s Nicholas Zhu sees adverse consumer selection and weak card fees. Consensus/disagreement/resolver: funding relief is genuine; durable credit/fee growth is not. H2 NIM after repricing fades, retail NPL formation, impairment and capital ratios resolve it.

ICBC

Profit missed the only robust public consensus by 3.1%; NIM beat, but impairment rose 22%, overdue/special-mention loans grew and mortgage/property NPL ratios worsened. Management says it prefers current risk recognition to manufactured growth. Confidence: high on managed stabilization, medium-low on rerating. (ICBC filing)

Bank of China

Common-shareholder profit beat Visible Alpha about 2% and NIM beat 5 bp, helped by a foreign-currency/overseas franchise that supplied 27% of profit. Mortgage NPL rose to 0.96% from 0.60% and asset impairment rose 18%. The dispute is whether overseas diversification makes BOC structurally better or merely less representative of domestic credit. (BOC filing)

Agricultural Bank of China

NIM beat the limited public estimate by 7 bp, but profit missed roughly 5%, fees fell, provisions rose and CET1/coverage declined. Bulls emphasize its rural deposit franchise; skeptics emphasize capital consumed by policy-led volume. H2 deposit cost and fee growth are the resolver. (AgBank report)

China Construction Bank

Profit missed Visible Alpha by 10%, although NIM beat by 4 bp and NII grew 8.5%. Credit impairment rose 20.8%, while mortgage, personal operating-loan and card NPL ratios worsened. The print supports a funding-cost floor, not a broad private-demand recovery. (CCB filing)

Bank of Communications

Q2 profit was about 1.8% above S&P’s limited estimate and NIM rose 2 bp; personal-loan NPL climbed to 2.02% from 1.58% and card NPL to 3.80% from 2.68%. Management expects the remaining deposit-repricing benefit in H2; Xu/Zhu’s sector caution is the counterweight. (BoCom filing)

5.6 Hafnia: exceptional Q2, visible Q3 normalization

Facts. Basic EPS beat by one cent, but a cleaner S&P comparison put EBITDA/profit slightly below expectations and the dividend above. Q3 is 80% covered at $30.7k/day versus Q2’s $44.1k average; net LTV is 13% and the company paid the 90%-of-profit maximum. (report)

Bull cluster (medium-high): management and Clarksons’ Frode Mørkedal see route fragmentation, inventory refill and winter strength. Bear cluster (medium): lower product availability and eventual Hormuz/Red Sea normalization can release tonnage before cargo recovers. Resolver: Q4 fixtures, rerouting persistence, product supply and fully committed LTV as newbuild cash leaves. Consensus: still very profitable; disagreement: duration. NYSE +2.17%.

5.7 CHAGEE: improving comps versus a weak premium-tea base

Facts. Revenue grew 2.5% but missed the credible range; Greater China GMV fell while overseas GMV more than doubled. GAAP profit benefited from cost control; adjusted profit declined. Management said July same-store sales were low-single-digit negative and August should turn positive. (release, call)

Bull cluster (medium): sequential comps, overseas growth and operating discipline point to a turn. Bear cluster (medium-high): premium discretionary demand, active members and company-store economics remain weak; store conversion can lift reported revenue without equivalent GMV economics. Macquarie’s Linda Huang upgraded at 11:53 a.m., contaminating the +4.35% reaction. Resolver: September/October comps without heavy promotion, active members and franchise revenue.

5.8 Sungrow: margin discipline inside a severe revenue reset

Q2 revenue missed the only public proxy by roughly 37%; H1 domestic revenue fell 55%, project-development revenue 85% and storage margin 749 bp. Offsets were 42.7% inverter margin, better Q2 gross margin, positive free cash flow and only a 10.6% overseas decline. (filing)

Bull cluster (medium): mix discipline, overseas concentration and cash show this is not a universal export-demand collapse. Bear cluster (high): storage recognition/margin, inventory and the H2 acceleration required by annual consensus are severe. Same-day broker commentary was unavailable. Resolver: H2 recognized storage revenue, storage margin and inventory conversion. Confidence: high on the miss; medium on whether it is timing or structural.

5.9 Zurich Airport: operating growth, capex dilution

Revenue, EBITDA and net income rose year over year but narrowly missed AWP. Zurich passenger growth and Brazil were constructive; Noida startup, higher depreciation/finance cost and a roughly flat FY EBITDA outlook dominated the debate. (H1 report, call transcript)

Bull cluster (medium): resilient premium travel, Brazil scale and long-term international optionality. Bear cluster (medium-high): Noida’s ramp, capex, leverage and dividend capacity create a lower near-term cash yield. Resolver: monthly Zurich traffic, Noida winter slots/passengers and 2027 EBITDA break-even. The selloff despite higher reported earnings supports the cash-yield critique.

5.10 Codelco: copper price masked production execution

Pre-tax profit surged with a realized copper price around 653¢/lb, but production fell 11% to 564kt and cash cost rose to 231.6¢/lb. El Teniente disruptions made the prior 1.33–1.36Mt annual target mathematically difficult; management did not restate it. (Reuters)

Bull cluster (medium): copper scarcity/pricing and productivity work support EBITDA/cash. Bear cluster (high): falling output, rising cost, capex/debt and operational disruption mean price is carrying the result. Consensus/disagreement/resolver: copper fundamentals are supportive; company execution is disputed. Monthly output, El Teniente normalization, cash cost and formal 2026 guidance resolve it.

5.11 BOCHK: credit-cost beat, capital-return ambiguity

Profit beat 3.5%, adjusted NIM widened and credit cost fell to 27 bp, while the impaired-loan ratio fell to 0.89%. The HK$10.5bn minimum additional-return plan was less than Goldman’s HK$12.7bn expectation and the special dividend yield was below UBS’s pre-event view. (results, SCMP)

Bull cluster (high): credit costs, CASA and excess capital are supportive. Bear cluster (medium): the top line was softer than statutory NII suggests, CRE risk remains elevated and capital-return phasing lacks specificity. Resolver: first 2027 special/regular DPS and H2 NIM/credit cost. Reaction awaits Aug. 31.

5.12 China Merchants Bank: wealth recovery, card stress

NII grew 5.6%, wealth fees 26.5% and retail AUM 8%; corporate loans grew while retail loans contracted. The NPL ratio stayed 0.94%, but consumer/card NPL formation worsened and expected credit losses rose 18.5%. (filing)

Bull cluster (medium-high): deposit funding advantage plus capital-market-sensitive wealth recovery. Bear cluster (high): falling cards, weak card fees and rising unsecured-retail stress validate Zhu’s adverse-selection warning. Resolver: Aug. 31 Q&A on NIM cadence, card vintages, collections and dividend. Friday prices were pre-print.

5.13 Montage: DDR5 earns today; CXL remains option value

Q2 revenue beat the public proxy 5.9%; interconnect revenue rose 28%, but H1 headline profit was inflated by RMB682m investment/fair-value gains and recurring profit grew 21%. Newer MRCD/MDB, PCIe Retimer, CKD and CXL products were roughly flat sequentially. (filing)

Bull cluster (medium-high): leading DDR5 generations and rising YoY interconnect margin. Bear cluster (medium): lower sequential margin and no new-product acceleration yet. Resolver: Sept. 8 briefing, Q3 new-product revenue and paid CXL qualification. Public post-print sell-side notes were unavailable.

5.14 Wuliangye: flagship stabilization, cash-quality veto

H1 profit landed almost exactly at the bottom of guidance, while Q2 revenue fell 13%, selling expense reached 47.5% of revenue and H1 operating cash flow turned negative. Flagship volume grew on lower average revenue per tonne; mass-market liquor collapsed. (filing)

Bull cluster (medium): China Galaxy and Xiao Zhuqing see flagship price stabilization near RMB800, inventory clearing and younger management execution. Bear cluster (high): Feikan Finance and Chai Jun emphasize promotions, cash, distributor economics and price-for-volume. Resolver: Mid-Autumn/National Day sell-through, flagship price, contract liabilities and positive cash flow. This is company normalization, not a broad China-consumer turn.

5.15 BOE: recurring earnings matter more than the preannouncement headline

Headline profit met the preannounced range; recurring profit missed the floor 2.3%. Q2 gross margin fell sequentially and fair-value gains supplied nearly one-third of pretax profit. OLED/MLED and lower capex were offsets; TV panel prices remained soft. (filing, TrendForce panel check)

Bull cluster (medium): OLED/MLED mix and simple free cash flow improve durability. Bear cluster (medium-high): display sales contracted, margin faded sequentially and marks inflated profit. Resolver: Q3 panel/OLED pricing, Gen-8.6 utilization, recurring profit and MLED margin. Aug. 31 is first price test.

5.16 COSCO Shipping Holdings: volume recovery versus rate/cost normalization

Q2 container volumes recovered and port profit improved, but operating profit remained below the year-ago level as freight economics normalized. Hormuz/Red Sea disruption supports ton-miles/rates but also raises bunker, insurance and network costs. (company filing, operating recap)

Bull cluster (medium): volume/share, port earnings and H1 delivery versus full-year profit aggregates. Bear cluster (medium): fleet-cycle risk and rate/margin erosion. Resolver: Q3 spot/contract rates, bunker/insurance costs, capacity deliveries and the duration of Hormuz disruption. No dependable public quarterly consensus or fresh named note was available.

5.17 Yankuang Energy: genuine Q2 improvement inside a one-off-heavy H1

CAS H1 profit rose 48%, but recurring profit rose only 4% because the Xintai sale added RMB2.84bn. Coal price/mix and chemicals improved; output fell and unit cost rose. (filing)

Bull cluster (medium): inferred Q2 recurring acceleration, Australian volume and tighter Chinese supply. Bear cluster (medium-high): one-offs, cash conversion, cost inflation and a back-loaded production/capex plan. Resolver: Aug. 31 Q&A, H2 output/unit cost, Kestrel funding and Q3 recurring profit. Public post-print named analyst opinion was unavailable.

5.18 China Shenhua: integrated resilience, weaker core coal economics

H1 revenue grew 8% and profit 2%; output fell, purchased-coal sales rose 33%, realized price rose and self-produced cost fell. Core coal and power PBT declined; chemicals swung strongly positive. The interim payout was 74%. (filing)

Bull cluster (medium-high): integration, rail/chemicals and payout. Bear cluster (medium): acquired perimeter, purchased-coal mix and declining core segment PBT. Resolver: Sept. 4 guidance on organic contribution, acquired assets, costs and capital returns. Friday’s closes were pre-print.

5.19 Weichai Power: selective industrial strength, not a broad capex boom

The print combined engine/AI-data-center exposure with softer heavy-truck demand and a still-important KION/European execution bridge. Headline acceleration included non-core items; public same-day consensus and analyst commentary were sparse. (HKEX issuer feed)

Bull cluster (medium): powertrain mix, data-center engines and KION recovery. Bear cluster (medium): truck cyclicality, Europe and cash conversion. Resolver: H2 KION margin/order intake, heavy-truck sales and core operating cash. The macro read-through is two-speed industrial demand.

5.20 Three-Circle: MLCC/optical demand with a valuation and cash test

Profit landed near the upper half of guidance and margins expanded, consistent with MLCC utilization/pricing and optical demand. Cash conversion lagged, working capital expanded and valuation was elevated. (company filing feed)

Bull cluster (medium-high): Changjiang’s Hu Shiyu expects Q3 volume/price acceleration from new capacity. Bear cluster (medium): aggressive broker forecasts are not yet supported by product-level disclosure or cash conversion. Resolver: Q3 MLCC price/volume, gross margin, receivable days and project conversion without inventory buildup.

5.21 CITIC Limited: market-sensitive recovery, not household reflation

H1 profit rose 8.1% and dividend 5%; financial services and advanced materials supplied much of the acceleration, while impairments, property and consumer weakness constrained breadth. (results feed)

Bull cluster (medium-high): brokerage cycle, bank margin, commodities and dividend. Bear cluster (medium): concentration in market-sensitive earnings and continuing property impairments. Resolver: Sept. 4 Q&A, securities revenue, bank credit cost, metal volume/price and disposal cash. The 0.70% move was only a partial reaction because subsidiaries had already reported.

6. Cross-event themes and notable contradictions

6.1 The policy constraint tightened while the growth mosaic softened

Warsh’s diagnosis was full employment, resilient private demand and insufficient inflation progress. Chicago PMI, French GDP and Michigan sentiment supplied a weaker growth signal; the BLS benchmark showed softer private employment than the total revision; Canada’s solid Q2 ended with flat July. The contradiction is real but not logically inconsistent: central banks can tighten into slower growth if inflation expectations, services and financial conditions remain the binding variables. The trade is therefore not simply “short duration because growth is strong”; it is “short front-end optionality while recession hedges remain relevant.” September U.S. jobs/CPI, national ISM and the FOMC resolve the tension.

6.2 Headline inflation and underlying inflation diverged across Europe and Japan

Spain’s 4.3% national CPI and France’s 2.4% CPI were energy-led; Spanish core eased. Tokyo’s rebased history softened, yet services/rents and subsidy effects kept the BOJ tightening case alive. The common thread is that energy shocks can raise headline prints without proving domestic overheating, but persistence in services/wages would still force a policy response. This is why European sovereign/fiscal risk and Japanese rate normalization can coexist with weak real demand.

6.3 Deposit repricing repaired bank income before household demand repaired

China’s large banks showed a genuine NIM floor and NII rebound, but retail lending was flat/down at several institutions, card fees were weak, consumer/mortgage NPLs rose and impairments increased. China Merchants Bank’s wealth fees recovered faster than card activity. The apparent contradiction—better bank earnings with weak households—is resolved by cheaper deposits, policy/corporate lending, bond books and capital-market income. It is not a broad household-reflation signal.

6.4 AI demand remained strong; capital capture split sharply

Victory Giant’s order book and Montage’s DDR5 growth showed that AI infrastructure demand is real. Victory Giant still fell 8.15% because gross margin, recurring profit and capex economics deteriorated; Montage’s newer CXL/retimer products had not accelerated sequentially and investment marks inflated headline profit. BOE’s AI-device narrative also lacked measurable AI revenue. The signal is consistent with the prior day’s AI results: demand clears the bar, but margin, free cash flow and the burden of capacity decide the equity outcome.

6.5 Geopolitics raised shipping earnings while normalization threatened their duration

Hormuz mine clearance and higher escorted flows pushed the physical market away from near-blockade conditions. Yet Frontline and Hafnia explained how waiting, dark fleets, ship-to-ship transfers, rerouting and inventory depletion created extraordinary TCEs. Their records can coexist with softer crude because shipping scarcity depends on effective vessel supply and distance, not only barrels. Q4 fixtures and independently verified transit/insurance normalization resolve how fast the windfall fades.

6.6 Reported growth repeatedly failed the clean-earnings test

BYD’s revenue miss sat inside a margin rebound; MINISO’s revenue beat sat inside a profit/margin reset; Wuliangye’s profit surge sat on a restated low base and negative operating cash flow; BOE and Montage were helped by fair-value gains; Yankuang by an asset sale; Shenhua by acquired assets and purchased-coal mix; Codelco by copper price despite falling production. The most durable signal across the day is to reconstruct recurring earnings, cash and organic volume before accepting the headline.

6.7 Monday carries an unusually large Asia price-discovery backlog

Most A/H-share filings crossed after local closes while the U.S. session was open. BYD, all five mega-banks, BOCHK, CMB, Montage, Wuliangye, BOE, Sungrow, COSCO, Yankuang and Shenhua therefore lack a clean post-result cash close. Any report that describes Friday’s A/H move as “after earnings” for these companies is chronologically wrong. Aug. 31 price/volume and the pending calls are part of the thesis, not missing decoration.

7. Coverage audit

7.1 Calendars and source sets checked

• Macro calendars/news inventories: Federal Reserve event calendar and Jackson Hole agenda; BLS and BEA release schedules; MarketScreener/Finwire global calendar; Trading Economics; Investing.com; Reuters/AP daily market coverage; Statistics Canada The Daily; Japan Statistics Bureau and Ministry of Finance schedules; INSEE; INE; euro-area national release calendars; Baker Hughes schedule.
• Earnings calendars: TipRanks global Aug. 28 calendar; MarketScreener global/emerging-market company calendars; Investing.com earnings calendar; Yahoo Finance U.S. calendar; Kiplinger/Briefing.com U.S. weekly calendar; HKEX result-announcement feed; SSE/SZSE/CNINFO filing feeds; company investor-relations calendars.
• Primary company sources: official releases, HKEX/SSE/SZSE/CNINFO filings, SEC 6-Ks where applicable, presentations, webcast replays and transcripts. Secondary consensus/reaction sources were used only when the definition and timestamp could be reconciled.
• Cross-check rule: a board date or calendar entry was not treated as a result until a public filing/release appeared. A close was not treated as a reaction unless the filing preceded that market’s trading window.

7.2 Borderline macro items excluded

• Germany unemployment: modest surprise, no material isolated U.S./global asset response and little incremental information beyond the larger Europe growth/fiscal set.
• Euro-area final confidence/business surveys, Norway/Sweden retail and unemployment, Finland GDP/trade, Portugal/Belgium/Italy confidence/orders: scheduled but no material U.S./global market reaction or unique thesis after cross-check.
• Kansas City services survey: lower-tier regional information, dominated by Chicago PMI and Warsh; no isolated market impact.
• BLS county employment/wages and state benchmark detail: useful reference data, but not a market-moving event separate from the national preliminary CES benchmark.
• Baker Hughes rig count: no material oil move attributable to the release; Hormuz and Warsh dominated energy pricing.
• India GDP: calendar pages showed Aug. 31, not Aug. 28; excluded as wrong date.
• Proposed U.S. food-processing legal order / meat comments: no executed order by cutoff and no meaningful Tyson/sector move; excluded as rhetoric without a market event.

7.3 Earnings candidates excluded or deferred

• Air China: Aug. 28 was board approval only; no H1 result was public by cutoff. Official filing and management Q&A are scheduled Aug. 31. The July profit warning (RMB2.1bn–2.6bn H1 loss) is not a new Aug. 28 result. (profit warning)
• China Resources Land: calendar/board date was Aug. 28, but no HKEX H1 filing was public by cutoff; expected results and the Aug. 31 briefing remain pending.
• Laurentian Bank: reported during a pending acquisition, held no call and produced no important sector move/read-through; below the materiality bar.
• Nordic American Tankers: Q2 release was dated Aug. 27, so it belongs to the prior trading-date report; no separate major Aug. 28 call catalyst was found. (company release)
• Haier Smart Home and Harmony Gold: calendar aggregators showed Aug. 28 in places, but primary releases were Aug. 27 and were already eligible for the prior report.
• Guangdong Investment and smaller HK interim filers: checked against HKEX/market-cap and price-move screens; no major index/sector move or unique global read-through at the cutoff. The HKEX feed contained more than 1,000 small interim filings; inclusion was based on size, realized move or material sector/macro signal, not filing count.

7.4 Calls, transcripts and analyst notes not yet available

• Calls scheduled after cutoff: China Merchants Bank and Yankuang on Aug. 31; Air China and China Resources Land result briefings Aug. 31; China Shenhua and CITIC Sept. 4; Montage Sept. 8.
• No same-day earnings call/transcript located: BYD, Wuliangye, BOE, Sungrow, China Shenhua, Montage, CITIC and Codelco. For BOCHK, the briefing occurred but a searchable analyst Q&A transcript was not public by cutoff.
• Machine/AI-assisted transcript caution: public transcripts for several Chinese banks were checked against filings; filing figures control where transcript summaries contained errors.
• Analyst-note scarcity: fresh public, named post-print sell-side notes were especially sparse for the after-close Chinese filings. Sector-level Fitch/Moody’s views and clearly labeled media/expert commentary were used only when company-specific notes were unavailable. Paywalled research was not represented as accessed.

7.5 Material data gaps

• First clean Aug. 31 price reaction for most Chinese A/H-share releases.
• Daily allocation of Japan’s ¥15.4tn intervention until MOF’s quarterly detail; scope of any U.S. Treasury/FIMA use.
• Independently audited Hormuz loaded/discharged volumes, war-risk insurance and signed corridor terms.
• Reliable public line-item consensus for many Chinese H1 filings; vendor EPS/revenue definitions often mixed cumulative, adjusted and IFRS/CAS figures.
• Company-specific Q&A on BYD’s volume target/battery constraint; Sungrow storage volume/recognition; BOE panel/OLED utilization; Codelco’s formal full-year production target.

Bottom line

The day’s highest-confidence signal is that Warsh restored a live tightening distribution while offering less guidance, and the market responded most clearly in the two-year yield, dollar, small caps, gold and bitcoin. The counter-signal is a softer growth mosaic, not a collapse. In companies, deposit repricing, exports, AI demand, tanker scarcity and commodity prices produced strong headlines; recurring profit, cash conversion, credit quality and capex determined whether those headlines deserved to be owned. The most important unresolved evidence arrives immediately: Monday’s Asian price discovery, then U.S. payrolls/CPI and the September Fed/BOJ decisions.