70 Years of Crashes: early indicators, lead times, and false alarms across the major economies

Macro Crash / Recession Early-Warning System — Evidence Base

Date context: late October 2026. Research run 2026-10-27 (America/Chicago).

How to read this report. “Hit rate / false positives” are retrospective reconstructions unless a primary publisher (Fed, Conference Board, FRED) is named — the samples are tiny (7–12 recessions), definitions of “a signal” vary by author, and several Oct-2026 aggregators that republish FRED series disagree by a few bp / a few days. Where sources conflict, both values are shown. Analyst opinion is labelled [opinion]. Figures I could not verify are flagged UNVERIFIED — none are estimated.

Headline conclusion (Part 3 in one line): As of 23–27 Oct 2026 the classic recession cluster (Sahm, claims, ISM, LEI, SLOOS, broad HY spreads) is quiet in the US, while the crash/fragility cluster is flashing: an exogenous oil shock (Brent ≈ $116), long yields at multi-decade highs (US 10-yr ≈ 5.0%, UK 10-yr 5.33%, France 10-yr ≈ 5%), CAPE > 40, record margin debt, CCC-only credit stress, a collapsing US housing pipeline, and Eurozone fragmentation centred on France (OAT–Bund ≈ 160–178 bp, widest since 2011/12). Historically that combination — supply-shock + bear-steepening long yields + extreme valuation/leverage, without labour/credit confirmation — matches 1973 and 1990 (shock into tightening) and 1987 (valuation + rates, no recession) better than it matches 2007 (credit-led). The system should be at AMBER for crash, GREEN-AMBER for US recession, RED-AMBER for France/Eurozone sovereign stress.

1. Ranked Indicator Table

Ranking = historical reliability for the stated job (recession warning vs crash warning), weighted by lead time usefulness and false-positive record. Lead = to NBER recession start unless stated.

Design implication: no single row justifies action. The historically robust recession signal is a cluster: curve inversion (12+ mo prior) → re-steepening on Fed cuts (bull steepener) → SLOOS ≳40% + HY widening through 500 bp → LEI 6-mo decline with diffusion <50 → claims rising + Sahm approaching 0.50. The robust crash signal is different: extreme valuation/leverage (rows 10–11) + a rates or funding break (MOVE spike, basis/repo stress, forced-seller loop like portfolio insurance 1987 / LDI 2022 / carry unwind 2024).

2. Part 1 — Per-Episode Forensic Timelines

NBER reference (primary, via Federal Reserve FEDS Note): Aug 1957–Apr 1958; Dec 1969–Nov 1970; Nov 1973–Mar 1975; Jan–Jul 1980; Jul 1981–Nov 1982; Jul 1990–Mar 1991; Mar–Nov 2001; Dec 2007–Jun 2009; Feb–Apr 2020 — https://www.federalreserve.gov/econres/notes/feds-notes/the-restrained-recovery-of-state-and-local-government-payrolls-from-the-pandemic-recession-20230804.html

2.1 United States

1957–58 (Eisenhower). Broke: tight-money inventory/durables recession. Fed raised the discount rate to 3.5% in Aug 1957 — the month the recession (hindsight-dated) began; S&P 500 fell ~21% Jun/Jul–Oct 1957 (≈ −17 to −20% on monthly-average reconstructions); unemployment 4.1% (Aug 1957) → 7.5% (Jul 1958); industrial production −~14%. Observable before: rising rates, falling equities and IP from mid-1957 — contemporaries could see tightening, not the NBER date. Failed: CAPE was only modestly above average — valuation gave no warning; the speed (Q1 1958 GDP −2.6% q/q, among the sharpest post-war quarters) surprised forecasters. Sources: https://moneyweek.com/investments/stockmarkets/601152/what-can-the-1957-stockmarket-crash-teach-us-about-the-coronavirus ; https://eco3min.fr/en/every-sp500-crash-since-1950/

1969–70. Broke: Fed tightening against Vietnam-era inflation + go-go valuation excess. S&P −36.1%, 29 Nov 1968 → 26 May 1970 (monthly reconstructions differ: −29%). Yield curve (10Y–3M) inverted from Dec 1968 (monthly avg), ≈12 mo lead. Sahm triggered Mar 1970 — 3 mo after the Dec 1969 peak (coincident, as designed). Failed: the 1966 inversion overhang meant curve-watchers had already cried wolf once this cycle (see §2.4).

1973–74 oil shock. Timeline: S&P peaked 11 Jan 1973 (−48.2% to 3 Oct 1974); 10Y–3M inverted Jun 1973 (≈5 mo lead); OPEC embargo Oct 1973, oil ~$3 → ~$12 in six months; recession Nov 1973–Mar 1975; CPI 8.5% (1973) → 12.3% (1974); unemployment → 9%. Observable before: curve inversion, rising CPI already ~8% in early 1973, Nifty-Fifty valuation excess, Bretton Woods collapse (1971). What the shock did: it was a supply break no domestic indicator predicted — the curve/inflation cluster warned of vulnerability, not of the embargo. Failed: Fed initially accommodated (Burns) — policy reaction function, not data, was the miss. Sources: https://manninglive.com/premium/stacker/stories/every-major-stock-market-correction-since-1950-and-how-long-recovery-took,194774

1980–82 Volcker (double-dip). Timeline: Volcker took office Aug 1979; 6 Oct 1979 shift to reserve targeting; fed funds 11% → 20% (late 1980 / Jun 1981), 10-yr >15%, 30-yr mortgages >18%, prime 21.5%; recessions Jan–Jul 1980 and Jul 1981–Nov 1982; unemployment 10.8% Nov–Dec 1982; S&P bear −27% Nov 1980–Aug 1982. Curve inverted from Nov 1978 (10Y–3M episode spanning both recessions, ≈14 mo to the first). Mexico's Aug 1982 default = EM contagion from US rates (template for §Family 7). Failed: consensus gradualism — almost no forecaster priced 20% funds; equities bottomed Aug 1982 while unemployment was still rising (equities lead out, labour lags).

1987 crash — the key control. Timeline: market +44% in ~7 months into Aug 1987; rising US rates, falling dollar, trade-deficit shock; 19 Oct 1987 Dow −22.6% (508 pts), S&P −20.5% in a day, −33.5% Aug–Dec. Mechanism: portfolio insurance — automatic futures selling as prices fell, liquidity/plumbing failure, no circuit breakers (introduced after). Outcome: no recession; GDP kept growing, Dow regained its high in <2 years; Fed supplied liquidity. Indicators before: valuation elevated and rates rising — fragility yes; LEI/labour/credit gave no recession signal, correctly. Lesson encoded in the system: a crash without credit/labour confirmation mean-reverts; do not convert an equity break alone into a recession call. Sources: https://www.federalreserve.gov — via secondary summaries https://legalclarity.org/what-is-a-market-crash-definition-causes-and-examples/ ; https://www.nasdaq.com/articles/25-years-later-lessons-crash-87-2012-10-19

1990–91. Broke: S&L crisis + 1989–90 credit crunch, commercial-real-estate bust, then a confidence shock. Curve inverted Jan–Sep 1989 (≈18 mo lead on 2s10s; first disinversion Jun 1989, recession 3 mo after the Mar 1990 re-steepening per Saxo reconstruction); Fed funds peaked 9.75%; S&P −19.9% Jul–Oct 1990; Iraq invaded Kuwait 2 Aug 1990 — after the Jul 1990 peak — oil doubled in ~2 months ($~15 → $41.15 intraday Sep 1990); CPI peaked 6.3% Oct 1990; unemployment kept rising to 7.8% (Jun 1992) — the original “jobless recovery”. S&L clean-up cost ≈ $153bn (≈2.6% of GDP). Failed: oil was blamed contemporaneously but followed the peak — a confound the system must code correctly (pre-peak signals only). Sources: https://Www.Home.saxo/content/articles/bonds/yield-curve-is-disinverting-lessons-from-past-crises-09082024 ; https://eco3min.fr/en/us-recession-history-since-1948/

2000–02 dot-com. Timeline: Fed 4.75% → 6.5% (Jun 1999–May 2000); CAPE 44.19 (Dec 1999) — all-time high; margin debt +80% in the year to Mar 2000 (~$300bn); 457 IPOs in 1999, avg first-day gain 71%; 2s10s inverted Feb 2000 (~13 mo lead), 10Y–3M Jul 2000 (~8 mo); Nasdaq peaked 5,048.62 on 10 Mar 2000 → 1,114 on 9 Oct 2002 (−78%); S&P −49%; recession only Mar–Nov 2001 (−0.3% GDP — an investment bust, not a consumer bust); HY OAS, having crossed 500 bp in 1998/2000, exceeded 1,100 bp in early 2002 as Enron/WorldCom credit frauds extended the bear after the recession ended; unemployment peaked later (6.3%, Jun 2003). Failed: earnings-based valuation (ex-items) rationalised the bubble in real time; equity breadth narrowed for ~2 years before the peak — a fragility gauge, not a timer. Sources: https://www.fool.com/investing/2026/10/25/witnessing-stock-market-do-something-3rd-time-in-100-years-history-crystal-clear-whats-next-for-stocks/ ; https://pitchbook.com/news/articles/high-yield-spread-500-bps-barrier-warning-sign

2007–09 GFC. Timeline (all publicly observable in real time): Fed 1% → 5.25% (Jun 2004–Jun 2006); Case-Shiller national +~124% 2000–06, peak Jul 2006; curve inverted 2006 (10Y–3M sustained Jul 2006, ~16 mo lead; 2s10s Jan/Feb 2006, ~22 mo); SLOOS tightening through 2007; HY OAS in the 200s in H1 2007 → through 500 bp Nov 2007 → peak Dec 2008 (2,182 bp on Pitchbook's series; FRED BAMLH0A0HYM2 all-time high cited as 19.88 pp on 16 Dec 2008 — series revisions/method differ; do not mix); Bear Stearns hedge funds Jun 2007; BNP Paribas fund freeze 9 Aug 2007 (LIBOR–OIS blew out); Northern Rock Sep 2007; S&P peaked Oct 2007; recession Dec 2007; Bear rescue Mar 2008; Lehman 15 Sep 2008; S&P −57% to 9 Mar 2009; unemployment 10.0% Oct 2009; Case-Shiller ultimately −33% national to Feb 2012. What worked: housing permits/sales (2005–06), curve, SLOOS, ABX/subprime delinquency, bank CDS/LIBOR–OIS — a full cluster, 12–18 mo of warning. What failed: rating-agency models, “contained to subprime” official commentary (spring 2007), and equity indices, which made highs 15 months after house prices peaked. Sources: https://docslib.org/download/4224159/charting-the-financial-crisis (Fed “Charting the Financial Crisis”); https://maseconomics.com/the-2008-financial-crisis-explained-causes-response-and-lessons-for-the-global-economy/

2020 COVID. Timeline: 2019 inversion first (see §2.4); S&P −34% 19 Feb → 23 Mar 2020 — fastest 30%+ fall on record; VIX close 82.69 on 16 Mar 2020; IG spreads +~250 bp in 3 weeks; Treasury-market dysfunction mid-Mar; WTI futures negative (−$37) 20 Apr 2020; recession Feb–Apr 2020 (2 months); unemployment 14.7% Apr 2020; Sahm spiked to 4.00 (Apr 2020 print). Fed to zero + unlimited QE + CARES ($2.2tn) → new S&P high by 18 Aug 2020. System lesson: exogenous shocks are not forecastable by the playbook; what the playbook did show pre-shock was late-cycle fragility (inversion, tight spreads, high leverage) that amplified the fall, and the coincident gauges (VIX backwardation, dollar spike, Treasury illiquidity) correctly identified panic in the event. Duration of signal ≠ duration of recession when policy responds at that scale.

2022 bond/equity bear (no recession). Timeline: CAPE crested 40 in the first week of Jan 2022; CPI → 9.1% (Jun 2022); Fed +425 bp in 2022 (fastest since Volcker); S&P −25% 3 Jan → 12 Oct 2022, Nasdaq ≈ −35% peak-to-trough; simultaneous bond bear — “worst bond market in US history” by several measures; 60/40 failed; H1 2022 printed two negative GDP quarters but NBER declared no recession (employment/income held). Curve inverted Jul/Oct 2022 → the false positive in §2.4. Failed: the 2-quarter GDP rule and the curve both “called” a recession that the broader NBER data never confirmed; HY widened but SLOOS/labour never reached recession cluster thresholds.

2.2 Non-US episodes

Japan 1990–92. BoJ held 2.5% after the Plaza Accord; Nikkei 38,915.87 on 29 Dec 1989 (P/E >70×, Japan ≈ half of world market cap); urban land +~200% 1985–90 in the six big cities; BoJ 2.5% → 6.0% (1989–Aug 1990) + MoF real-estate lending caps (Mar 1990); Nikkei −39% in 1990, −~60% by 1992 (below 17,000), land fell for a decade+; NPLs ≈8% of GDP by the late 1990s, jusen/Yamaichi failures 1995–97; deflation and zero rates (1999), QE (2001). Nikkei did not regain 38,915 until Feb 2024. Observable before: valuation, credit growth, land-price/ income ratios — all extreme for years; policy tightening was the trigger. Failed: “Japan is different” / keiretsu cross-holdings (≈20–30% stable shareholders) suppressed float and sell signals — breadth/liquidity gauges understated fragility. Sources: https://ccab.mkirsanov.com/speculative-bubbles/ ; https://grokipedia.com/page/japan_and_the_world_economy

UK ERM — Black Wednesday, 16 Sep 1992. Entered Oct 1990 at DM 2.95 (floor ≈2.78) into recession, while post-reunification Bundesbank rates stayed high. On the day: Bank intervention (Treasury cost ultimately ≈£3.3bn), base rate 10% → 12%, a rise to 15% announced for the next day and rescinded on exit that evening; sterling $1.8937 → $1.7435 in four days (BoE series). Recovery followed exit (rates to 6% by Jan 1993, inflation targeting from Oct 1992). Template: a peg/FX level inconsistent with domestic rates is itself the indicator — reserves burn + forced rate rises into weakness = break imminent. Sources: https://www.economicsobservatory.com/the-birth-of-inflation-targeting-why-did-the-erm-crisis-happen ; https://aitimeline.in/black-wednesday-uk-erm-exit-timeline-15691/

UK 2008. UK was a GFC transmission case (Northern Rock — first UK bank run since 1866, Sep 2007; RBS/HBOS rescues Oct 2008); UK unemployment 5.2% → 8.1% (2011). The early-warning content duplicates §GFC, plus a UK-specific gauge: wholesale-funding dependence of mortgage banks (Northern Rock's model was public).

UK gilt / LDI crisis, Sep–Oct 2022. Timeline: 23 Sep 2022 “mini-budget” — ~£45bn unfunded tax cuts, no OBR forecast; 30-yr gilt ≈3.6% → >5% within days (+>100 bp in a few sessions, +127 bp in one session per one reconstruction); sterling record low $1.035 on 26 Sep; leveraged LDI (used by ~60% of DB schemes) faced collateral calls → forced gilt sales → yields up → more calls; 28 Sep BoE announced up to £65bn temporary long-gilt purchases (bought ≈£19.3bn); Kwarteng sacked 14 Oct; Truss resigned 20 Oct (45 days). No recession was caused by the gilt crisis itself — it is the modern template for speed, not level: a >100 bp long-yield move in days breaks leveraged hedges regardless of solvency. Sources: https://uk-debt.info/bond-market-gilts ; https://www.ipe.com/analysis/rising-gilt-yields-trigger-deja-vu-for-uk-db-industry/10132881.article

Eurozone sovereign crisis 2010–12. Greece admitted deficit misreporting (2009/10); Greek 10-yr ~5% → ~30%+; Ireland bailout Nov 2010, Portugal May 2011; contagion to Italy/Spain by mid-2011 (Italian spreads ~150 → >500 bp within months per Monti recollection); bank–sovereign doom loop (domestic banks holding domestic sovereign debt, no fiscal union, no sovereign lender of last resort). Turning point was political, not data: Draghi “whatever it takes” 26 Jul 2012 and OMT (6 Sep 2012, never activated) compressed Spanish/Italian spreads by ~50–60 bp on announcement alone (ECB research). Failed before: pre-2008 spread convergence (≈0 bp for Greece vs Bunds) — the market priced redenomination/default risk at zero; debt/GDP and current-account divergences were public for years and ignored. Sources: https://www.ecb.de/pub/pdf/scpwps/ecbwp1707.pdf?938b7d20c8b82b5450f2a9ac5de68beb ; https://cnonsensus.com/eurozone.html

Asia 1997–98 → Russia/LTCM 1998 (contagion + liquidity template). Thailand floated the baht 2 Jul 1997 after reserves were drained defending the peg; contagion to Indonesia, Malaysia, Korea; IMF/World Bank packages >$100bn. Pre-signals, all public: short-term foreign-currency debt > reserves, pegged FX + open capital accounts, property/equity booms, deteriorating current accounts. Russia defaulted/devalued Aug 1998; LTCM — notionally >$1tn, equity ~$4.8bn — lost >$4bn in months; NY Fed brokered a $3.5bn private recapitalisation (Sep 1998) and the Fed cut three times (Sep–Nov 1998, incl. an inter-meeting cut). US equities fell ~19% then recovered by Dec 1998 — no US recession. Failed: US curve watchers: 2s10s briefly inverted Jun 1998 and HY crossed 500 bp — a genuine global-liquidity signal misread if scored as a US-recession signal. This is the archetype for scoring funding/liquidity indicators on a global stress axis, separate from the recession axis. Sources: https://www.jbic.go.jp/en/information/today/today_202401/jtd_202401_column1.html ; https://froggyweb.com/2023/03/25/factbox-the-biggest-financial-crises-of-the-last-four-decades/

China equity crash 2015. Shanghai Composite +~150% in 12 months to 5,166–5,178 on 12 Jun 2015 while GDP growth slowed to a 25-year low; official margin debt RMB 2.2tn (~$355bn; margin financing +5× in a year to ~$434bn on another measure incl. shadow/umbrella trusts) — among the highest leverage-to-market-cap ever recorded; CSRC capped broker margin at 4× net capital and probed off-exchange leverage on 12–13 Jun; index −~30% in 3 weeks (~$3tn erased), >1,400 companies (>half the market) suspended trading by 8–9 Jul; second leg −8.48% on 24 Aug 2015 (“Black Monday”) hit global markets. Yuan devaluation (11 Aug 2015) compounded it. Template: margin-debt growth rate + retail froth + regulator tightening = trigger; state buying delayed, did not prevent, the fall. Sources: https://api.finexus.net/api/news/events/77a97eff-bd05-4745-9a2b-1fd7c76c6d94/html ; https://www.rstreet.org/commentary/15-sheep-explain-the-chinese-stock-market-crash/

China property downturn 2021– . “Three red lines” (Aug 2020) cut developer refinancing (liabilities/assets <70%, net gearing <100%, cash/short-term debt >1×); Evergrande (~$300bn liabilities) warned Guangdong of a cash crunch Aug 2021, missed payments, Fitch “restricted default” on offshore debt Dec 2021; contagion to Kaisa, Fantasia, Sunac, Sinic, Modern Land, then Country Garden's offshore default (2023). Annual property sales roughly halved from ~RMB 18.2tn (2021) to ~RMB 9tn (projected 2025, JHU/BIPR summary of Reuters polls); S&P estimated 762m m² of completed unsold housing; home-price falls continued into 2025–26 (see Part 3). Unlike 2008-US, transmission ran through local-government land finance, presale completion risk and household wealth — a slow balance-sheet recession, not a bank run; Western bank/credit gauges (CDS, interbank) stayed quiet throughout, which is itself the lesson: use China-specific series (property sales/investment, land-sale revenue, credit impulse, developer offshore spreads), not US templates. Sources: https://bipr.jhu.edu/documents/2026-05-11-BIPRBlogMarcelKolm.pdf ; https://sherborneinvestmentclub.com/Market%20Report/sicmarketreportchinahousingburstapril2026refinedgapfix.pdf

2.3 (cross-reference) UK/US 2022 are covered above; both are also false-positive controls for GDP and curve rules.

2.4 False-positive controls — inversions / clusters that did NOT deliver

Sources for this section: https://assetwhisper.com/yield-curve-inversion-recessions/ ; https://eco3min.fr/en/ny-fed-recession-probability-model-2/ ; https://eco3min.fr/en/sahm-rule-false-signals-history/

3. Part 2 — Indicator Playbook by Family

3.1 Yield curve & rates

• Series: FRED T10Y3M (academic/NY Fed standard), T10Y2Y (market standard, from Jun 1976), T10YIE breakevens, DFII10 real 10-yr, NY Fed recession-probability probit. A −82 bp 10Y–3M spread ≈ 50% 12-mo recession probability in the NY Fed model; −240 bp ≈ 90% (model estimates, pre-2022 sample).
• Pattern that mattered: sustained (≥2 months of negative monthly averages) inversion, then re-steepening. Recessions typically began after disinversion, and the dangerous form is the bull steepener (Fed cutting into weakness, front end falls fastest). The 2024 disinversion was a bear steepener (long yields rose) — historically the benign form — consistent with no recession following. In Oct 2026 the curve is compressing from a small positive peak (+41 bp on 8 Sep 2026 per one reconstruction) — that source's “compression-after-peak” recession claim is [opinion, single-source, do not encode as a rule without independent validation].
• Type: leading (12–18 mo, highly variable). See ranked table rows 1–2.

3.2 Credit spreads & lending standards

• Series: ICE BofA HY OAS BAMLH0A0HYM2, IG OAS BAMLC0A0CM, CCC OAS BAMLH0A3HYC; Gilchrist–Zakrajšek excess bond premium (Fed FEDS); SLOOS net tightening (Fed, quarterly, ~1-mo lag).
• Thresholds: HY <350 bp complacent / 350–500 elevated / >500 stress / >800 panic-zone historically associated with recession underway; change matters: +300 bp from a 12-mo low is a strong deterioration signal in secondary research. CCC–broad divergence (CCC blowing out while broad HY stays tight) = late-cycle idiosyncratic stress, seen in Oct 2026 (see Part 3) and pre-2007 in single names before it generalised.
• Type: spreads = leading-to-coincident (lead equities ~months, lag the curve); SLOOS = leading for credit growth (~12 mo) but with the 2023 false positive scored.

3.3 Labour & broad economy

• Sahm (row 3): coincident onset detector; always pair with claims (row 7) and payrolls breadth/temp help. A Sahm rise driven by participation with claims flat (2024, and again not present in Oct 2026) should be down-weighted.
• LEI (row 4): use the CB's own 3Ds rule (6-mo change depth + diffusion <50) rather than any single monthly fall; LEI leads peaks by ~7 mo on CB's current methodology page (verified in the Sep 2026 release).
• ISM (row 8): new orders <50 sustained + headline <48 = manufacturing recession; services ISM matters more for the modern US economy; ISM prices is an inflation gauge (77.9 in Sep 2026 — see Part 3), not a recession gauge.

3.4 Equity valuation / breadth / froth

• CAPE, forward P/E, equity risk premium (excess CAPE yield ≈0.84 pp in Oct 2026 — thin), Buffett indicator, price/sales, concentration (top-10 weight — 37.6% Jul 2026, above the ~27% top-10 share at the 2000 peak on secondary measures), equal-weight vs cap-weight divergence. All are fragility/condition gauges with zero timing content; their job in the system is position-sizing / hedge-budget, not alarms. Breadth deterioration preceded 2000 and 2007 peaks by many months and also occurred in non-events (2023–26 narrow leadership persisted for years).

3.5 Leverage & funding liquidity

• FINRA margin debt (row 11; monthly, ~3-week lag — stale by construction); leveraged-ETF and 0DTE volumes are not in FINRA margin and are a known blind spot [analyst caveat]; repo: NY Fed SOFR vs IORB and Standing Repo Facility take-up (Sep 2019 repo spike is the template: SOFR 5.25% intraday stress with no recession); MOVE (row 12); VIX backwardation ≥3 days = coincident panic confirmer; dollar: broad dollar / DXY spike + widening cross-currency basis (EUR/JPY basis) = global dollar-funding stress (Mar 2020 template). Current cross-currency basis level: UNVERIFIED in this run — no dated public source found; the system should source it from BIS/central-bank publications, not blogs.

3.6 Housing & bank stress

• Permits (leading, in LEI), NAHB HMI, pending sales (leads closings 1–2 mo), starts; prices (Case-Shiller) lag ~6–12 mo; delinquency (Freddie/MBA) lags further and grades bank loss content: 2007-style stress requires delinquency + bank CDS + SLOOS tightening together. Bank stress gauges: regional-bank equities/CDS, unrealised securities losses (2023 SVB template — duration, not credit), CRE exposure (office), and in Oct 2026 additionally bank loans to non-bank financial institutions (H.8: ~$2.10tn, ~14.8% of bank loans, +21.5% YoY per one 18 Oct 2026 research summary — [single-source, verify against H.8 directly before operational use]).

3.7 Global

• China: credit impulse (change in credit flow as % GDP; NY Fed research notes the impulse dipped negative in late 2024 and turned up through Q1 2025, with smaller cycles than pre-2020); total social financing growth; NBS property investment/sales/prices (70-city); developer offshore HY spreads. Credit impulse historically leads Chinese/global industrial activity by ~9–12 mo in vendor research [label: analyst convention, not a settled estimate].
• Japan / carry: BoJ policy rate path vs Fed, USDJPY level and speed, CFTC leveraged-fund yen shorts, 10-yr/30-yr JGB yields, Japanese investors' foreign bond holdings (¥768.7tn at end-2025 and $1.07tn of Treasuries per one Oct 2026 analysis — [single-source]). Template event: 5 Aug 2024 (BoJ hike 31 Jul + soft US payrolls 2 Aug → Nikkei −12.4% in a day, VIX intraday extreme, S&P −8.5% peak-to-trough, resolved in ~a week because positioning, not solvency, was the problem).
• Eurozone: OAT–Bund and BTP–Bund 10-yr spreads (see rows 14), ECB TPI/OMT backstop credibility, French fiscal path. The doom loop is bank holdings of domestic sovereigns — watch EZ bank equities alongside spreads.
• EM dollar debt: JPM EMBI GD spread; frontier spreads ≳1,000 bp = distressed cohort (2022 template: Sri Lanka/Argentina/Egypt). A strong dollar + US 10-yr >5% is the classic tightening of EM financial conditions even when spreads have not yet moved (Sep–Oct 2026 configuration — managers trimming EM HY while spreads sit at 2007 tights, per Bloomberg reporting).

3.8 Commodities as signal

• Oil spike (supply) = stagflation impulse: raises headline CPI within 1–2 months, forces central banks hawkish, squeezes real incomes — 1973, 1979, 1990, 2022, and Oct 2026. Oil collapse (demand) = coincident recession confirmation (2008 H2, 2020). Distinguish by cause: if copper is rising with oil, it is more likely demand; if oil rises while copper/gold and equities fall, it is supply/geopolitical.
• Copper/gold: falling ratio = growth pessimism vs fear bid. Handle with care in Oct 2026: copper $6.62–6.66/lb (+~28% YoY, near 12-mo highs — electrification/AI demand is a structural bid that contaminates the cyclical signal) while gold >$4,000 reflects real-rate/fiscal/geopolitical hedging, so the ratio's message is ambiguous by construction right now.

4. Part 3 — Current Baseline, Late October 2026

All values dated and sourced; “verified primary” = the publisher's own release was fetched in this run. Others are index reads of the named publisher's data via reputable republication — check the primary series before acting.

4.1 United States — baseline table

US risk reading: Recession families quiet (Sahm 0.00, claims <200k, ISM 54.5, LEI flat-to-up, SLOOS neutral, NFCI loose, broad HY 338 bp). Crash families flashing (CAPE >40, record margin debt, CCC distress, MOVE >100, housing pipeline collapse, oil >$115, 10-yr ~5% with a December hike priced). Overall: AMBER — a valuation/rates accident is plausible at any time; a textbook recession needs labour or broad credit to break, and neither has. The two tripwires that would upgrade to RED: (1) HY OAS through 500 bp with SLOOS re-tightening, (2) claims 4-wk >260–300k and Sahm >0.30 rising.

4.2 Eurozone — highest sovereign-stress reading in the developed world

• Flash composite PMI 51.9 Oct (Sep 53.1, 4-mo low); France 48.4 (contraction), Germany 52.8; France business confidence lowest since 2012 outside COVID; hiring near standstill. Verified via Reuters text fetch: https://www.reuters.com/world/europe/euro-zone-growth-slows-4-month-low-amid-rising-costs-french-turmoil-pmi-shows-2026-10-23/
• HICP 3.8% Sep (Aug 3.2%), energy +18.9%, core 2.5% (Eurostat, 16 Oct, via https://bullquill.com/ecb-october-2026-meeting-rate-decision-preview/). ECB deposit rate 2.50%, hold expected 29–30 Oct, Reuters poll expects +25 bp in Dec.
• Bund 10-yr 3.36% (26 Oct); France OAT 10-yr 4.96% (26 Oct) after a 5.14% peak the prior week — highest since 2002; OAT–Bund 160 bp on 26 Oct after 178 bp intraw-week — widest since 2011/12; Italy BTP 4.49%, BTP–Bund 113 bp — note France, not Italy, is now the spread problem, inverting the 2011 hierarchy (France 10-yr yields more than Italy's). Moody's held France's rating on 24 Oct, triggering the partial retracement. Sources: https://exante.eu/press/market-updates/3311-will-ai-continue-to-shatter-records/ ; https://www.aieon.ee/eng/2026/10/french-10-year-yields-hit-5-14-as-barnier-warns-of-financial-crisis/ ; https://en.sedaily.com/international/2026/10/27/french-bond-yields-retreat-below-5-percent-after-moodys
• Context [analysis]: French public debt ~€3.5tn, deficit path repeatedly missing EU targets, Apr 2027 presidential election with both leading challengers proposing more spending — the fiscal-credibility channel of 2011, but the ECB now has TPI/OMT backstops it lacked in 2010.
• Reading: RED-AMBER. PMI still >50 for the bloc and spreads retraced on the Moody's hold, but a core sovereign at 160–178 bp over Bunds with a contracting economy is the one DM configuration that historically forces a policy response. Tripwire: OAT–Bund >200 bp or a failed French auction.

4.3 United Kingdom — resilient activity, punitive gilt curve, Budget event risk (28 Oct)

• Flash composite PMI 53.1 Oct — strongest since Feb, new orders fastest in 2 years, firms hiring again; price pressures strongest since Jun. Verified via Reuters: https://www.reuters.com/world/uk/uk-businesses-report-strongest-month-since-february-pmi-shows-2026-10-23/
• Bank Rate 3.75% (held 6–3, three members for a hike); a Nov 5 hike to 4.00% priced at ~86–95% across reports. Inflation prints in the sources range 3.1–3.5% by month/release — treat the exact latest month as partially unverified; direction (above target, re-accelerating on energy) is consistent across sources.
• 10-yr gilt 5.33% (26 Oct) after 5.53% on 8 Oct — highest since Jul 2007; 30-yr 5.89–6.04% — highest since Jan 1998 (above the Sep 2022 crisis peak level, but reached by a grind, and the 2022 lesson is that speed, not level, breaks LDI — buffers are now regulatorily larger). Unemployment reported at 4.7–4.9% across Oct releases (release-dependent); sterling $1.32–1.34. Sources: https://www.morningstar.com/news/dow-jones/202610085220/uk-gilt-yields-hit-multidecade-highs-as-boe-officials-signal-inflation-concerns ; https://dailybusinessgroup.co.uk/2026/10/rising-gilt-yields-add-pressure-to-uk-business-costs/ ; https://exante.eu/press/market-updates/3311-will-ai-continue-to-shatter-records/
• Reading: AMBER. Activity/labour argue resilience; the gilt curve + fiscal headroom “cut nearly in half by borrowing costs” + a Budget the day after this report is written is a textbook 2022-style event window. Tripwire: 30-yr gilt moving >50 bp in days (not the 6% level itself) or sterling and gilts falling together.

4.4 Japan — tightening into an AI-led equity wobble; carry not (yet) unwinding violently

• BoJ policy rate 1.25%, expected hold this week with hawkish signalling (underlying 2% inflation reportedly to be declared “met”); next hike expected Dec/Jan; terminal 1.75–2.00% by mid-2027 (CNBC survey of economists, 26–27 Oct). Core CPI ex-fresh food 2.1% Sep; real wages +1.5% (Aug, 8th positive month). Sources: https://asiainsiders.net/bank-of-japan-is-expected-to-hold-rates-this-week-but-stay-hawkish-cnbc-survey/ ; https://tru.express/news/d0fbb65e-6321-4ae9-96fd-9ebb1156eacd
• 10-yr JGB 2.79–2.81%, 30-yr 3.81–3.83%; USDJPY 157.76 (27 Oct), survey range 157–158 for the next month — i.e. the yen has not spiked; leveraged-fund yen shorts actually fell (33,673 → 22,890 contracts, 13→20 Oct, CFTC via the funding compilation above) — no 2024-style crowded unwind in progress. Hedged math [analysis]: at Fed 3.75–4.00% vs BoJ 1.25%, a currency-hedged 10-yr Treasury yields a Japanese investor ~2.3% vs ~2.9% on a 10-yr JGB — the repatriation incentive is now real and grows with each BoJ hike.
• Nikkei 225 62,190 (26 Oct, 3-month low), −4.9% on 15 Oct (3rd-largest point fall on record), chip/AI names leading (SoftBank, Tokyo Electron, Kioxia); G7/G20 finance ministers stated no financial-stability concern from the AI selloff (Japan MoF, mid-Oct). Sources: https://en.infomaxai.com/news/articleView.html?idxno=145150 ; https://cryptobriefing.com/japan-mof-g7-g20-ai-selloff/
• Reading: AMBER. The risk is a combination: BoJ hike + Fed hold/cut narrowing the differential from both ends while AI equities are already falling — the Aug 2024 template with larger JGB yields. Tripwires: USDJPY −5%+ in a week, 10-yr JGB through 3.1% quickly, Nikkei −10%+ weeks.

4.5 China — property balance-sheet recession continues; growth stabilised by exports, not by property

• Q3 GDP +4.5% YoY (Q2 +4.3%, Reuters, NBS data 19 Oct); retail sales +1.1% YoY Sep; industrial production +5.5%; surveyed urban unemployment 5.2%. Property investment −20.6% YTD Sep (worsening), residential sales area −13.7% YTD, new-home prices −2.8% YoY (monthly falls moderating: −0.14% m/m new, −0.26% used). Sources: https://985theriver.com/2026/10/18/chinas-q3-gdp-growth-accelerates-to-4-5-beating-forecast/ ; https://news.marketflux.io/news/china-q3-gdp-grows-4-5-year-on-year-beating-forecasts-on-strong-442e34e6b17c792b5c6a5f7308a00ab4.html ; https://aitimeline.in/china-export-boom-consumer-struggle-timeline-19132/
• Structural claims to label: deflation for ~10 consecutive quarters and true growth well below official prints (Rhodium Group “rightsizing” analysis) are [analyst opinion], contested, and should not be entered as data. An S&P Global Ratings note (early Oct 2026, via republication) sees tier-1 prices possibly recovering in 2027 and a national bottom only by Q3 2028 — https://stockxpo.com/2026/10/08/chinas-real-estate-market-a-potential-turnaround-amid-policy-shifts/
• Credit impulse current numeric value: UNVERIFIED in this run (only a qualitative “high” in one analyst's Oct 2026 liquidity model).
• Shanghai Composite ~3,713 (15 Oct). No 2015-style leverage froth was found in any Oct 2026 source — the equity-leverage family is quiet in China this time; the risk remains the slow property/local-government channel and export dependence if the AI cycle or EU trade turns.
• Reading: AMBER (chronic, not acute). Property is still contracting at −20% investment rates five years in; that has historically been a drag, not a contagion trigger, provided banks are shielded by the state — EM/DM contagion from China property has been minimal since 2021 for exactly that reason.

4.6 EM contagion check

EMBI dollar spread ≈ 189 bp (Bloomberg, via 12 Oct reporting) vs 447 bp at the Oct 2011 EZ-crisis peak; EM dollar-bond spreads described as tightest since 2007 (28 Sep), with managers trimming HY EM (Colombia, Ecuador, Zambia) because valuations offer no cushion against US yields at 20-year highs. Analyst view [opinion] that EM is better armoured than in 2011 (fiscal/external balances, light positioning). Sources: https://businessmirror.com.ph/2026/10/12/emerging-markets-better-armed-for-renewed-european-debt-turmoil/ ; https://bworldonline.com/bloomberg/2026/09/28/782638/traders-grow-wary-on-emerging-dollar-bonds-as-us-yields-soar/ Reading: GREEN on current pricing, AMBER on vulnerability — tight spreads here are a coincident complacency reading, not an all-clear; the 1997/2022 templates both began with spreads that looked fine until the dollar/rates shock was underway.

4.7 Overall risk matrix

Note on US Treasuries/credit vs history, as asked: Treasuries sit in the opposite configuration to every post-1970 pre-recession setup except 1973–74/1979–80: pre-recession curves were inverted or bull-steepening on cuts with falling long yields; Oct 2026 has a positively sloped curve with long yields at 19–24-year highs and the Fed contemplating hikes. Credit is split the same way: broad HY at 338 bp resembles mid-2007's complacency (200s) more than any pre-recession widening, while the CCC tail at ~1,200 bp resembles the first stage of 2007, when stress was confined to the weakest borrowers for ~6 months before generalising. That CCC/broad divergence is the single most informative US credit gauge to run daily.

Could not verify

• Cross-currency basis (EUR/JPY USD basis), late Oct 2026: no dated public source found — do not fill from memory.
• MOVE exact close for 23–27 Oct 2026: only early/mid-Oct values (~104–114) sourced; direction (elevated >100) is supported, the latest print is not.
• FINRA margin debt after Jun 2026 (Jul–Oct): sources lag; the record $1.502tn is a June figure.
• Temp-help employment, Oct 2026; China credit impulse, numeric, Oct 2026; EZ unemployment, Oct 2026; Japan unemployment/GDP, Oct 2026: not sourced in this run.
• UK CPI / unemployment exact latest month: Oct reports give 3.1–3.5% CPI and 4.7–4.9% unemployment depending on release/month — the range is reported, a single “latest” figure is not asserted.
• Conflicting prints noted, not resolved: 10-yr Treasury early-Oct peak (5.31% vs 5.35% by source/session); Brent on 23 Oct (one Japan report says Brent “briefly $94” while US/European sources put Brent >$100 all month — the $94 figure is treated as an outlier/possible contract error and not used); HY OAS Dec-2008 peak (19.88 pp FRED-cited vs 21.82 pp Pitchbook series); S&P drawdown depths differ by daily-close vs monthly-average methodology throughout Part 1 (both bases are flagged where they matter).
• Small-n warning: every hit rate in §1 rests on 7–12 episodes with author-chosen signal definitions; treat them as orders of magnitude for ranking, not probabilities. Several Oct-2026 data aggregators used here are low-editorial-oversight republications of FRED/BLS/ISM data — values were only used where a second source or a primary publisher corroborated the ballpark.

Sources

Primary / publisher (fetched in full, 27 Oct 2026): - Federal Reserve, SLOOS July 2026 — https://www.federalreserve.gov/data/sloos/sloos-202607.htm - Conference Board LEI Sep 2026 (PR Newswire release, 22 Oct 2026) — https://www.prnewswire.com/news-releases/the-conference-board-leading-economic-index-lei-for-the-us-was-flat-in-september-302917003.html - Reuters, Eurozone flash PMI Oct 2026 (23 Oct 2026) — https://www.reuters.com/world/europe/euro-zone-growth-slows-4-month-low-amid-rising-costs-french-turmoil-pmi-shows-2026-10-23/ - Reuters, UK flash PMI Oct 2026 — https://www.reuters.com/world/uk/uk-businesses-report-strongest-month-since-february-pmi-shows-2026-10-23/ - Federal Reserve FEDS Note (NBER chronology) — https://www.federalreserve.gov/econres/notes/feds-notes/the-restrained-recovery-of-state-and-local-government-payrolls-from-the-pandemic-recession-20230804.html - ECB Working Study on OMT effects — https://www.ecb.de/pub/pdf/scpwps/ecbwp1707.pdf?938b7d20c8b82b5450f2a9ac5de68beb

Data series / reconstructions (index reads, 27 Oct 2026; each is a republication or reconstruction of the named primary series): - FRED T10Y2Y/T10Y3M histories & Oct 2026 values — https://eco3min.fr/en/t10y3m-recession-signal-10y-3m-spread-explained/ ; https://eco3min.fr/en/yield-curve-inversion-history-2s10s-spread/ ; https://eco3min.fr/en/yield-curve-inversion-history-dataset/ ; Cleveland Fed record via https://thefinsense.io/yield-curve-recession-indicator/ - Sahm rule history & 2024 false signal — https://eco3min.fr/en/sahm-rule-false-signals-history/ ; current 0.00 — https://whatisarecession.com/current-probability-2026 - FRED BAMLH0A0HYM2 (HY OAS 3.38%, 23 Oct 2026) — https://govspending.org/series/BAMLH0A0HYM2/ ; IG/CCC 2 Oct — https://marketbriefer.com/the-global-close-october-7-2026-the-10-years-24-year-high-ends-the-record-run/ - US yields Oct 26 — https://www.briefs.co/news/treasury-yields-hold-steady-as-investors-await-fed-decision/ ; EZ/UK yields Oct 26 — https://exante.eu/press/market-updates/3311-will-ai-continue-to-shatter-records/ - ISM Sep 2026 — https://www.thefinancialeconomy.com/ism-manufacturing-pmi/ ; CAPE — https://www.fool.com/investing/2026/10/25/witnessing-stock-market-do-something-3rd-time-in-100-years-history-crystal-clear-whats-next-for-stocks/ ; margin debt — https://primexbt.com/news/record-margin-debt-surge-flashes-historic-warning-for-the-sp-500/ - VIX/SKEW/S&P, 26 Oct — https://www.home.saxo/content/articles/options/paying-for-the-index-not-the-names---options-brief---26-october-2026-26102026 ; MOVE — https://www.tokenpost.com/news/investing/29460 ; SOFR/IORB — https://beepboop2025.github.io/financial-evidence-skills/media/briefs/2026-10-11-dollar-funding/ - Housing Sep/Oct 2026 — https://metapress.net/economy/2026/10/21/housing-is-flashing-warning-signs-just-as-the-fed-weighs-more-interest-rate-hikes/ ; https://newslink.mba.org/mba-newslinks/2026/october/chart-of-the-week-mortgage-originations-forecast/ - Commodities 27 Oct — https://price-chart.com/ ; https://www.kitco.com/news/off-the-wire/2026-10-27/gold-retreats-stronger-dollar-traders-brace-fed - France spreads — https://www.aieon.ee/eng/2026/10/french-10-year-yields-hit-5-14-as-barnier-warns-of-financial-crisis/ ; https://en.sedaily.com/international/2026/10/27/french-bond-yields-retreat-below-5-percent-after-moodys ; UK gilts — https://www.morningstar.com/news/dow-jones/202610085220/uk-gilt-yields-hit-multidecade-highs-as-boe-officials-signal-inflation-concerns - Japan — https://asiainsiders.net/bank-of-japan-is-expected-to-hold-rates-this-week-but-stay-hawkish-cnbc-survey/ ; https://en.infomaxai.com/news/articleView.html?idxno=145150 ; China — https://985theriver.com/2026/10/18/chinas-q3-gdp-growth-accelerates-to-4-5-beating-forecast/ ; EM — https://businessmirror.com.ph/2026/10/12/emerging-markets-better-armed-for-renewed-european-debt-turmoil/

Historical episode sources are cited inline in §2 next to each claim.