Independent economic research
Macro. Markets. Credit. Risk. In one place.
My personal framework for understanding economic cycles, what’s happening now, and the signals I’m watching next.
Seven gauges. Two clocks. Never blended.
Current feeling: Oh shit.
Two model dials from the CVETF framework, computed by the weekday indicator watch from 10-year percentile ranks and EMA-smoothed. Updated with the daily snapshot.
Risk Dial · R(t)
59/100
Composite risk dial, 0-100: credit, labor, growth, market, inflation and tail-risk percentiles, smoothed.
Credit Sub-Dial · C(t)
71/100
Credit sub-dial, 0-100: HY/IG/Euro HY/CCC spreads, private-credit defaults and bank lending standards, smoothed.
As of Nov 13, 2026
The cockpit's health bars, on the same scales, zone colors, thresholds and chips: good on the left, danger (red) on the right, solid marker now, faded markers 1 and 3 months ago, and a chip for the one-month move. CDG has no history yet, so its bar carries the level marker plus a state dot (red = FIRED) instead of ghost markers.
One panel per indicator, each on its own scale, with the suite's trigger levels as dashed gold lines and NBER recessions shaded on the long panels. CSG plots the sourced anchors through Sep 2026, then readings computed daily from SPY's published holdings; nothing interpolated. CDG begins at its 2026-10-14 seed and accrues one reading per daily run; no earlier history exists and none is reconstructed.
789 readings · Jan 1961 – Sep 2026
Model-computed monthly history (logistic model, FRED ingredients). Dashed line: the daily watch fires at 20%.
Daily tier as of Nov 15, 2026 · +1.9 since yesterday
922 readings · Jan 1950 – Oct 2026
Model-computed monthly history. Dashed line: trigger zone at 60.
Daily tier as of Nov 15, 2026 · unchanged since yesterday
922 readings · Jan 1950 – Oct 2026
Model-computed monthly history. Dashed line: trigger zone at 60.
Daily tier as of Nov 15, 2026 · unchanged since yesterday
1522 readings · Jan 1900 – Oct 2026
Three-leg blend (log-CAPE, Buffett deviation, ECY) for every start month; all three legs from Jan 1962, earlier months averaged over the legs then available. Dashed line: zero.
Daily tier as of Nov 15, 2026 · unchanged since yesterday
794 readings · Sep 1960 – Oct 2026
Model-computed monthly history (expanding percentiles, no look-ahead). Dashed lines: 85 escalation, 98 watch fire.
Daily tier as of Nov 15, 2026 · unchanged since yesterday
12 readings · Dec 1990 – Nov 2026
Sourced anchors through Sep 2026 plotted as points, then readings computed daily from SPY's published holdings. Nothing interpolated. Dashed line: watch fires at 42%.
Daily tier as of Nov 15, 2026 · unchanged since yesterday
33 readings · Oct 2026 – Nov 2026 · begins at the 2026-10-14 seed
Seeded 2026-10-14; the gauge did not exist before, so no earlier history exists or is reconstructed. One reading accrues per daily run (trends-log.json). Dashed line: ARMED at 60.
Daily tier as of Nov 15, 2026 · unchanged since yesterday
Logistic probability that a recession begins within 12 months, from five standardized features: 10Y-3M spread (long-history GS10 minus TB3MS), Sahm, industrial production 6m change, building permits YoY, payroll 6m growth. Accuracy: AUC 0.88 in-sample; 0.83 out-of-sample (fit 1959-1999, tested 2000-2026). Only 3.2% of non-recession months ever printed above 50%. Median probability around past peaks: 25% a year before, 29% at -6m, 28% at -3m, 33% at the peak — this model speaks in odds, not alarms, and it rarely exceeds ~35% before the fact. Honest flaw: the yield curve does nearly all the work (curve-only AUC 0.87); the other features mostly calibrate. And because the curve re-steepened in Dec 2024, RPI has already forgiven the 2022-24 inversion — if that inversion was the signal and the lag is simply unprecedented, RPI will be the last to know. The post-inversion clock: re-steepening was 2024-12, 21 months ago on current data. In the four sampled cycles, recession followed re-steepening within 1-11 months — so either this cycle's lag is the longest on record, or the inversion's signal has expired. RPI's low reading assumes the latter; the clock is printed here so the assumption stays visible.
Labor only, 0-100: Sahm scaled (0.50 = full weight), claims YoY, payroll 6m momentum, temp-help YoY (1990+). Pure labor, because every postwar recession is ultimately a labor event and labor is where false alarms go to die. Accuracy at ≥60: matched 8 of 8 recessions since 1970 (all of them), false-positive episodes 3 — 2003 (jobless-recovery echo), Jul 2024 (the supply-driven Sahm summer), and Sep 2025 (a genuine soft patch that healed). Episodes: 1970-02→1969-12 (-2m) · 1974-08→1973-11 (-9m) · 1980-04→1980-01 (-3m) · 1981-10→1981-07 (-3m) · 1990-08→1990-07 (-1m) · 2000-12→2001-03 (+3m) · 2003-02→none · 2007-09→2007-12 (+3m) · 2020-03→2020-02 (-1m) · 2024-07→none · 2025-09→none. Note the +3m leads in 2000 and 2007: claims and temp help turn before headline unemployment does, which is the lead Sahm alone never gives.
Credit channel, 0-100: a widening leg that takes the worse of Baa-Aaa widening from its 12m low and HY daily widening from its 6-month low (the HY leg exists from Oct 2023, FRED's ICE window; before that, Baa carries the history), plus spread level and NFCI. Accuracy at ≥60: matched 6 of 7 recessions since 1970 (the miss is 2001 — the one postwar recession with no financial-stress event to measure; it peaked at 50), false positives 5 (1971 and 1974 recession tails, 1984 Volcker stress, 2011 euro crisis, 2022 inflation shock — real stress, no new recession). Verdict: FSM is a coincident severity gauge, not a leader — when it fires, the question is no longer whether, but how deep. Today it reads 33 (Oct 2026 monthly reading), and the composition is the story: the entire reading comes from the HY leg — spreads widened off the late-August low while equities sat at record highs, the divergence the tripwire board flagged — while the Baa-Aaa and NFCI legs still read zero. Stress is being priced in the daily market before it reaches monthly credit or bank conditions, which is exactly the blindness the Oct 2026 improvement analysis fixed (the pre-fix gauge read 0). Episodes: 1971-01→none · 1973-09→1973-11 (+2m) · 1974-10→none · 1979-06→1980-01 (+7m) · 1981-10→1981-07 (-3m) · 1984-02→none · 1990-10→1990-07 (-3m) · 2007-12→2007-12 (+0m) · 2011-10→none · 2020-03→2020-02 (-1m) · 2022-07→none
The tier below HY is flashing louder than HY itself. ICE BofA CCC & lower OAS closed Oct 8, 2026 at 12.52%, +3.48pp off its May 2026 low of 9.04% in five months, while BB widened ~0.44pp and BBB ~0.10pp over the same span; the CCC−HY quality gap of 9.37pp was then the widest in FRED's window. FRED truncates this series to ~3 years, but TradingView's chart of the same series carries the full history (verified 2026-10-11): since Dec 1996 the all-time high is 36.7% (Nov 2008) and the stress peaks were 23.9% (Dec 2000), 14.5% (Sep 2011), 19.8% (Feb 2016), 17.9% (Mar 2020) and 12.81% (Sep 2022). That 12.52% sat 0.29pp under the entire 2022 rate-shock peak, around the 85th percentile of months since 1996 — and in every prior episode, a momentum crossing of ~12% overshot by +1 to +24pp rather than topping (the first >12% monthly close in 2008 came in March, eight months before the peak, with a head-fake dip back below 12% in between). Since that print the tier has widened through the 2022 peak: CCC peaked at 13.56% on Oct 23, 2026 — the CCC−HY gap reached 10.18pp the same day — and closed Nov 9, 2026 at 13.11%, with the gap at 9.77pp (HY 3.34%). CCC and BBB widening wires joined the tripwire board on 2026-10-11; both CCC wires seeded tripped and BBB is calm, so the stress is so far contained to the weakest tier — BBB joining is the contagion tell.
Expected next-10-year real total return as an equal-weight blend of three legs, each fit by its own regression to the same target (forward 10y real total return, Shiller data): log-CAPE, the Buffett indicator's deviation from its long-run trend (Fed Z.1 equities over GDP), and ECY — the CAPE earnings yield minus the ex-post real 10-year yield. No single valuation ratio carries the reading alone. The blend was adopted only after it beat the CAPE-only gauge out of sample: fitted on start months through 1999 and tested on 2000-2013 starts, it cut forecast error by a third (RMSE 3.77 vs 5.54) and turned a negative out-of-sample R² (−0.475) positive (+0.316). On the full common sample (621 start months, 1962-2013) the blend's R² is 0.425. The CAPE leg's quintile ladder — monotone across 125 years — is the long-history context:
| CAPE quintile (1900+) | CAPE range | Median actual next-10y real return |
|---|---|---|
| Q1 | 5-11 | +10.8% |
| Q2 | 11-14 | +8.7% |
| Q3 | 14-17 | +5.5% |
| Q4 | 17-22 | +5.1% |
| Q5 | 22-44 | +4.0% |
Today the legs disagree, and the disagreement is the information. CAPE at 40.09 implies −0.4%/yr real on its own; the Buffett leg implies −2.5%/yr; the earnings-yield leg, which sets what equities earn against the real yield on the 10-year, still implies +5.2%/yr. The blend reads +0.8%/yr real. Honest precision: the blend's residual SD is 4.2pp, so the ±1 SD band runs −3.4% to +4.9%/yr — even its optimistic edge sits below the ~+6.5% long-run real average, and its pessimistic edge is a lost decade. A valuation statement, not a forecast with timing.
Mean of five extremity percentiles (expanding, no look-ahead): CAPE, Buffett trend deviation, household equity allocation (37.8% of total household assets, Fed Z.1 — the positioning gauge), top-10 concentration (range-normalized across sourced anchors; no continuous official series exists), and margin leverage (FINRA/NYSE margin debt YoY growth, 1959+). The leverage leg today: margin debt hit a record $1,502B in Jun 2026 and stood at $1,480B in Sep 2026, +31.4% YoY (82nd percentile since 1959), with debit balances at 7.0x free credit cash balances — the thinnest cash cushion in the dataset. Leverage is the amplifier: it does not start declines, it converts them into forced selling. Accuracy — as a return gauge, excellent:
| DRG band | Months | Median actual next-10y return (nominal, annualized) |
|---|---|---|
| 0-50 | 320 | +10.9% |
| 50-70 | 299 | +6.5% |
| 70-85 | 82 | +1.2% |
| 85-101 | 93 | +1.1% |
Accuracy — as a crash predictor, deliberately shown mixed. DRG at the start of every major drawdown:
| Peak | Drawdown | DRG at peak |
|---|---|---|
| 1968-11 | -36% | 93 |
| 1973-01 | -48% | 54 |
| 1980-11 | -27% | 29 |
| 1987-08 | -34% | 50 |
| 1990-07 | -20% | 32 |
| 2000-03 | -49% | 86 |
| 2007-10 | -57% | 70 |
| 2020-02 | -34% | 47 |
| 2022-01 | -25% | 74 |
Valuation-driven tops (1968, 2000, 2022) started from DRG 74-93. Policy/exogenous crashes (1980, 1987, 1990, 2020) started from 29-50. DRG measures how much air is under the market, not whether something will push it. At 93 today, the air is as thin as at the thinnest of those peaks.
Your concentration point, quantified two ways. History through Sep 2026 is 11 sourced anchors (J.P. Morgan, S&P Dow Jones Indices, FactSet/RQI, Goldman Sachs, constituent-cap calculations). Since Nov 2026 the gauge is computed daily from State Street's official SPY holdings file — the index's own published weights, with dual share classes combined by issuer (Alphabet counts once) — so the current reading no longer waits on anyone's research note:
| Date | Top-10 share of S&P 500 market cap |
|---|---|
| 1990-12-31 | 20.0% |
| 2000-03-31 | 26.6% |
| 2015-12-31 | 19.0% |
| 2020-12-31 | 28.6% |
| 2022-12-31 | 24.0% |
| 2023-10-31 | 32.0% |
| 2024-12-31 | 37.0% |
| 2025-12-31 | 41.0% |
| 2026-05-31 | 40.8% |
| 2026-08-31 | 37.8% |
| 2026-09-30 | 39.2% |
| 2026-11-12 | 38.67% (first computed reading) |
Two facts, and they pull in opposite directions. Cap concentration is near its record: 38.67% (Nov 12, 2026, computed from SPY holdings), just under the 41.0% anchor printed for Dec 2025, vs. 26.6% at the 2000 peak — and the index's leaders are the same names across the S&P 500, Nasdaq-100, and growth indexes, with a 36-month S&P/Nasdaq return correlation of 0.96. (That correlation's percentile is only 38th since 1971 — the two indexes have moved as one for most of two decades, so sync alone is a weak discriminator; the concentration level is the real break with history.) But the earnings support is real this time: the top 10's cap-vs-earnings gap is roughly 40% of cap vs. ~37% of expected earnings (Goldman, Oct 2026) — a ~3pp unsupported-weight gap, vs. ~12pp in 2000 (27% of cap, ~15% of earnings). Concentration here is a single-point-of-failure risk (a handful of earnings reports now move the index), not the hollow froth of 2000. If those earnings crack, there is no diversified index left underneath — that is the collapse mechanism this gauge watches: top-10 earnings revisions, which the other five indicators would only catch after the fact.
The first six indicators watch the whole market. CDG watches one chain: in the AI buildout, a small number of private hubs (today, effectively OpenAI) have signed purchase commitments now measured in the hundreds of billions; those commitments are financed in part by the hub's own suppliers and investors, and public neoclouds have borrowed against them. The exposure chain runs both directions: hub → neocloud contracted revenue → neocloud bonds, loans and equity → public equity and credit indexes. It is concentration plus circularity plus a funding gap, with every leg exposed to the same counterparties. Spec: cross-default-gauge-spec-2026-10-14.md. Headline discipline: only definitive and closed deals are ever summed; LOIs are never added to totals (announced-but-unfunded headline figures circulating for this chain exceed $1.4T and mix statuses).
Composite level: 76 / 100 (0.35·CON + 0.30·CIR + 0.35·GAP). State: FIRED — the level alone can only reach ARMED; CONFIRMED additionally requires the market layer MKT ≥ 60 plus a defined confirmation event in the trailing 6 months; FIRED requires MKT ≥ 80, a Type-4 hub event, or distress CDS bands. Clearing steps down one rung at a time with hysteresis (two consecutive monthly readings; quarterly for ARMED→NORMAL).
CON 81: CoreWeave top-2 customers = 62% of revenue (Q2 2026 10-Q; top-1 was 67% in FY2025, so direction is improving at the anchor); Applied Digital top-2 = 84% (FY2026 10-K). CON = 0.5·median + 0.5·max of the banded subscores (70, 85). Nebius is excluded — it does not disclose customer revenue shares; it is carried as a commitment memo only (~$32B firm of >$40B reported commitments from Microsoft and Meta). Reverse concentration memo: S&P's Jul 2026 Oracle downgrade named OpenAI at nearly half of Oracle's $638B backlog.
CIR 70 (provisional): 6 qualifying circular-financing events in the trailing 12 months (window from 2025-11-01): C-06b, C-08, C-09, B-02, B-03, B-04. Dollar intensity is not computable today — warrant values, backstop utilization, and Nvidia's share of the OpenAI round are undisclosed — so the layer is velocity-scored and flagged provisional until filings disclose the values. The one computable pair, Microsoft/OpenAI (~$13.8B invested vs a $250B Azure commitment, ~6%), is a memo, not the layer.
GAP 75: OpenAI commitments of $750B through 2030 (Reported Jul 2026) = $150B/yr against a ~$50B annualized run rate (net basis, FT, 2026-10-08) — ratio 3.0, band 75. Every figure is a reported estimate, not an audited financial; bases are never blended. Sensitivity (commitment / run-rate basis → ratio (score)): $600B / gross $70B → 1.71 (55) · $600B / net $50B → 2.4 (75) · $750B / gross $70B → 2.14 (75) · $750B / net $50B → 3.0 (75). Even the most favorable pairing scores 55. Anthropic memo (not scored): ratio ~1.4-1.6, band 55.
MKT 73: Oracle 5Y CDS ~298 bps (2026-11-06, two independent reports, band confirmed) → 65; CoreWeave 5Y CDS ~1155 bps (2026-11-06) → 80 SINGLE SOURCE, provisional; primary market → 70 (CoreWeave senior notes at 9.75% Apr 2026 and 9.625% Jun 2026; the Aug 2026 $2.6B term loan flexed ~+100-125 bp wider); CRWV equity drawdown 49.2% from its 52-week high → 75. A CDS band may change only on two independent reports; a single report sets a provisional band with a flag. Readings go stale at 45 days and drop out at 90. The suite's credit tripwires (CCC/HY/BIZD/Fitch) are displayed alongside CDG in the cockpit but are never scored into MKT — counting them twice would double-count the same evidence.
Confirmation events already in the trailing 6 months (why the state is FIRED, not merely CONFIRMED):
What fired it: FIRED requires MKT ≥ 80, a Type-4 hub event (payment delay, renegotiation, commitment reduction, missed funding milestone, flat/down round, or failed raise), or distress CDS bands — Oracle 5Y CDS > 300 bps or CoreWeave 5Y CDS > 1,000 bps, each on two independent reports. The band condition fired first: in the week of Oct 20-24, 2026, Oracle printed 343 bps and CoreWeave 1,167 bps, both two-report confirmed, and the state stepped up to FIRED on the Oct 26, 2026 reading. A Type-4 hub event has since occurred as well — Microsoft's commitment reduction toward its neocloud suppliers (Nov 6, 2026; in the events above) — and it is currently the condition holding the state at FIRED: the latest CDS prints no longer satisfy the band condition (Oracle ~298 bps, back under 300; CoreWeave ~1,155 bps on a single report, so its band stands unconfirmed; both Nov 6, 2026). What the gauge is not: not a probability, not a timing tool, not a claim that default is likely; CON and CIR describe an exposure structure, GAP a funding arithmetic, and only MKT says anything about likelihood. It also says nothing about whether the buildout is economically justified — only about what happens to the chain's creditors and shareholders if it is not.
| ID | Announced | Parties | Structure | Financing value | Status | Counts |
|---|---|---|---|---|---|---|
| C-01 | 2025-09-22 | Nvidia ↔ OpenAI | equity for capacity * | $100.0B | LOI | velocity |
| C-02 | 2025-10-06 | AMD ↔ OpenAI | warrants for capacity | — | definitive | velocity |
| C-03 | 2025-10-28 | Microsoft ↔ OpenAI | equity stake plus purchase commitment | $135.0B | definitive | velocity |
| C-04 | 2025-09-10 | Oracle ↔ OpenAI | purchase commitment | — | definitive | memo |
| C-05 | 2025-10-13 | Broadcom ↔ OpenAI | co development | — | definitive | memo |
| C-06a | 2025-09-01 | Nvidia ↔ CoreWeave | residual capacity backstop | $6.3B | definitive | velocity |
| C-06b | 2026-01-26 | Nvidia ↔ CoreWeave | equity investment | $2.0B | closed | velocity |
| C-07 | 2025-09-25 | CoreWeave ↔ OpenAI | purchase commitment | — | definitive | memo |
| C-08 | 2026-03-31 | Amazon ↔ Nvidia ↔ SoftBank ↔ OpenAI | funding round * | $122.0B | closed | velocity |
| C-09 | 2026-03-01 | Nvidia ↔ Nebius | strategic investment | $2.0B | closed | velocity |
| C-10 | 2025-11-03 | Microsoft ↔ IREN | purchase commitment | — | definitive | memo |
| C-11 | 2025-09-01 | Microsoft ↔ Nebius | purchase commitment | — | definitive | memo |
| C-12 | 2025-11-01 | Meta ↔ Nebius | purchase commitment | — | definitive | memo |
| B-01 | 2025-10-23 | Google ↔ Anthropic | supply agreement | — | definitive | memo |
| B-02 | 2026-04-01 | Google ↔ Broadcom ↔ Anthropic | capacity financing | — | definitive | velocity |
| B-03 | 2026-04-01 | Amazon ↔ Anthropic | equity plus purchase commitment | $5.0B | definitive | velocity |
| B-04 | 2025-11-01 | Microsoft ↔ Anthropic | equity plus purchase commitment | $5.0B | definitive | velocity |
* disputed: C-01's headline total is disputed by later reporting (the $100B figure is an LOI-era ceiling; Bloomberg reported in Feb 2026 the deal was “nowhere near finalized”); C-08's circulating claim of a completed $30B Nvidia investment in OpenAI is contradicted by Nvidia's own 10-K (filed 2026-02-25) and 10-Q (filed 2026-05-20). Statuses: LOI / definitive / closed / disputed / completed. Full notes, sources and conditions are in the register file.
Scripts: ~/workspace/valuation-model/work/analysis3.py (RPI/LIG/FSM/RPS/MSI), analysis4.py (DRG), cdg.py (CDG; data: cdg-deals.json, cdg-con.json, cdg-gap-log.json, cdg-mkt-log.json). Concentration research: work/concentration-research.md. Prepared for Michael Davis, 2026-10-10; CDG added 2026-10-14.
The Data Room. Every series behind these gauges — latest value, previous value, change, as-of date and source — in one table: All indicators →
Ranked slow-moving fault lines that standard gauges can miss, from the November 2026 volcano hunt. Eruption means the fault becomes a first-order macro or market event within 24 months. Probabilities are judgment ranges from the research; each size figure keeps its measured, estimate, or contested label. As of Nov 1, 2026. Editorial assessment, not a live feed. Click a card for the mechanism and the strongest counter-argument.
AI financing complex
Size. AI-related debt ~$1.03–1.07tn; 2026 hyperscaler capex guided ~$630–800bn [Estimate]. Off-balance-sheet SPV borrowing >$120bn in 18 months [Measured].
Trigger. Frontier-model revenue disappointment, a hyperscaler capex guide-down, or GPU rental prices collapsing.
Watch. Hyperscaler capex & FCF (filings); Oracle RPO concentration; CoreWeave debt & GPU rental price indices; the CDG gauge above.
Mechanism ›Mechanism. Vendor financing plus circular equity. Chipmakers invest in and guarantee customers; customers sign take-or-pay leases; SPVs borrow against those leases; private credit and insurers fund the SPVs; hyperscaler cloud revenue is partly their own capex returning as customer spend. GPUs depreciate in 3–6 years while the debt and power infrastructure amortise over 15–40. This is 1999 telecom vendor financing at roughly 30–50 times the scale.
Transmission. Capex is now a measurable share of GDP growth, so a capex stop is a demand shock. The borrowers, lenders, landlords, and chip suppliers are the same names that dominate the equity indices, so correlation goes to 1 inside the index itself.
Strongest counter. Unlike 2000, the core hyperscalers are among the most profitable companies in history and fund most capex from operating cash flow; much capacity is pre-leased to creditworthy tenants. A 2000-style equity bust without a 2008-style credit event is a coherent, arguably base, case.
Private credit / insurer nexus
Size. Global private credit ~$1.5–2.0tn [Estimate range]. About 40% of borrowers have negative free cash flow, up from 25% in 2021 [Measured, IMF].
Trigger. A redemption wave meeting gated funds, a large sponsor-backed default cluster, or an insurer regulatory action on affiliated offshore assets.
Watch. KBRA DLR and Fitch private-credit default & PIK indices; BDC PIK income share and non-accruals; redemption requests vs caps in fund filings.
Mechanism ›Mechanism. Opacity, PIK income, and retail-isation. Reported defaults run ~2% versus effective defaults of ~5–6.4% including selective defaults and liability-management exercises. In 2026 semi-liquid retail vehicles hit redemption caps across major managers, and one Blue Owl fund halted redemptions permanently. US life insurers allocate ~30% of invested assets to private assets, with PE-owned insurers running originate-to-own-affiliate structures where valuation, origination, and ownership sit in one group.
Transmission. Banks (credit lines plus ~$300bn direct), insurers and annuitants, retail wealth channels, and sponsor-owned employers. If one manager's marks are doubted, all are — informational contagion.
Strongest counter. Private credit is only ~2.8% of US private-sector liabilities; most capital is locked-up institutional money that cannot run; fund-level leverage is modest versus 2008 banks. A painful 8% default cycle is absorbable — it becomes systemic only via the bank, insurer, and retail channels, which is why those channels are the watch item.
US fiscal / term premium
Size. Debt held by the public crossed 100% of GDP in June 2026; FY2026 deficit ~$1.9–2.1tn [Measured / contested range]; net interest ~$1.0tn a year.
Trigger. A failed or very weak Treasury auction, a debt-ceiling accident, an inflation re-acceleration forcing hikes into this supply, or a foreign-buyer strike.
Watch. Treasury auction tails and indirect-bidder share; NY Fed ACM term premium; net interest as a share of revenue.
Mechanism ›Mechanism. r > g arithmetic with a primary deficit near 2% of GDP. Every point of yield the market demands raises issuance, which raises the yield demanded. Market functioning is the fragility — 2019 repo and March 2020 showed Treasury depth can vanish. In Sep 2026 the Treasury tripled buybacks to $6bn and long yields rose on the news, the dress rehearsal for a repricing.
Transmission. The Treasury yield is the discount rate for everything else on this page. A term-premium repricing is simultaneously a fiscal, equity-multiple, mortgage-rate, and CRE-refinancing event. This is the volcano whose eruption triggers the others.
Strongest counter. No fixed debt ratio triggers crisis for the reserve-currency issuer borrowing in its own currency; auctions through Oct 2026 drew normal demand and Treasuries rallied as a haven in the October sell-off. The risk is a repricing, not default — but a repricing is damaging enough on its own.
Concentration / valuation amplifier
Size. CAPE 40.7, 98.9th percentile since 1881; top 10 companies ~39% of the S&P 500; households hold a record 48.2% of financial assets in equities; margin debt $1.48tn [Measured].
Trigger. Anything else on this list erupting, or an earnings disappointment in the top-10 names — which are also the AI borrowers of Rank 1.
Watch. Fed Z.1 household equity allocation; FINRA margin debt; the CSG and DRG gauges above.
Mechanism ›Mechanism. Not a standalone volcano — the transmission belt that converts any other eruption into a demand shock. A 25–35% equity drawdown, the CAPE mean-reversion zone, hits the cohort that does roughly half of consumer spending, while margin calls force mechanical selling at the margin. The top 10% of households own 87–88% of equities, and top-decile investment-to-checking transfers now fund over 10% of their spending — the wealth effect is financing current consumption.
Transmission. Wealth-effect consumption reversal; margin and leveraged-ETF deleveraging; retirement-account confidence effects.
Strongest counter. Household assets-to-debt is above 10:1; even a 20% equity fall leaves it higher than at any time before late 2025. Most equity sits in retirement accounts that historically are not panic-sold, and excess CAPE yield is still positive — equities are less stretched versus bonds than in 2000. Valuation predicts poor 10-year returns, not a crash.
Leveraged loans / software distress
Size. US leveraged loans ~$1.5tn; $65bn trading below 60 cents, the most since Mar 2020; software alone $25–40bn distressed, a record [Measured].
Trigger. The 2027–28 software maturity wall meeting a closed refinancing market, or a marquee sponsor default.
Watch. Morningstar LSTA share of loans below 80 cents and LME share; CCC spread (ICE BofA); JPMorgan distressed tallies.
Mechanism ›Mechanism. 2020–21 LBO software deals were underwritten on perpetual ARR growth at 6–7 times leverage; AI now attacks exactly that ARR assumption, while floating rates keep debt service high. Covenant-lite structures remove the early tripwire, so distress is restructured rather than defaulted — liability-management exercises are 64% of combined defaults plus LMEs — and losses arrive later and larger, concentrated in CLO mezzanine and equity and in the insurers and private credit that fund CLOs.
Transmission. CLOs to insurers, banks, and Japanese holders; sponsor portfolio-company employment; private-credit marks on the same borrowers.
Strongest counter. Payment defaults remain under 1%; CLO structures survived 2020 and AAA tranches have never lost historically; distress is heavily one sector's repricing rather than broad deterioration. Loan prices below 80 have repeatedly marked buying opportunities when the macro held.
France / euro sovereign-bank loop
Size. OAT–Bund 10-year spread above 140bp, widest since 2012 and wider than Italy or Greece; French debt ~119–120% of GDP; deficit 5.4% of GDP [Measured].
Trigger. Budget failure or government collapse, a rating-downgrade cascade, or the 2027 election pricing a fiscally expansionist winner.
Watch. OAT–Bund spread and auction bid-covers (Agence France Trésor); French bank CDS; ECB TPI signalling.
Mechanism ›Mechanism. The 2011 loop in the core. Fiscal slippage widens spreads; French banks holding OATs take balance-sheet damage (their CDS already widening, Italian CDS following); credit tightens; growth and the deficit worsen. France is too big for an ESM-style rescue — the ECB's TPI is the backstop and its activation is political. Market depth is more than 50% below normal.
Transmission. Euro bank equities and CDS into global bank funding; the euro; European credit creation; US multinational earnings.
Strongest counter. 2012 spreads reached ~190bp without a French default; the ECB now has TPI and PEPP flexibility that did not exist in 2011; much of the 2026 widening was a positioning shock rather than pure fundamentals. Eruption means a spread crisis requiring ECB action, not a euro breakup.
CRE / regional banks
Size. $875bn of commercial mortgages mature in 2026 and $652bn more in 2027; office CMBS delinquency 12.2%, second-highest ever [Measured, Trepp].
Trigger. The wall itself if the 10-year stays above 5%, or a regional-bank earnings miss that forces marks.
Watch. Trepp CMBS delinquency, especially office; MBA maturity survey; Fed H.8 bank CRE delinquency; SLOOS CRE standards.
Mechanism ›Mechanism. 2019–21 loans written at 3–4% refinance at 6–8% against lower net operating income and values down 50% or more in some office markets — DSCR failure, maturity default, then extend-and-pretend or loss recognition concentrated in regional and community banks. Extensions are being granted at below-market rates, which delays but enlarges the reckoning.
Transmission. Regional bank capital into local lending contraction — the 1990–91 credit-crunch channel; CMBS investors including insurers; municipal tax bases.
Strongest counter. The most telegraphed wall in history. Lenders have extended for three years, regional-bank net charge-offs clustered at just 29–51bp in Oct 2026 earnings, and losses are arriving slowly enough to be earned against. Multifamily and industrial are not office. Slow bleed, not eruption.
Japan carry unwind
Size. Carry scale is the widest measurement gap on this list — ~$250bn leveraged (BIS) to ~$2.3tn on a broad proxy (Jefferies) [Contested]. BOJ policy rate 1.25%, highest in 31 years.
Trigger. Hawkish BOJ guidance on the terminal rate, or a sharp yen spike through intervention levels starting the buy-back mechanically.
Watch. CFTC yen positioning; TIC Japanese Treasury holdings; USD/JPY realised volatility; long-end JGB auctions.
Mechanism ›Mechanism. August 2024 proved the loop: a 25bp BOJ surprise plus soft US data sent the yen up 12% in under four weeks and forced global deleveraging. The trade rebuilt because the differential remained — and each BOJ hike now lands on a larger stock. Leveraged funds were net short ~102k yen futures contracts in late July 2026 with USD/JPY briefly at a 40-year yen low.
Transmission. Forced selling of US and European assets, including Japanese Treasury holdings at the margin; CLO demand (Norinchukin, the largest single CLO holder at ~$54bn, increased holdings after $9.3bn of losses — a concentrated node); EM carry unwinds.
Strongest counter. Japan is ~90% domestically funded, pension holders are unlevered, and banks sit on over ¥400tn of surplus cash as natural JGB buyers. There is no levered forced seller of LDI scale. Episodic volatility is likely; a structural unwind is a grey swan, and the 2024 episode itself healed in weeks.
Consumer credit tail
Size. Card balances 90+ days delinquent at 13.1% in the NY Fed panel, highest since 2011 — while bank-reported card delinquency fell to 2.85% [Measured, both]. Subprime auto 60+ at series highs (Fitch).
Trigger. Labour-market softening hitting the same lower-income cohort, or ABS funding spreads widening for subprime issuers — historically the kill-switch.
Watch. NY Fed Household Debt & Credit early-stage transition rates; FRED DRCCLACBS; Fitch ABS delinquency indices.
Mechanism ›Mechanism. K-shaped exhaustion. Lower-income households financed post-2021 inflation with revolving credit at 22%+ APRs; early-stage transitions feed serious delinquency with a two-to-three-quarter lag — the 2006 pattern in miniature. Student-loan serious delinquency is back above 10% as repayment restart bites, and BNPL risk is migrating to outside buyers, including private credit and insurers.
Transmission. Direct losses are small versus bank capital. The channel is consumption — bottom-quintile spending retrenchment — and subprime lender and ABS funding.
Strongest counter, and it is strong. The two card series measure different universes. The NY Fed panel shows crisis-level tails; bank data show delinquency falling for five straight quarters and the flow into serious delinquency flat year over year, because the mortgage book — two-thirds of household debt — is clean, fixed at low rates, with large equity cushions. On bank data, the consumer crack already peaked in 2024 and is healing. The tail stress is real and persistent; an aggregate consumer crisis is not in the 2026 data.
China LGFV / property
Size. Local-government plus LGFV debt ~90% of GDP at end-2025, LGFV alone ~51%; LGFV interest coverage below 1.0 [Measured, IMF]. Land-sale revenue has more than halved from peak.
Trigger. A disorderly LGFV default breaking the implicit-guarantee assumption in a weak province, or a policy error at the June 2027 platform-exit deadline.
Watch. LGFV bond spreads by province (Wind/ChinaBond); NBS property investment, starts, and prices; Ministry of Finance land-sale revenue.
Mechanism ›Mechanism. A balance-sheet recession managed by swap-and-extend. Growth drags, PPI deflation is exported through excess capacity, and fiscal capacity is consumed by debt service at exactly the moment stimulus is needed. About 63% of 2026 local bond issuance merely refinanced the past.
Transmission. Mostly slow for the US — commodity and EM demand, multinational earnings, deflationary goods prices. Financial contagion is limited by capital controls and state ownership of banks; losses can be socialised over decades, Japan-1990s style.
Strongest counter. This volcano has been erupting for five years and the state has demonstrated both the tools and the willingness to stretch resolution over a decade. The risk to a US reader is a growth and deflation headwind, not a Lehman moment.
Secondary watch
Nonbank mortgage servicers
Size. Nonbanks service 66% of the loans in more than $9tn of Ginnie Mae / GSE securities, up from 27% in 2014, and originate 76% [Measured, GAO].
Trigger. A delinquency spike forcing servicer advance obligations up precisely when warehouse lenders pull lines.
Watch. Ginnie Mae issuer financials and PTAP usage; MBA servicer advance data; delinquency in Ginnie Mae pools.
Mechanism ›Mechanism. Nonbank servicers must advance principal, interest, tax, and insurance on delinquent Ginnie loans from their own liquidity, funded by short-term warehouse lines that can vanish in downturns — with no federal prudential regulator. GAO found monitoring gaps at Ginnie Mae and FHFA, and the liquidity backstop FSOC recommended in 2024 does not yet exist. A large servicer failure means servicing transfers, borrower harm, Ginnie fiscal exposure, and an origination shutdown for exactly the borrowers with no alternative.
Strongest counter. The sector survived its true stress test in 2020 without a major failure, and capital and liquidity standards have since been raised. Probability is low — but this is the candidate where a failure would most surprise consensus, SVB-style.
Housing lock-in / insurance shock
Size. Insurance premiums up 46% since 2021, Florida averaging $6,800–8,000+ per year [Measured / Estimate]. About 80% of mortgaged owners still hold rates below 6%.
Trigger. A major hurricane season landing on a thin, reinsurance-dependent Florida market, or a large carrier withdrawal.
Watch. Florida OIR carrier actions and Citizens policy count; ICE / Insurify premium indices; FHFA and Case-Shiller prices.
Mechanism ›Mechanism. Not a default volcano — mortgage delinquency is low and equity cushions large. It is a transactions and mobility freeze plus a regional cost shock. Insurance non-renewals can make homes unmortgageable in coastal Florida, Louisiana, Texas, and California, forcing cash-only discounts, while locked-in owners cannot move for jobs — a labour-market efficiency loss that compounds other frictions.
Strongest counter. No leverage, no forced sellers, large equity, and prices flat rather than falling nationally for a year. This is an affordability and regional insurance problem with no plausible path to a national financial crisis on its own within 24 months — though a severe regional event, such as Florida post-hurricane, is a coin flip.
