Shiller CAPE20%
41.4
4.93
—
98th pct since 1881; web 41.4 vs computed 47.5; divergence 6.1 pts - earnings vintage suspect; 10yr real earnings carried from 2023-06-01 (40 months)
Eight valuation, credit and fiscal instruments plus a nineteen-voice sentiment roster, read against the deployment rules.
At composite 4.5+, historically a 40% probability that a 20%+ real decline begins within 36 months (29% for 30%+) - but 4.5+ zones also produced the longest melt-ups. The signal is regime, not timing.
98th pct since 1881; web 41.4 vs computed 47.5; divergence 6.1 pts - earnings vintage suspect; 10yr real earnings carried from 2023-06-01 (40 months)
median ~85%, dotcom peak 146-172%; 200%+ is the top of the historical range
1/CAPE 2.42% - real 10y 1.80% (nominal 5.28% - trailing 10y CPI 3.48%)
real price vs its 120-month average; 91st pct historically
trailing 12m real price change, 65th pct. Froth marker, not a sell signal - the classic top is momentum turning DOWN while valuation stays red
EARLY-WARNING TRIPWIRE. +31bps vs the 2026-09-28 report (WIDENING). Tight = complacent; past 500bps alongside an equity drawdown is the staged-deployment trigger
positive after a post-inversion re-steepening = 3.0 (*) - recessions historically arrive 12-18 months AFTER un-inversion, not during the inversion
federal debt/GDP 123%. ~6% deficits at full employment scores 4.5
THESIS BROADENED from AI/semis to housing and the credit cycle - which is why 36m goes to -3. His mortgage data is the real-economy version of what the HY tile is showing
'The Big Short is about to repeat itself': rates up six straight weeks, mortgage applications -6% w/w, purchase applications -14% y/y, refis -56% y/y, and serious 90+ day delinquencies up for the first time in five months, +19% y/y. Separately questions how fast Nvidia's AI chips lose economic value, citing the 1968 computer-financing cycle, and calls AI infrastructure commitments liabilities in 'hypergrowth mode'
No new statement inside the fortnight, but the tape is validating the April letter and his Norges Bank remarks: private-credit default rates hit a record 6.0% (Fitch, April) and HY OAS has widened 54bps since the last report. 'Credit will perform worse than people expect' is becoming measurable rather than rhetorical - the rhetoric>action gap is closing from the data side, not from him
CNBC: hyperscaler AI spending could hit $1 trillion next year (JPM India conference)
latest sourced statement is June 2026 - inside the 6-month window but not fresh this fortnight
Calls this 'the most expensive market in American history'; a reversion to trend would be closer to a wipeout than a correction, and the odds an AI bubble does not bust are 'slim to none'
Held: Fed commentary, no equity view
Yahoo Finance interview: Warsh's speech 'utterly conventional' - 'tight money until we see a lot better data'
24m cut to -4 and conviction to the maximum: he has put a clock on it
Says the US is 'on the brink' and that if the fiscal path does not change a debt shock could hit within two years
Held: bonds over stocks, a recession surprise ahead; talk only, no positioning
Kitco interview: one hike is not a mistake but four or five would be; 'block your nose and buy' Treasuries; expects bonds to outperform stocks within months
12m trimmed to +1: a 10y near 5% and TIPS ~2.6% compete with equities, equity risk premium ~2.4pp, 'no major near-term upside catalyst'
WisdomTree 'Higher Rates Test Stocks, But Growth Endures': 'If long-term yields decisively move above 5%, it becomes much more difficult for stocks to advance'
Held: probably not one-and-done, but not a long cycle either
Schwab Market Perspective: stocks fell sharply after Warsh's press conference; slow hiking cycles have historically produced milder declines
CUT and GAP CLOSED: reportedly cut AI holdings to ~20% of their level six months earlier (FT, via secondary source), calls AI a likely 'profit bubble' and rates too low
Piper Sandler closed-door meeting (FT): Fed members calling policy restrictive are 'just ridiculous'; 'We are likely in a profit bubble because this AI construction boom will eventually end'
Deployment continued into Q3 under Abel: ~$39.4bn of equities bought across H1, ~$17bn of it Alphabet via the AI-funding secondary, plus a $6.8bn Taylor Morrison homebuilder acquisition closed in July. Full Q3 13F is not due until November. Cash was still $365.5bn at 30 June, so this is deployment, not capitulation - and note he is buying a homebuilder into the exact housing weakness Burry is shorting
Abel's reversal confirmed: 14-quarter net-selling streak broken with ~$20bn net bought, $4.5bn of buybacks against none in 2025
Held: relative preference for bonds, not a call against stocks
Daily Spark 'Yield Levels Starting To Look Juicy': income without excessive risk even if rates rise further
CALLED IT: in August he said the Treasury buyback was financial engineering that left long rates untouched. The 10y then went through 5%. His new line - interest-rate risk becoming equity and credit risk - is precisely what the HY tile is now showing
Expects the 10y to stay around 5% even if oil falls, citing a structural bond supply/demand imbalance; 'volatility will be a feature, and elevated yields will be a feature for the rest of the year', and 'the big danger' is rate risk turning into equity and credit risk
MOVED CAUTIOUS from neutral: own nothing but 1-year T-bills, be cautious on stocks at record levels because higher yields will compress margins
Investment Outlook / FT op-ed 'Don't own bonds and be cautious with stocks': 'be cautious with stocks at record levels as higher yields over time will contract profit margins'
No new memo inside the fortnight, but his private-credit memo is the live thesis: the US private-credit default rate hit a record 6.0% in April (Fitch) and the Proskauer index ran 2.73% in Q1 against 1.84% two quarters earlier - while loan spreads were still tightening. Exactly the 'not being paid for the risk' setup he described
Oaktree memo 'Shall We Repeal the Laws of Economics - Part III': buybacks are an ice pack on a fever; long rates stay elevated
Held: post-hike newsletter calls the Fed a sideshow next to tech-driven deflation (paraphrase only)
Wood newsletter (via Crypto Briefing): the hike is a sideshow; innovation pushes inflation toward 0-1%
Held: the loudest bull, now openly split from Yardeni
CNBC Closing Bell: S&P could 'easily be above 8,200 by the end of the year' as long as the AI trade isn't extinguished; Q4 'one of the biggest rallies'
Held: formal note on the cut, 8,400 pushed to mid-2027, 70/30 Roaring-2020s/bear odds
'US MARKET CALL: Investors Curbing Their Enthusiasm...': 'Our new year-end target is 7,900'
RAISED from +1: not worried about an AI bubble; rising real yields reflect AI capex and growth, not a bond crisis
Cernobbio interview (Bloomberg video 'Roubini Says He's Not Concerned About an AI Bubble'): stocks rise with growth, though 'some corrections' possible
HIKED 25bps to 3.75-4.00% on 2026-09-16, unanimously - the first increase since 2023 - with the dot plot pointing to one more hike this year (median 4.1% end-2026, no change in 2027, cuts in 2028-29). the headwind became an actual tightening. Warsh: inflation has been 'too high for too long' and the FOMC's confidence standard 'has not been satisfied'. Core PCE has run above 3% every month of 2026
The single most important event since the last report: a unanimous 25bp HIKE to 3.75-4.00%, the first since 2023, with one more signalled. My Fed block is no longer a headwind, it is active tightening.
My tripwire, and it is moving: 270bps in August to 324bps on 1 October, widening nearly every session since 25 September. Watch this series, not the index, for the next three weeks.
'If the 30-year must trade at 5.5%, that isn't a crisis. It is an invoice.' The cleanest statement of why my fiscal tile and my rates tile are the same tile.
Defaults rising while loan spreads were still tightening is the Marks memo stated as data - and the leading edge of the public HY widening my tripwire just caught.
The mortgage data behind his housing turn - refis -56% y/y, serious delinquencies +19% - is the household-side mirror of a 5.28% 10-year, and it is testable week to week.
Composite 4.33 (frothy / hold dry powder). Base DCA continues regardless of score. Above 3.5, so the DCA-doubling rule stays dormant; it arms at 3.5 (+0.83 away). Hold dry powder. Direction matters more than the level here: spreads widened 31bps since 2026-09-28, and the 500bps staged-deployment trigger is 176bps away - roughly 5.7 more moves of that size. Have the tranche sizes decided before it gets there, not after. HY spreads at 324bps are still inside 500bps, so the staged dry-powder trigger is not armed. Credit is the tripwire: it goes first. And read the top-signal line above before acting on the composite alone: a flat composite with falling momentum and widening credit is not the same market as a flat composite with both rising.