Theory Watch

If the theory is true, it should start showing here.

Every era has its big calls — a famous investor lays out a possible future, and the arguing starts. We don’t argue. For each theory we write down the steelman (the strongest honest version), then pre-register the evidence: the specific, measurable places it would show up first if it’s right. The gold zone on each chart is written down in advance — when the line enters the zone, the telltale reads SHOWING. No moving the goalposts, in either direction.

Two things we hold ourselves to: “not yet” is not “never” — being early and being wrong look identical for a while. And “showing” is not proof — it’s evidence, and we say so plainly either way. Fair to the person, sharp on the claim.

Gordon Chang · 2001–present
0 of 3 showing
“China is collapsing.”
Not a lone voice — also on record: Kyle Bass (the banking-system short), Jim Chanos (“treadmill to hell”, the original short thesis), Hal Brands & Michael Beckley (“peak China”). Michael Pettis argues the careful version — a long structural slowdown, not a collapse — and deserves to be quoted as saying that, not lumped in here.

The steelman. The strong version isn't “China has problems” — it's that the problems compound into a rupture: a property bust too big to paper over, local-government debt that can't be rolled, deflation, a shrinking population, and capital trying to leave. If THAT is what's happening — rupture, not slowdown — it can't hide, because it has to show up in three prices at once: China's own equities, the currency Beijing defends, and the metal China buys half the world's supply of. One tripping is a bad year. All three together is the theory.

If it’s true, it starts showing here:
China's own market cratersNOT YET
the zone · -30%’22’24’26-630

Not a bear market — a collapse-sized markdown in Chinese large-caps (MCHI), the zone starting well past ordinary pain at -30%.

MCHI (China large-caps), distance from its 3-year high · latest -21.77%
zone record: China '15 ✓ · Q4 '18 ✓ · 2022 bear ✓ · trips outside the analogs since ’11: 5
The defended line breaksNOT YET
the zone · 7.5’22’24’2678

Beijing manages the yuan hard; a genuine rupture overwhelms the defense. Watch the yuan weakening past the line they've fought for — the zone starts at 7.5.

yuan per dollar — higher means weaker yuan · latest 6.8
no analog on record — if this trips, it’s genuinely new information
The world's factory stops buyingNOT YET
the zone · -25%’22’24’26-260

China buys roughly half the world's copper. A rupture in Chinese demand shows up in the global price falling hard from its recent high — the zone starts at -25%.

global copper price, distance from its 2-year high · latest 0.0%
zone record: GFC ✓ · China '15 ✓ · COVID ✓ · trips outside the analogs since ’00: 10

Fair is fair: This is the longest-running collapse call in finance — the book was published in 2001, and being early for a quarter century is its own verdict. But fair cuts both ways: the property bust, the deflation, the demographics are real, and “it hasn't happened yet” is not proof it can't. That's exactly what the zones are for.

So far: Not showing. The tape isn't voting for this one yet — and 'not yet' is not 'never'.
Dario Amodei · 2025
1 of 5 showing
“AI will wipe out a huge share of entry-level white-collar jobs — fast.”
Not a lone voice — also on record: Geoffrey Hinton (mass job loss), Jim Farley (Ford: half of white-collar work), Andy Jassy (Amazon: the corporate workforce shrinks).

The steelman. The CEO of an AI lab said the quiet part with numbers: AI could eliminate as much as half of entry-level white-collar work within one to five years, pushing unemployment into the double digits. The serious mechanism: junior analysis, junior code, junior support is exactly what the models do first, so companies stop hiring at the bottom rung before they ever fire anyone — the door closes before the layoffs make headlines.

If it’s true, it starts showing here:
The temp layer sheds firstNOT YET
the zone · -5%’22’24’26-16-2

The unexpected-but-necessary one: before any full-timer is touched, the CONTRACT layer goes — temp staffing is the flexible edge of white-collar work, and it thins first, quietly. Watch temp-help employment falling below its 2-year high — the zone starts at -5%.

temp-help employment, distance from its 2-year high · latest -3.81%
zone record: dot-com ✓ · GFC ✓ · COVID ✓ · trips outside the analogs since ’00: 5
The young get shut out firstNOT YET
the zone · 4.5 pts’22’24’2624

Entry-level is the claimed epicenter, so the gap between young-worker unemployment (20–24) and overall unemployment should blow out past its normal couple of points — the zone starts at 4.5 pts.

unemployment rate, 20–24 year olds minus everyone · latest 3.03 pts
zone record: GFC ✓ · COVID ✓ · trips outside the analogs since ’00: 5
Knowledge-work headcount shrinksSHOWING
the zone · -5%’22’24’26-80

Not a market wobble — actual jobs gone: information-sector employment (tech, media, data) falling well below its recent high while the rest of the economy holds — the zone starts at -5%.

information-sector jobs, distance from their 3-year high · latest -7.56%
zone record: dot-com ✓ · GFC ✓ · trips outside the analogs since ’00: 5
The hiring door closesNOT YET
the zone · 6.0M’22’24’26612

The mechanism is hiring freezes before layoffs: total job openings sliding below the roughly seven million that counted as normal before the frenzy — the zone starts at 6.0M.

US job openings (JOLTS), millions · latest 7.59M
zone record: GFC ✓ · COVID ✓ · trips outside the analogs since ’00: 5
The machine signature: productivity boomsNOT YET
the zone · 3.0%’22’24’26-23

The separator. If the machines are doing the work, output per hour ACCELERATES while hiring stalls — job losses plus a productivity boom is AI; job losses without one is just a recession wearing an AI costume. The zone starts at 3.0%.

productivity growth, year over year (long-run normal is under two) · latest 2.65%
zone record: dot-com ✓ · GFC ✗ · COVID ✓ · trips outside the analogs since ’01: 17

Fair is fair: Amodei runs an AI company — he was warning about his own product, which cuts against interest and earns the claim a fair hearing. He also said it isn't inevitable. Plenty of others made vaguer versions; his is dated and quantified, so his is the one a chart can watch.

So far: Partially showing (1 of 5). Worth honest attention — watch the remaining telltales.
Michael Burry · 2025
0 of 4 showing
“The AI trade is a bubble, and the unwind will be brutal.”
Not a lone voice — also on record: Jim Covello (Goldman’s “too much spend, too little benefit”), Torsten Slok (Apollo: bigger than 1999), David Cahn (Sequoia’s $600B question), Jim Chanos (the data-center build-out).

The steelman. The man who called the housing crash spent late 2025 warning that AI capex and the handful of mega-caps carrying it had gone parabolic — and disclosed bearish positions against marquee AI names. The serious version isn't 'AI is fake'; it's that the SPENDING is running far ahead of the cash it returns, and when that gap closes, the stocks that led the market up lead it down.

If it’s true, it starts showing here:
The four checkbooks slowNOT YET
the zone · 0%’22’24’26-272

The separator, and the mechanism itself: the entire AI trade is funded by the capital spending of Microsoft, Alphabet, Amazon and Meta. Stocks falling while the checkbooks still grow is a valuation argument; the checkbooks themselves CLOSING is the unwind engaging. The zone: combined capex growth turning negative — under 0%.

the four hyperscalers' combined capital spending, growth of the trailing-year total · latest 71.5%
zone record: GFC ✓ · trips outside the analogs since ’05: 15
Growth leadership breaksNOT YET
the zone · -10%’22’24’26-180

The Nasdaq has led the market for years. If the air is coming out, that leadership cracks first: the QQQ-to-SPY ratio falls well below its 12-month high — the zone starts at -10%.

distance of the QQQ/SPY ratio from its 12-month high · latest -2.24%
zone record: dot-com ✓ · GFC ✓ · 2022 bear ✓ · trips outside the analogs since ’00: 6
A real bear in the leadersNOT YET
the zone · -20%’22’24’26-340

Corrections happen every year. A popped bubble is different: the index that owns the AI trade drops into a true bear — the zone starts at -20%.

QQQ, distance from its all-time high · latest -4.66%
zone record: dot-com ✓ · GFC ✓ · 2022 bear ✓ · trips outside the analogs since ’00: 6
Credit starts to careNOT YET
the zone · 5.0’23’25’2635

Equity wobbles are opinions; credit stress is money. A real unwind shows up as high-yield spreads pushing through 5.0.

ICE BofA US High Yield spread, points over Treasuries · latest 2.75
zone record: dot-com ? · GFC ? · China '15 ? · COVID ? · trips outside the analogs since ’23: 0

Fair is fair: Burry was early on housing too — early and right can look identical to wrong for years. We're not grading his timing; we're watching for his evidence.

So far: Not showing. The tape isn't voting for this one yet — and 'not yet' is not 'never'.
Ray Dalio · 2025
0 of 3 showing
“A debt spiral: money flees bonds and the dollar, and gold is the refuge.”
Not a lone voice — also on record: Paul Tudor Jones (the “debt bomb”), Stan Druckenmiller (spending like “drunken sailors”), Jamie Dimon (a “crack in the bond market”), Niall Ferguson (interest passing defense).

The steelman. Dalio's book-length warning: deficits this size eventually force a choice — default, austerity, or print. History says print. In that world the classic refuge (government bonds) becomes the problem, the dollar sags, and hard assets — gold first — absorb the flight.

If it’s true, it starts showing here:
Gold outruns stocksNOT YET
the zone · -5%’22’24’26-420

Not gold merely rising — gold beating equities, with the gold-to-stocks ratio pressing a 3-YEAR high (reworked 2026-07-04 from a 12-month lens: the old zone geometry only fired at extremes; change logged publicly, as promised). Showing = the ratio within a few points of that high; the zone starts at -5%.

distance of the GLD/SPY ratio from its 3-year high · latest -30.53%
zone record: GFC ✓ · euro '11 ✓ · COVID ✓ · trips outside the analogs since ’04: 6
Nowhere to hideNOT YET
the zone · -10%’22’24’26-2016

The debt-crisis signature: stocks AND long bonds falling together, so the classic 50/50 refuge mix bleeds — its 6-month return in the zone below -10%.

6-month return of a 50/50 SPY + TLT mix · latest 5.07%
zone record: GFC ✓ · COVID ✗ · 2022 bear ✓ · trips outside the analogs since ’02: 1
The dollar sagsNOT YET
the zone · -5%’22’24’26-90

Flight from the currency itself: the broad dollar index falling well below its 12-month high — the zone starts at -5%.

broad trade-weighted dollar, distance from its 12-month high · latest -0.85%
no analog on record — if this trips, it’s genuinely new information

Fair is fair: Dalio has warned early before, and 'eventually' is doing heavy lifting in any debt argument. The telltales below are his own claimed signature, not our paraphrase of doom.

So far: Not showing. The tape isn't voting for this one yet — and 'not yet' is not 'never'.
The crowd, about Warren Buffett · 2024–2025
0 of 2 showing
“Berkshire's record cash pile means the greatest investor alive is bracing for a crash.”

The steelman. Berkshire let cash grow to historic size while trimming big equity stakes. The popular reading: when the best capital allocator of all time can't find anything worth buying, prices are too high — and he's holding dry powder for the moment they aren't.

If it’s true, it starts showing here:
The waiting has to winNOT YET
’24’25’26100142■ the market (SPY, dividends in)■ T-bills (the cash pile)

Cash earns the bill rate while it waits. The theory is vindicated the day the market line drops below the T-bill line — everything above that gap is what the waiting has cost so far.

growth of 100 since mid-2024 · latest 137.64%
The price he'd load atNOT YET
the zone · -20%’22’24’26-240

The moment the thesis pays: a real markdown. The market's distance from its all-time high reaching the zone below -20%.

SPY, distance from its all-time high · latest -3.22%
zone record: dot-com ✓ · GFC ✓ · COVID ✓ · 2022 bear ✓ · trips outside the analogs since ’00: 4

Fair is fair: To be fair to the man himself: Buffett never predicted a crash — he has said, for decades, that he simply buys when things are cheap and waits when they aren't. The prediction belongs to the crowd's reading of his cash, so that's the claim we watch.

So far: Not showing. The tape isn't voting for this one yet — and 'not yet' is not 'never'.

Priced-in: what the market is claiming

A famous price is a famous prediction. We watch where it would crack.

These aren’t calls, long or short — they’re the market’s own claims, written down, with the crack lines pre-registered. Fundamentals update on filing days (earnings day is the day the claim gets tested); relative strength updates daily. Nothing here suggests a trade, in either direction. Educational, not advice.

the crowd, via the price · 2026
cracks: 0 of 3
“NVDA: AI demand compounds for years, and no one competes away the margin.”

The steelman. At a seventy-plus-percent gross margin and this multiple, the price asserts two things at once: the AI buildout keeps accelerating, AND Nvidia keeps taking almost all of the profit from it. Both have been true for three years running. The price says they stay true.

The price is the claim. If the story cracks, it cracks here first:
The demand leg cracksNOT YET
the zone · 25%’22’24’26-12208

The claim needs hypergrowth. Trailing-year revenue growth slowing under 25%.

NVDA revenue, growth of the trailing-year total · latest 70.68%
The moat leg cracksNOT YET
the zone · 65%’22’24’265676

The claim needs the margin. Gross margin slipping under 65%.

NVDA gross margin (a crack here = competition arriving) · latest 74.1%
The market votes noNOT YET
the zone · -20%’22’24’26-530

Relative strength versus the market breaking well below its 12-month high -- -20%.

NVDA/SPY ratio, distance from its 12-month high · latest -15.79%

Fair is fair: This isn't a call that the price is wrong -- prices have been 'impossible' and right before. It's a list of what has to KEEP being true, and where a crack would show first.

So far: No cracks. The story is delivering -- so far, the price is earning itself.
the crowd, via the price · 2026
cracks: 0 of 3
“MU: this time, memory isn't cyclical.”

The steelman. Micron is priced as an AI-infrastructure toll booth. But memory has been a brutal boom-bust business for forty years -- the price only works if AI demand has genuinely killed the cycle. That's the claim: the most cyclical business in tech just became a compounder.

The price is the claim. If the story cracks, it cracks here first:
The cycle comes homeNOT YET
the zone · 30%’22’24’26-1573

Memory downturns always arrive through price -- gross margin sliding under 30%.

MU gross margin (the cycle's fingerprint, every single time) · latest 72.6%
Growth goes negativeNOT YET
the zone · 0%’22’24’26-49167

Not a slowdown -- the cyclical signature itself: trailing-year revenue shrinking, 0%.

MU revenue, growth of the trailing-year total · latest 166.98%
The market votes noNOT YET
the zone · -30%’22’24’26-510

Relative strength breaking hard from its high -- -30%.

MU/SPY ratio, distance from its 12-month high · latest -19.61%

Fair is fair: Every memory cycle top has had a reason the cycle was dead. Sometimes for years. The zone doesn't argue -- it waits.

So far: No cracks. The story is delivering -- so far, the price is earning itself.
the crowd, via the price · 2026
cracks: 1 of 2
“PLTR: worth years of flawless hypergrowth, paid today.”

The steelman. Palantir carries one of the richest sales multiples any large company has sustained. The strong version: government-grade AI software is winner-take-most, and Palantir is the winner, so decades of growth deserve to be paid for now.

The price is the claim. If the story cracks, it cracks here first:
An ordinary quarter arrivesNOT YET
the zone · 25%’22’24’261668

Perfection pricing needs hypergrowth. Trailing-year growth under 25%.

PLTR revenue, growth of the trailing-year total · latest 67.71%
The market votes noCRACKED
the zone · -25%’22’24’26-700

Relative strength breaking from its high -- -25%.

PLTR/SPY ratio, distance from its 12-month high · latest -42.32%

Fair is fair: A perfection multiple isn't proof of a bad business -- Palantir's business has repeatedly beaten the skeptics. It's proof there is no room for an ordinary quarter.

So far: Cracks forming (1 of 2). The story has stopped delivering on a line it needs.
the crowd, via the price · 2026
cracks: 2 of 3
“TSLA: an AI and robotics company that happens to make cars.”

The steelman. Tesla's multiple has never been a car multiple. The claim is a category claim: robotaxis and humanoid robots turn a manufacturer into a software-margin platform. If that's true, the financials have to MIGRATE -- away from carmaker economics, visibly, in the filings.

The price is the claim. If the story cracks, it cracks here first:
Carmaker growthCRACKED
the zone · 10%’22’24’26-360

Platform stories grow; carmakers cycle. Trailing-year revenue growth under 10%.

TSLA revenue, growth of the trailing-year total · latest 2.25%
Carmaker marginsCRACKED
the zone · 5%’22’24’26416

The category claim lives in the margin: net margin at or under 5%.

TSLA net margin (platform economics vs auto economics) · latest 3.9%
The market votes noNOT YET
the zone · -30%’22’24’26-62-3

Relative strength breaking from its high -- -30%.

TSLA/SPY ratio, distance from its 12-month high · latest -26.25%

Fair is fair: Tesla has embarrassed the 'it's just a carmaker' argument more than once. The zones don't take sides; they mark where the category claim stops being visible in the numbers.

So far: Cracks forming (2 of 3). The story has stopped delivering on a line it needs.
the crowd, via the price · 2026
cracks: 0 of 2
“LLY: GLP-1 is a decade of compounding, not a product cycle.”

The steelman. Lilly is priced like a tech platform because the market believes obesity drugs are the largest pharmaceutical franchise ever -- with years of supply-constrained growth ahead and competition permanently behind. The biggest non-AI claim on the board.

The price is the claim. If the story cracks, it cracks here first:
Growth normalizesNOT YET
the zone · 20%’22’24’26-647

The platform claim needs franchise-scale growth. Trailing-year growth under 20%.

LLY revenue, growth of the trailing-year total · latest 47.44%
The market votes noNOT YET
the zone · -25%’22’24’26-35-0

Relative strength breaking from its high -- -25%.

LLY/SPY ratio, distance from its 12-month high · latest -1.3%

Fair is fair: The franchise IS delivering -- guidance was raised again this year. The zones watch for the two things pharma history says end these runs: growth normalizing, and competition compressing price.

So far: No cracks. The story is delivering -- so far, the price is earning itself.
the crowd, via the flows · 2026
cracks: 1 of 1
“IBIT: institutional adoption killed the bitcoin cycle.”

The steelman. The ETF era's claim: now that bitcoin lives in brokerage accounts and model portfolios, the old four-year boom-bust -- minus eighty percent, twice a decade -- is over. Permanent flows, shallower drawdowns, a maturing asset.

The price is the claim. If the story cracks, it cracks here first:
The old cycle reassertsCRACKED
the zone · -40%’24’25’26-50-2

Adoption-era drawdowns are claimed to be shallower. The zone: a cycle-sized markdown from the high -- -40%.

IBIT, distance from its 12-month high (the pre-ETF cycle's signature) · latest -50.47%

Fair is fair: Two prior 'this cycle is different' eras ended at minus eighty. This one genuinely has different plumbing. The zone is set where the OLD pattern would reassert itself -- not at ordinary volatility.

So far: Cracked on every pre-registered line.

Telltales recomputed from the data 2026-07-04. Levels are pre-registered and only change with a public note explaining why. Educational, not advice.

When a telltale trips, the wringer covers it.
The free read: most weeks, one famous claim — a system, a guru, a “sure thing” — tested honestly and delivered win or lose. (The Risk Reducer’s live read and the day-it-flips alert are the members’ side.) No forecasts. No spam. Unsubscribe anytime.

When a theory starts showing, you’ll hear it from the tape — not from us shouting.

Most weeks we put a famous claim through the wringer. Members also get the Risk Reducer’s live read and the alert the day it flips. About a coffee a week.