Meb Faber #654: It's a Bubble, So Buy It

Contents

1. A ruler that only looks at price
2. Why he would rather look only at price
3. From “it’s a bubble” to “so buy it”
4. The conditions under which the call breaks
5. What he admits the ruler cannot do
6. A few other things worth keeping
7. Where to look next
8. One thing to take away

I woke to only the pillow and the mat; the mists and glowing clouds I had been walking through were gone.

—— Li Bai, “A Dream Journey to Tianmu, on Parting” (Tang dynasty, c. 745; translation mine)

In The Meb Faber Show #654, published 16 October 2026, Dirk Willer — head of global macro asset allocation and emerging market strategy at Citi — says something that sounds self-contradictory: guys, it’s great, it’s a bubble, let’s buy it. His definition is borrowed from the asset manager GMO: when a market’s real price runs more than two standard deviations above its long-term trend, call it a bubble. By that ruler, US equities entered bubble territory around the middle of 2025. He also gives the condition that voids the call: if you believe earnings growth over the next five years will look like the last two or three, then it probably isn’t one.

A ruler that only looks at price

Willer says the word is argued over hard inside Citi, and that he runs a large quant team — and if you want to backtest something, you first need a definition that isn’t a feeling. So they took GMO’s: compare a market’s real price, stripped of inflation, with its own long-term trend, and anything more than two standard deviations above it is a bubble.

That ruler earns him a lot of flak, and the complaint is always the same — it’s price-based, not valuation-based. He thinks that is the point.

Why he would rather look only at price

This is the stretch of the episode I replayed, because his reasoning turns around and hits the valuation camp.

Bubbles show up in earnings too. Some of the gross margins the high flyers are reporting now are unsustainable, which is the first layer. The second is more interesting: when something is in scarce supply, people order a lot more than they actually want, and companies then use that stack of double bookings to underwrite their own earnings estimates. Price goes up, earnings get inflated alongside it, and the multiple ends up looking almost reasonable.

Stack those two layers and the flaw in a valuation definition is visible: at the moment you should be most alert, the number it hands you reads “still fine.” A price definition doesn’t care whether the bubble shows up in earnings or in the multiple. It only asks how far you are from trend.

He leaves himself one fallback: if you insist on valuation, the Shiller CAPE — the multiple computed on ten-year average earnings — tells you the same thing. Both rulers point the same way.

From “it’s a bubble” to “so buy it”

Here is where most people’s first reaction lands, mine included: if you’ve called it a bubble, shouldn’t you be cutting?

His answer sits inside the question of what separates a bubble from a bull market. His separation: in a bubble, you give back a fair chunk of the money you make along the way. So the call tells you how big the payback will be, and what to do today has to be decided separately. He is blunt about it — how big depends on how far the bubble inflates first, and he thinks mid-2025 levels are reachable again. That leaves three instructions: don’t be over-levered, keep some hedges on the book, know that it lands badly.

Then he sizes the payback. In 2000, US equity market cap was a bit above 100% of GDP, and the bursting bubble delivered a shallow recession. Today that ratio is 240%. The market is a bigger lever on the economy, so the wealth effect bites harder, and he expects a larger recession with the Federal Reserve reacting to it — rates back to zero is not implausible to him, he thinks the Fed would buy the back end of the curve too, and gold goes straight up.

What struck me is that he puts the bullish view and the bearish view on two different clocks and then admits both arrive. He even shortens his own bullish horizon out loud: long-dated real yields are around 3.3%, and he asks whether the S&P 500 can beat 3.3% real if you believe the bubble eventually bursts. His own answer is maybe not. Sell-side strategists rarely talk that way.

The conditions under which the call breaks

Three of them, all his own.

The first is earnings. If you believe the next five years of earnings growth will replicate the last two or three, it isn’t a bubble. That sentence is the live wire running through the whole argument — two standard deviations measures deviation from a long-term trend, and whether a new earnings level rewrites the trend itself is a question the ruler doesn’t answer.

The second is liquidity. Withdrawal of liquidity is the main agent that pops bubbles, and 2000 is the cleanest case: the Fed hiked 175 basis points, and before that, to cover the Y2K bug, it had pushed a lot of liquidity into the market in the fourth quarter of 1999 — part of why the Nasdaq rose 50% in that one quarter. Pull it back out in the first quarter of 2000 and the market peaks in March. His line: the first Fed hike is usually not a problem, the last one is. And first and last coinciding has happened once, in 1997, followed by the Russian and Asian crises. On that framework, this Fed is some distance from popping anything.

The third is sentiment. Bubbles usually come with aggressive sentiment and retail investors jumping in with both feet near the end — Korea just ran that play, up 80% in a quarter, driven mostly by highly levered local retail. In the US, the American Association of Individual Investors survey is currently bearish; people are not excited. As for breadth narrowing to a handful of names, he says that is a feature of this kind of market, with everyone chasing the same shiny thing.

What he admits the ruler cannot do

It gives you no timing. The historical bubbles the definition finds vary in both length and extent — they can run for years, or end quickly as in 1929.

The ideal is to trade in line with fundamentals: work out when compute goes into overcapacity, note that stocks tend to peak six to nine months before fundamentals do, and turn cautious nine months ahead. Easier said than done. His own fundamental analysts think compute overcapacity is far off, while his read on human behaviour is that the bigger the opportunity the more we overcapitalise it, so overcapacity arrives eventually. He doesn’t reconcile those two. He leaves them both on the table.

So he built a separate trigger, which they call the generals indicator. Every 1 January you take the seven largest names in the S&P 500 as the generals leading the market; when the bubble indicator is switched on and four of those seven break down below their long-term moving average — from his description, the 200-day — you get out. He says it saves a lot of money in historical bubbles, and it pulls him out six to nine months earlier than turning fundamentals would.

The structure is what I keep coming back to. The definition says what state you are in; the indicator says when to act; he keeps those two jobs apart. Where people get stuck is collapsing them into one sentence — I think it’s too expensive, so I’m leaving. Expensive is a state, leaving is an action, and the trigger in between has to be supplied by you, in a form someone else could check.

A few other things worth keeping

• The level of rates barely matters for equities; the volatility of rates does. They watch the MOVE index, the implied volatility of the Treasury market — once that jumps, equities start to care. The MOVE index has moved over recent weeks while the S&P went roughly sideways, which he reads as impressive price action.
• Decompose the long-end sell-off and all the work was done by real rates: inflation expectations never moved, and asset swap spreads never moved either. The first means the market never said the Fed was losing credibility; the second means it never priced this as a fiscal problem. The market is saying growth will be strong so rates have to be higher — and that version of higher rates is one equities can keep ignoring while they rise.
• November and December seasonality is stronger in good years. His explanation: what you forecast for next year leans on what happened this year, so a profitable year gets you positioned early for another one, and there is less tax-loss selling in those months.
• Brazil is a rare idiosyncratic play. Most emerging markets are a two-factor model of oil and the AI story; Brazil has its own. The post-election government is expected to be market friendly, and if the fiscal tightens, those 9–10% real rates have a long way to fall. His most memorable lesson is also Brazilian: in 2002, when Lula first won, the real strengthened aggressively almost from the day he came in — and the S&P bottomed that same day. What was moving was the global cycle. The lesson he drew: emerging markets are the tail being wagged by global markets, with global forces roughly 70% of the game.
• Chinese GDP growth and Chinese equity returns don’t line up. Compare equity markets and earnings per share against GDP growth across countries and China shows no correlation, while India does. His guess is that stakeholders other than shareholders take some of the spoils, plus hypercompetitive local markets. So he grants that Chinese AI is both competitive and cheap, and still doesn’t believe buying the index gets him those factors.
• TIPS carry a higher Sharpe ratio than nominal bonds, so a permanent bond holder can own TIPS and lever them a little. They remain a duration product: in a stagflation scenario they work only if the Fed sits still. Inflation plus hikes and all bonds get sold.
• Copper is still a valid signal. They trade it off a global data change index — copper’s Sharpe ratio is much higher when that momentum is rising, and it is rising now, with European data improving too. The copper-gold ratio stopped working because of the denominator, with gold on a tear.

Where to look next

• The Meb Faber Show #654 (16 October 2026), guest Dirk Willer
• Dirk Willer, Trading Global Macro Markets (published spring 2026) — hundreds of charts and written-down rules
• GMO’s two-standard-deviation definition of a bubble; Robert Shiller’s public CAPE series
• The American Association of Individual Investors weekly sentiment survey; the MOVE index
• The US Treasury’s quarterly refunding statement — he flags the November one as a chance to shorten duration of issuance and sell more bills

One thing to take away

Judging that something will break and deciding to leave today are two different questions. Collapse them and you walk out while there is still a harvest, or you stay past the point where you can explain why.

Something I’ve tried: take one thing you know won’t hold, and it needn’t be an investment — a relationship, a role, a flat, a habit you are propping up. Write two sentences. The first says what you are still there to collect. The second says what signal gets you out, and that signal has to be a fact someone else could see — “two months without one real conversation,” “passed over for the list a third time” — rather than “when it starts to feel wrong.” I’ve done this a few times, and the ones where I couldn’t write the second sentence are the ones I dragged out longest and paid most for.

Raw Markdown

This article is an educational discussion of investment method. It is not advice to buy or sell any individual security, offers no target prices, and does not analyze any current holding. Investing carries risk; make your own decisions or consult a qualified professional.