The Market Already Ate the Good News: Notes on a Fight About 2027

Be the first to feel the world’s sorrows, the last to enjoy its pleasures.

—— Fan Zhongyan, Record of the Yueyang Tower (Northern Song, 1046; translated by the author)

Fan Zhongyan was describing moral character. He also, accidentally, described a market: it grieves ahead of you, celebrates ahead of you, and by the time the thing actually happens it has moved on to something else.

What this episode is about

The August 25, 2026 episode of The Compound and Friends spends more than forty minutes on a single idea: the stock market is an anticipation machine.

The trigger is an op-ed. Bill Dudley, formerly of the New York Fed, wrote in Bloomberg listing five reasons the bubble bursts by the end of 2027. That combination is rare — a recently departed central banker not only calling a bubble but dating its end. One host thinks that’s remarkable. The other spends the segment picking it apart, not because he disagrees with the conclusion, but because he already knew every piece of the argument.

From there: seven things worth watching into Nvidia’s print, why Airbnb needed six years to dig out of its IPO valuation, how far the product-launch spectrum has drifted toward the casino, and a mystery chart neither the audience nor most listeners would guess.

The highlights

1. Stocks price the future, not the present. This is the spine of the episode. Bond investors want their money back plus interest, so they don’t want surprises. Equity investors want a return on capital, so price follows outlook. This year’s capex and earnings are, in one host’s words, possibly as good as anything we’ll see in our lifetimes — and the market priced that in back in January and February. Which is why companies posting the best contracts and earnings of their lives can sit in deep drawdowns. People seeing that usually conclude one of two things: the market is wrong, or I don’t understand what’s happening. Neither has to be true. The market can be right and you can understand it perfectly — you’re just looking at different points in time.

2. The bear case is not really about valuation. The opening indicators — a CAPE of 41, a Buffett indicator at 240% — are table-setting. The actual chain of reasoning is: what matters for growth is the change in investment, not its level. The 2026-over-2025 increase is enormous and 2027 is unlikely to match it. As investment growth slows, the earnings growth of the suppliers falters, and the multiple people pay for decelerating growth shrinks. So you get hit twice. The unpleasant part: none of this requires earnings to fall. “Not as good as before” is sufficient.

3. Whether that argument is worth anything is a separate question. The rebuttal is blunt: this is the consensus bear case, and there is nothing in it I don’t already know. He also flags a structural problem with CAPE — averaging today’s earnings with 2017’s treats two different worlds as one. And on the bubble word itself: a company growing earnings 99% trading in the low twenties on forward earnings, Meta at fifteen — those aren’t bubble numbers. His standard is worth stealing: if you’re going to call your shot, tell me something I don’t know. He’d be entirely unsurprised if 2027 is a hard year. The issue isn’t the conclusion, it’s the absence of new information.

4. If Nvidia has a bubble, it’s the most innocent kind. The useful distinction here is between a valuation bubble, an earnings bubble, and a pull-forward. He argues for the third: everyone is worried about compute, worried about getting their slice, worried about not delivering what they’ve contracted to customers — so they stockpile. That is a completely different illness from “everyone was stupid and bought things nobody needs.” The supporting analogy is good: build a steel mill and you’re likely running much of the same equipment twenty years on. Data centers aren’t like that — chips are roughly half the build cost, and today’s GPUs are very unlikely to still be in service a decade out. So even if new construction halves, replacement demand on what’s already built doesn’t go away.

5. Semis are cyclical, and that’s the real reason the multiple compressed. Why does a company growing quarterly earnings 99% trade below the market multiple? Three layers: competition from custom silicon (the hyperscalers building their own), investor fatigue and saturation (it’s already 7% of the S&P — if you’re not an indexer, do you want to be 9%? 10%?), and sheer size, which the market struggles to digest at five trillion dollars. But the deepest layer is the first point: three years of de-rating is the investing public acknowledging these are cyclical businesses. A historical aside worth keeping: software has long carried higher multiples than semiconductors precisely because software is less cyclical and can outgrow a cycle in a way hardware historically hasn’t. Maybe not this time — but that’s a claim requiring proof.

6. The starting multiple governs your return more than the growth rate does. The most practically useful stretch of the episode. Airbnb went public in December 2020, unprofitable, at 40 times sales — and needed six years to grow out of that hole. A cited study: buy above 50x earnings, and even with 15–20% growth over the following five years, the average five-year annualized return is -2%. Cisco really did grow 20% a year. Microsoft really did compound earnings through the 2000s. Both stocks went nowhere for a long time. You were right on the fundamentals. You paid too much.

7. The product shelf keeps sliding toward the casino. The hosts had Claude build what they call a “capital cooker” — a ladder from most boring to most degenerate: cash and T-bills → target-date funds → plain beta → factors and sectors → engineered outcomes (buffer funds, option overlays) → crypto and crypto treasuries → high-distribution option-income products → leveraged ETFs → gambling ETFs (event contracts, even league-season futures) → zero-day and intraday leveraged products. The newest rung: single-stock ETFs resetting leverage hourly, and an interface for betting on where a crypto price lands in the next fifteen minutes. One host reaches for Vegas Vacation — the sleaziest casino downtown, no blackjack, no roulette, just flip-a-coin and pick-a-number. He isn’t against any of it existing. He just wants listeners to hear one thing: nobody is sitting in a lab coat asking whether this helps investors. The boardroom question is “will people buy this?” Those are not the same question.

Going further

”The numbers were spectacular and the stock fell. Am I being played?”

This is probably the most common unanswered question retail investors carry. A company posts 97% revenue growth and 99% earnings growth, and the stock does nothing — or drops. The instinctive explanations are that someone is distributing into you, or that the market has lost its mind.

The episode offers a third, and it’s more reliable than either: the price already paid for that.

One layer down. The question is never “is this company good.” It’s “is it better or worse than what was already expected.” Those answers frequently diverge. A company expected to print $92 billion and 97% growth that prints $95 billion and 99% has, in absolute terms, made history — and in expectation terms, merely met the bar. Meeting the bar, at a price that already assumes the bar is met, shouldn’t produce a return.

Another layer: when does “meeting” become bad news? When the market is paying for acceleration rather than speed. That’s the one genuinely valuable sentence in the bear case — what matters is how much investment is increasing, not its level. A company decelerating from 100% growth to 60% is still extraordinary in absolute terms, but if the price embeds 100%, then 60% is a disappointment.

So the practical version: before a print, write down the expectation — consensus, or the company’s own guidance. After the print, compare to that first, and only then look at the absolute numbers. Most “irrational” price reactions stop being irrational at that step. As one host puts it: knowing the numbers and fetishizing the growth rates is interesting, but it guarantees you nothing, because the market is smart and it’s way ahead of this.

”Is this an AI bubble? Should I get out?”

Here’s the counterintuitive part: the disagreement between the two hosts isn’t bull versus bear. It’s a disagreement about definition — and nearly every bubble argument is stuck at exactly that point.

Three distinct illnesses get conflated:

A valuation bubble — price is too high relative to earnings. Easy to diagnose; you look at the multiple. By this standard the leaders don’t especially qualify.

An earnings bubble — the earnings themselves are fake or built on something unsustainable. Nobody laughs at this one, but the observation is fair: if it’s an earnings bubble, it’s been going on for years rather than appearing in a single quarter after a product launch. Bubbles aren’t usually that patient.

A pull-forward — the demand is real, the money is genuinely being spent, it’s just been dragged into the present. The consequence differs from the other two: it doesn’t zero the asset, it makes some future year look ugly because the orders were moved.

Your response differs completely by diagnosis. Valuation bubble: wait for price. Earnings bubble: leave. Pull-forward: adjust your forward growth curve and watch whether the replacement cycle holds. Collapsing all three into the word “bubble” sounds vigilant but tells you nothing to do.

Second layer: how do you judge whether a bear case deserves a change in your positioning? The standard set here is high — it has to tell you something you don’t know. CAPE is elevated, the Buffett indicator is elevated, growth will slow, supply will increase, rates pressure multiples. None of those is new, and five non-new things don’t add up to a new thing. That doesn’t make the conclusion wrong. It makes the conclusion already priced, which means it isn’t a reason to act.

I’d add one more test: a bear case that doesn’t state what would prove it wrong isn’t analysis, it’s a posture. Dating the burst looks brave, but if the end of 2027 passes quietly, the call gets rewritten as 2028 — and a forecast that can always slide is a forecast you can’t use.

”I was right about the company and still lost money”

The sharpest number in the episode is that starting-multiple study: buy above 50x, and even with genuine 15–20% growth for five years, your average annualized return is negative.

Your judgment can be entirely correct and you still lose. Cisco compounded 20%. Microsoft compounded double digits through the 2000s. Airbnb genuinely became a good business. Buyers at 40x sales still waited six years to get back to even. This isn’t a lesson about picking wrong companies. It’s a lesson about picking right ones not being enough.

Why is a multiple so hard to work off? Because it’s compounding subtraction. A 50x multiple means the market has prepaid several years of growth. The company then grows 20% while the multiple retraces from 50 to 25 — two forces in opposition, and the contraction usually moves faster. Every point of earnings growth is servicing an old optimism. The irony, also noted in the episode: these companies got 50x because the growth was explosive. A high multiple is the reward for excellence and an advance against your future return.

The fix isn’t avoiding great companies. It’s separating two evaluations: will this business win, and does this price let me share in the winning. The first is commercial judgment; the second is arithmetic. Most people do the first and blame luck for failing the second.

One closing thread worth connecting. The mystery chart: two retailers widely seen as head-to-head, one up 78% while the other fell 53% — then the roles fully reverse, the beaten one doubling while the leader flatlines. Walmart and Target. The admission that follows is honest: you cannot be unbiased about consumer-facing companies. McDonald’s, Coca-Cola, Disney, Netflix — your brain has already ruled. The companies whose products you never touch are the ones you can read numerically.

And the snap judgment isn’t a defect, it’s equipment. The explanation is good: you see a group that looks dangerous, you turn around, you live to pass on your genes. Humans have a hundred thousand years of evolution. Stocks have about four hundred. The instinct that keeps you alive runs backwards in markets.

Worth reading

• The Compound and Friends, August 25, 2026 — the source for these notes
• Bill Dudley’s Bloomberg column on why the bubble bursts by end-2027. Read the original rather than anyone’s summary, including this one
• Chip War, named in the episode, for the cyclical nature of semiconductors
• For the starting-multiple point, find any long-run scatter of entry P/E against subsequent returns — more persuasive than ten essays

The one thing to take away

Whenever you evaluate anything, what you’re actually doing is comparing it to a prior expectation — and you almost never write that expectation down, so you can’t tell whether you’re judging the thing or judging your own imagination.

That’s the entire reason a stock falls on great earnings. It’s also the entire reason a meal disappointed you, a person let you down, a project felt like it missed. The gap is measured against a line you never said out loud. And because it was never written down, it moves afterwards — when things go well you say “that’s what should have happened,” when things go badly you say “I knew something was off.” Neither version teaches you anything.

The exercise for today: pick one thing happening this week that you will definitely form an opinion about — a meeting, a kid’s test, a get-together you’ve arranged, a film you’re already excited for. Before it happens, write two lines on paper:

1. What I expect it to be like — specific enough to be judged right or wrong. Not “probably fine.”
2. What result would surprise me.

Then let it happen and go back to the page.

Three outcomes. It matched — in which case your satisfaction or disappointment says more about where you set the line than about the thing. It beat the line — record that, it’s evidence your judgment runs conservative. It missed — before blaming the other party or bad luck, look at what you drew the line from.

Do it three times and something surfaces: most of what moves you isn’t the event, it’s the line you won’t admit you drew. Write it down and it stops moving. This has nothing to do with markets, and it’s what markets do every single day.

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.