Nobody warns you before a bubble bursts
- Robin Powell
- Mar 18
- 7 min read
Before every major bubble in history, the experts were overwhelmingly bullish. New research from Harvard Business School reveals why — and why it means nobody's coming to protect your portfolio when the next bubble bursts.
Right now, 334 analysts cover the seven largest stocks in the world. Of those, 277 — 83% — hold buy ratings. Five of the seven have zero sell ratings. Not one analyst covering Microsoft, Nvidia, Alphabet, Amazon or Meta recommends selling.
That's not a divided room. It's a standing ovation.
If you find that unanimity reassuring, the historical record has uncomfortable news. This is what the run-up to a bubble looks like. Not panic. Not debate. Consensus. Near-total, confident consensus that the good times will continue.
The problem isn't that the experts are hiding something. They believe their own forecasts. The analysts issuing those buy ratings are genuinely convinced the stocks they cover will keep climbing. Research published this year explains how.
"The problem isn't that the experts are hiding something. They believe their own forecasts."
Think of it as a fire alarm wired backwards. Every sensor in the system — analyst forecasts, short-seller activity, media coverage — should flash red when danger approaches. But during the episodes that end in crashes, every sensor reads 'safe'. The more confidently they signal all-clear, the higher the probability that things are about to fall apart.
None of this means today's market is a bubble. But it does mean the absence of warnings tells you less than you think.
Inside a bubble, every expert signal points the wrong way
You'd expect the run-up to a crash to be noisy with dissent. Short sellers piling in. Analysts turning cautious. Journalists asking awkward questions. The research tells a different story.
Stolborg, C. & Greenwood, R.M. (2025), in a working paper from Harvard Business School and Copenhagen Business Schools, examined thousands of boom-bust episodes in individual US stock prices between 1980 and 2023. They tracked every available measure of expert sentiment to see what the professionals were doing as prices surged toward their peaks.
The most striking finding concerns short sellers — investors who borrow shares and sell them, betting the price will fall. During run-ups that end in crashes, short interest doesn't rise. It falls. At the peak, it averages 5% of shares outstanding — below the market average. The people who should be betting against overvalued stocks aren't there.
The other signals are no better. Analysts forecast long-term earnings growth of roughly 30% at boom-bust peaks — double the market average. Their expected one-year return for these stocks averages 52%, and only 7% carry negative return expectations at peak. Disagreement among analysts doesn't increase during the run-up; the entire distribution shifts upward. The variation reflects degrees of optimism, not the presence of genuine sceptics. Even media coverage offers no counterweight — fewer than 10% of articles mention overvaluation or use the word 'bubble', even after the crash has begun.
This research examines individual US stocks rather than whole-market bubbles, but the forces it identifies — extrapolation, conformity, career risk — operate at every scale.
"The louder the all-clear, the worse the fire."
Here's the punchline. When Stolborg and Greenwood expand their sample to include run-ups that don't end in crashes, the measures of optimism predict which ones collapse. Higher analyst growth forecasts and higher expected returns are associated with greater crash probability. Earlier work by Greenwood, R.M., Shleifer, A. & You, Y. (2019) established the framework: when an industry's net-of-market return exceeds 100% over two years, the probability of a 40% drawdown within two years rises from roughly 20% to 53%. At a 150% threshold, it reaches approximately 80%.
The louder the all-clear, the worse the fire.
Why the alarm never sounds
Three forces explain the silence — one cognitive, one institutional, one strategic.
The first is extrapolation. When a stock has doubled, analysts don't see a warning sign. They see confirmation the company is exceptional — and forecast more of the same. Research by Barberis, N., Greenwood, R., Jin, L. & Shleifer, A. (2018) models how this creates self-reinforcing bubbles: good news triggers buying, which drives prices higher, which breeds more optimism, which triggers more buying. Survey evidence from Greenwood, R. & Shleifer, A. (2014) confirms the pattern — people consistently expect high returns after recent high returns, the opposite of what rational models predict. Analysts who raise their forecasts during a run-up aren't lying. They're doing what human brains do: projecting the recent past forward and calling it analysis.
The second is career risk. As Shiller, R.J. (2002) documented, conformity pressures and reputational herding create a brutal asymmetry: being wrong and bullish costs far less than being wrong and bearish. A fund manager who rides a bubble up and crashes alongside everyone else faces a bad quarter. One who sits out the run-up faces redemptions, ridicule and the sack. Julian Robertson's Tiger Management learned this the hard way. Robertson called the late-1990s tech frenzy a 'Ponzi pyramid destined for collapse'. He was right. But his investors didn't wait. They withdrew billions, and Robertson was forced to close the fund in March 2000 — the same month the NASDAQ peaked. Being early and correct was indistinguishable from being wrong.
"Being early and correct was indistinguishable from being wrong."
The third force is the quietest. The few sophisticated investors who did recognise the dot-com bubble didn't sound the alarm. They profited from it. Brunnermeier, M.K. & Nagel, S. (2004) found that hedge funds were heavily overweight technology stocks throughout the bubble, their aggregate exposure peaking roughly six months before the crash. They rode the wave deliberately, timing exits at the individual stock level with considerable skill. The sceptics existed. But their scepticism never reached the public signals that ordinary investors rely on. They weren't warning anyone. They were making money.
"The sceptics existed. But their scepticism never reached the public signals that ordinary investors rely on."
Nobody is coming to save you
If the analysts are believers, the media is quiet, and the few who see through the optimism are too busy profiting to warn anyone — who's going to tell you when to get out?
Nobody.
And the discomfort doesn't end there. A globally diversified index fund — the standard evidence-based recommendation — still holds bubble stocks in proportion to their inflated market caps. If technology shares account for 30% of a global index because the sector has surged, 30% of your portfolio rides that surge. You can't own the market without owning its excesses. Diversification spreads your exposure across sectors and geographies, but it won't eliminate bubble risk.
So the protection has to come from the structure of your portfolio, not the opinions of its observers.
Broad diversification is the starting point. You're not all-in on a single sector, country, or theme. When a bubble bursts in one part of the market, the rest absorbs the blow. Not immunity — damage limitation. In a world where nobody can reliably call a bubble in advance, damage limitation is a serious form of defence.
Mechanical rebalancing may be the most important piece. A disciplined schedule forces you to trim winners and buy relative losers at set intervals — the opposite of what extrapolation bias tells you to do. It counteracts the cognitive pattern that fuels bubbles, and it works because it doesn't ask you to recognise what's happening. It just requires a calendar.
Cost discipline matters because you're not paying someone to fail at the same task. Active managers face the same biases, the same career pressures, the same information failures. Low costs mean those errors don't compound against your returns.
"The system works because it doesn't require you to be right. It requires you to be disciplined."
Staying invested through the crash prevents the behaviour gap — the well-documented cost of selling low and buying back high. Morningstar's annual Mind the Gap research finds that the average dollar invested in US funds earns roughly 1.2 percentage points per year less than those funds' reported returns, mostly because of poorly timed transactions. The worst damage in a bubble doesn't come from the crash. It comes from the panic sale that makes the loss permanent.
None of this makes you bubble-proof. But it removes the need for anyone — analyst, fund manager, journalist, or yourself — to correctly call the top. The system works because it doesn't require you to be right. It requires you to be disciplined.
The silence isn't safety
Hundreds of analysts. Near-unanimous optimism. Barely a sell rating in sight. Every alarm showing green.
That should trouble you — not because a crash is coming, but because history says this silence tells you nothing useful. It's not evidence of safety. It's what the system looks like when it's failing.
The evidence-based investor doesn't need to know when the fire starts. They need a portfolio built on the assumption that the alarm won't sound. Not certainty. Not bubble-proofing. Something more practical: a system designed to survive what nobody will see coming.
That's not pessimism. It's the most rational form of confidence there is.
Resources
Barberis, N., Greenwood, R., Jin, L. & Shleifer, A. (2018). Extrapolation and Bubbles. Journal of Financial Economics, 129(2), 203–227.
Brunnermeier, M.K. & Nagel, S. (2004). Hedge Funds and the Technology Bubble. Journal of Finance, 59(5), 2013–2047.
Greenwood, R. & Shleifer, A. (2014). Expectations of Returns and Expected Returns. Review of Financial Studies, 27(3), 714–746.
Greenwood, R.M., Shleifer, A. & You, Y. (2019). Bubbles for Fama. Journal of Financial Economics, 131(1), 20–43.
Shiller, R.J. (2002). Bubbles, Human Judgment, and Expert Opinion. Financial Analysts Journal, 58(3), 18–26.
Stolborg, C. & Greenwood, R.M. (2025). Optimism Everywhere: Beliefs during Stock Price Bubbles.
For financial advisers
The content problem nobody talks about
Clients expect regular, credible communication. Writing it at midnight isn't sustainable — and outsourcing it to someone who doesn't understand evidence-based investing is often worse. Regis Media produces content that actually reflects what you believe. Email Robin or message him on LinkedIn to find out how it works.
For investors: Find an evidence-based adviser near you
