A bubble is not a bubble unless it bursts

Published in The Sunday Times on 25th October 2026.

For a long time, I have pushed back at the idea that bubbles are only identifiable in hindsight, as many academics would argue. It seems intuitive to me, that if prices rise dramatically, the more inevitable the eventual reversal.  

And I’ve witnessed a few. In my investing career, I have worked through two extreme episodes – the boom and subsequent crash associated with the dotcom bubble and the somewhat less dramatic boom that preceded the arguably more dramatic crash associated with the great financial crisis.

A strong run in prices does not portend a bust

It turns out my intuition is letting me down.

Historically, it has been a mistake to assume that because something has risen dramatically, mean reversion must come next. Enormous price rises don't reliably predict subsequent falls.

If prices do subsequently collapse, we can confirm in hindsight that it was a bubble, but that’s not a testable definition of one. There would need to be a systematic way of identifying the bubble in real time and predict that there is some end to it. That’s a high bar.

AI Bubble 

This has never been more topical. References to an ‘AI bubble’ are pervasive. In fact I’ve written about it myself in this paper no less than three times in the past twelve months.   

Part of the problem is definition. Calling something a bubble usually implies that its price has become detached from fundamental value. But that requires us to agree on a valuation model. That’s controversial. 

Price-based definitions are less contentious: agree on what constitutes a price boom and what qualifies as a crash. Still subjective, but easier to get a consensus on.

The practical issue for investors then, is whether a rapid increase in prices predicts a rapid future price drop. The historical rarity of bubbles, makes this question difficult to answer. But a recent CFA Institute paper by Roger Ibbotson sheds some interesting light on this. The paper goes back as far as 1792 and looks at US stock market for evidence.

A boom is more likely to be followed by further price rises 

And here’s the punchline - stock market booms are more likely to be followed by a further boom than by a crash.

Of the episodes in which the market rose by at least 50% over three years, on 37% of occasions the market gained at least another 50% within five years. A full reversal only occurred on 13% of occasions. Even very large multiyear or one-year runups in the US equity market have historically been more likely to be followed by further gains than by full reversals.

Several aspects of this analysis warrant attention.

The popular intuition that a significant runup in stock prices is a reliable signal of an impending crash is not supported by 232 years of US data.

What a boom tells you more reliably is not the direction of the next move, but that the range of possible outcomes has widened. So a boom is a reliable predictor of future volatility, but not of market direction. 

Booms are common. Booms that fully unwind are not. A 50% surge might look bubbly, but historically fewer than 2% have subsequently given all those gains back within three years. Such a small sample renders any conclusion to the territory of anecdote, not evidence.

It is an understatement to say that my working life experience is very unusual

I have worked through two of the most memorable boom-and-bust episodes in modern markets - they loom large precisely because they were exceptional, not because they were normal. It may also explain my somewhat unhelpful obsession with trying to understand the nature of bubbles and an overly cautious approach to investing.  

True bubbles are rare. This isn't to say bubbles don't exist. Nor is it to say valuation doesn't matter. It simply says that identifying a spectacular rise in prices is not enough to tell us, in real time, that we're looking at one. 

The author does acknowledge the survivor bias in using US stock market data. The US market is exceptional in that it has survived wars, crises, and regime changes. It is dangerous to infer too much from only one run of history in the most successful market in the world.

All booms are different

All stock market hype phases are different. The current hype around AI is evident everywhere. The bear case is easily constructed and compellingly told. But the most intellectually honest conclusion as to whether this is a bubble or just a boom is - we don’t know. Those last three words are used far too sparingly in finance, a profession with an unfortunate preference for certainty. 

With global equities up c. 17% year-to-date in euros you are being paid to bear this ambiguity. If it’s certainty you are after, I’ve plenty of bonds or money market funds I can offer you.

Source: Data is sourced from Bloomberg as at market close 31st December, returns are based on total indices in local currency terms, unless otherwise stated.

Gary Connolly is Investment Director at Davy. He can be contacted at gary.connolly@davy.ie or on X at @gconno1.