Reflexivity

If the market is efficient — if prices already reflect every known fact — how did the same market trade Pets.com at a $300 million valuation in 2000 and at zero eighteen months later, with no new fundamental information that justified either price? The dominant theory of how markets work — the Efficient Market Hypothesis — says that prices reflect all available information. Buy a stock and you are paying a price that already incorporates everything anyone could reasonably know about the company. By construction, the average investor cannot beat the market because the market is the aggregation of everyone's information. This theory has won Nobel prizes, populated business school textbooks, and built the index-fund industry. It is also wrong, in a specific and important way, about the part of market behavior that matters most: bubbles and crashes.

George Soros, who spent forty years compounding capital at roughly 20% annually, named the framework that actually explains those episodes: reflexivity. The core idea is simple. Markets are not just measurement devices that passively reflect underlying reality. Prices change the reality they are supposed to be measuring. A rising stock price lets a company issue new shares cheaply, raise capital, acquire competitors, hire engineers, and grow. The improved fundamentals then "justify" the higher price. Skeptics get squeezed out, more capital flows in, and the cycle reinforces itself. On the way down, the same loop runs in reverse: falling prices tighten credit, force layoffs, suppress earnings, and confirm the lower prices.

The two functions that produce the loop

Soros distinguished two ways participants relate to markets. The cognitive function is the attempt to understand the world — to figure out what assets are worth. The manipulative function is the way participants' beliefs change the world — through buying, selling, lending, hiring, investing. Standard economic theory assumes these are separate: there is the world, and there is our understanding of the world, and the second describes the first. Reflexivity says the two functions feed each other. Our understanding of the world changes the world; the changed world then alters our understanding.

When this feedback runs strongly in one direction, markets do not move toward equilibrium. They move away from it. Prices and fundamentals drift further from any sustainable relationship until something breaks — a regulatory change, a confidence crack, an interest-rate hike, a leverage limit. The break triggers the reverse loop. In a single episode, perception can pull fundamentals up by 50% and then push them down by 70%, all without any change in the underlying productive capacity of the assets involved.

Six classic reflexive episodes

The 1980s conglomerate boom worked exactly this way. Companies that traded at high P/E ratios issued stock to buy companies trading at lower P/E ratios. Earnings per share rose mechanically (the acquired earnings were now divided by fewer shares relative to the higher market value), confirming the high multiple, which then enabled more acquisitions. The loop ran until accounting changes and market saturation broke it, at which point the same conglomerates traded at deep discounts to the sum of their parts.

The dot-com episode of 1995–2000 is the canonical case. Rising internet stocks made venture capital cheap; cheap capital funded thousands of internet startups; the proliferation of internet activity seemed to "validate" the sector's prospects; valuations climbed further. By March 2000 the NASDAQ traded at over 200 times earnings of the companies that had any. The break came from a combination of cash burn rates, Fed tightening, and one quarter of disappointing announcements. The unwind erased 78% of NASDAQ market capitalization in two years.

The 2003–2007 housing bubble followed the same shape. Rising home prices made loans look safer (the collateral was always worth more next year), which loosened underwriting, which brought more buyers, which raised prices. The break came when subprime defaults started rising — even modestly — and forced the unwind of the loop. Tesla's stock from 2019–2021 worked similarly: high stock price let Tesla raise $12B in 2020 alone, the capital let Tesla expand production and pay engineers, the expansion seemed to validate the valuation. Bitcoin in 2017 and again in 2021. Meme stocks in 2021. AI-themed tech stocks in 2023–24. The shape is the same every time.

Recognizing reflexive markets in real time

The diagnostic checklist that distinguishes a normal market from a reflexive one is short and worth memorizing. Are valuations high relative to historical norms? Is media attention and retail participation rising sharply? Is the trend self-reinforcing — more buying produces higher prices which produces more buying? Are fundamentals "improving" specifically because of high prices (companies issuing stock, raising capital, hiring, paying bonuses in stock)? Are skeptics being mocked or losing money for being right too early? Is leverage increasing in the underlying market? Five or more "yes" answers means a reflexive bubble is in late stages. Three or four means a setup is forming.

The practical implication for an investor: in reflexive markets, traditional valuation is a moving target. The right question is not "what is this asset worth?" but "where in the reflexive cycle is it?" Early-stage reflexive bubbles (yes on two or three checklist items, valuations within a standard deviation of historical norms) can offer enormous returns. Late-stage reflexive bubbles (yes on all six items, valuations multiple standard deviations from norm) are the most dangerous markets in the world, because they continue past every rational sell signal until the break comes from outside the system.