Kahneman, prospect theory, and Munger explain how individuals misjudge, and Marks describes the sentiment that results. George Soros contributed something distinct and deeper. His idea is about how markets themselves behave when they're full of biased, thinking participants. It challenges the very foundation of the intrinsic-value approach you spent Phase 3 learning, and you'll need to hold the two in productive tension. Soros set it out in The Alchemy of Finance (1987), and he calls it reflexivity.
How does reflexivity work?
The core idea. Classical finance assumes a one-way street. Fundamentals determine value, and prices passively reflect (or gravitate toward) that value. Prices are a thermometer, reading a temperature they don't affect. Soros says that's false whenever the participants are thinking humans whose decisions affect the thing they're trying to predict. In such systems two functions run at once. There's a cognitive function, where participants' perceptions are shaped by the situation. And there's a manipulative or participating function, where participants' perceptions, acting through their decisions, shape the situation. When both operate, you get a two-way feedback loop. Perceptions influence fundamentals, and the changed fundamentals influence perceptions, around and around. Price isn't a thermometer at all. It's a thermostat that changes the temperature it reads. Under reflexivity, markets needn't tend toward equilibrium. They can move away from it in self-reinforcing trends, because the trend changes the fundamentals in a way that justifies the trend, for a while.
Advertisement
Where does reflexivity matter most in investing?
Why this matters mechanically. The cases where price is a fundamental. Reflexivity is strongest wherever a company's valuation feeds back into its actual business results:
- Companies that use their own stock as currency. A serial acquirer whose shares trade at a high multiple can issue those richly valued shares to buy other companies cheaply. That boosts reported earnings per share, which appears to validate the high multiple, which lets it issue more shares to buy more companies. It's a self-reinforcing boom, as in the 1960s US conglomerate wave and the roll-up frauds of later decades. The high price is reflecting value, and it's also manufacturing the earnings growth the market cites to justify the high price. When the multiple stops rising, the machine reverses. The currency loses value, acquisitions stop being accretive, growth collapses, the multiple falls further. The bust is faster than the boom.
- Financials, where confidence is the fundamental. A bank or NBFC funds long-term loans with short-term borrowing, and its access to that borrowing depends entirely on the market's confidence in its solvency. A high credit rating and a rising share price lower its cost of funds. That raises its profits, which appears to confirm its strength, which supports the rating. Then something cracks confidence. Funding evaporates, forcing asset sales that realize losses that further destroy confidence, and you have a run. That's the IL&FS and DHFL failure mode: a reflexive collapse in which the perception of insolvency caused the insolvency by cutting off the funding the model required. For a bank, "confidence" isn't sentiment sitting on top of the fundamentals. It is a fundamental, and Soros's framework is why.
- Currencies and commodities, where a trend (a strengthening currency, a rising oil price) changes the economic behavior (capital flows, investment, hoarding) that sustains the trend.
What is the boom-bust sequence?
The boom-bust sequence. Soros's model of a full reflexive process has a characteristic shape you can learn to recognize. It starts with an initial trend that nobody has recognized yet. Then comes a period of acceleration, as the trend and the prevailing bias reinforce each other. A testing period of setbacks follows, and if the trend survives it, the bias gets reinforced further ("it dipped and recovered, so it's unstoppable"). The gap between expectations and reality then becomes unsustainable (the "moment of truth"). There's a twilight or plateau where people keep playing a game they no longer fully believe, then the crossover where the trend reverses. Finally comes the crash, a self-reinforcing collapse far more violent and rapid than the boom, because the same feedback now runs in reverse. Booms that build slowly and busts that happen fast are a direct prediction of the model and a recurring feature of real markets.
How does reflexivity fit with intrinsic value?
The synthesis with intrinsic value, and the tension you must hold. Reflexivity seems to contradict everything in Phase 3. If price changes fundamentals, how can there be a price-independent "intrinsic value" to anchor to? The mature resolution, the Soros-versus-Buffett/Munger synthesis, is that both are true in their domains, and you need both. For most stable, cash-generative businesses most of the time, the intrinsic-value approach is right and reflexivity is a second-order ripple. Buffett and Munger largely ignore macro reflexivity. They buy businesses sturdy enough that the feedback loops don't threaten them, and that's a valid, lower-stress way to win.
The specific situations above are different. Think of leveraged financials, serial acquirers, anything where confidence or the stock price feeds the fundamentals, and whole markets at manic or panicked extremes. There reflexivity dominates. "Intrinsic value" is unstable because it depends on the price, and ignoring the feedback loop will get you killed. A "cheap" NBFC isn't cheap if a confidence crack will destroy the book. Soros makes his money by identifying and riding these reflexive processes and by recognizing the moment they turn. Buffett makes his by avoiding businesses fragile to them. Your job is to know which regime a given situation is in. The tell is one question about any company: does its share price or the market's confidence feed back into its actual results? If it does, you're in Soros's world. Analyze the loop as well as the snapshot. Worked Example 3 quantifies one such loop.