George Soros' Theory of Reflexivity

Do stock price changes cause fundamental changes rather than the other way around?

Note: Capital Gains now has a referral program! Rewards include (but aren't limited to) access to my database of book reviews (1), an invite to the new Capital Gains Discord (2), stickers (10), and a mug (25). Details at the bottom for existing readers, or subscribe here to get your unique link!

The standard story for stock prices, both among theorists and practitioners, is that they reflect the fundamentals of the assets in question. For example, if a company earns $1 per share and the stock trades at $15, then for the stock to get to $20 you’ll need some combination of:

  • A one-time reset in earnings, such that it's making $1.33 but with the same expected growth rate as before,

  • A change in future growth expectations, such that $1 of earnings today implies a higher number in the future than it used to, or

  • Some decrease in the perceived riskiness of the earnings stream (an example of this might be a company that has one direct competitor with frequent pricing wars, where that other company shuts down or otherwise indicates it's not going to be such a fierce competitor any more).

This theory, that fundamentals are what inform stock prices, makes enough sense that it’s hard to imagine any alternative—but that’s exactly what George Soros did, and it’s how he made his very large pile of money.

Soros’ reflexivity theory flips the argument on its head, and states that instead of the fundamentals being what move stock prices, it’s the stock prices that move fundamentals. Over the years, he’s explained the idea in various books and op-eds, but originally it was revolutionary, and detailed for the first time in a 1970s memo that explained how his firm was thinking about real estate investment trusts (the memo was so popular that the firm would sometimes get requests for a fresh copy, because it had been forwarded by fax so many times that it became unreadable).

The Soros argument works best for companies that use a little leverage, and it goes like this:

  • Suppose there's some sector where companies are chugging along, investing, growing, and borrowing some money to do it.

  • Suddenly, investors decide that the sector is an exciting growth area. So they start buying more shares, driving stock prices up.

  • With all of that investor enthusiasm, companies in the sector are now able to borrow more cheaply. And that's what they do—as long as borrowing gets cheaper faster than returns on investment go down, everyone is making more money.

  • So the investor’s thesis is validated. The companies are growing! What’s more, now some of them have started acquiring each other just to grow faster—which gives investors a sense of what a “sophisticated buyer” would pay for that sort of asset, further increasing investor confidence that the assets in question are worth owning.

  • All is not well, though: the sector eventually saturates its market. It's either overpaying for existing assets or overbuilding new ones, and either way, returns start to go down.

  • Suddenly, all those healthy dynamics reverse; lenders don't want to pour money into a shrinking industry, so the cost of borrowed capital goes up. Meanwhile, some of the most over-levered companies collapse entirely. Losses abound.

  • Eventually things stabilize at a more rational level, following fundamental theory. And sometimes, the reflexivity cycle begins anew.

This idea is quite clever, and is somewhat unique in that it's both a description of a real investment phenomenon and a framework for making trades: the basic strategy is to buy when some category of stocks is going up for no reason, and to sell when they're going up for a very good reason. It's also a paradoxical idea, a bit like someone being famous-for-being-famous. It imagines a sort of Kardashian economy where industries are profitable because they're seen as profitable, not because of fundamental characteristics.

That said, it’s not "contrarian" in the most literal sense of buying low and selling high, since it proposes that you buy high and sell higher. But it is contrarian in the broader sense of finding a very clever reason to do something that a reasonably smart person would assume is the exact opposite of what makes sense.

You'll notice in the example above that reflexivity works for companies that borrow. But it also works for countries—Soros made a lot of money chasing momentum in currencies over his career. In fact, one paper from quant firm AQR even argues that, if you strip away quantitative factors like momentum and trend-following, Soros actually added negative value as an investor.¹

But there are other cases where it works. Tech in the last decade has found itself in a reflexive situation, where rising stock prices make equity compensation appealing, which allows tech companies to outbid other industries for talent. And since tech companies sell some of their products to other tech companies, these capital inflows actually create better fundamentals for the overall industry. As long as there's more software worth writing, more chips worth designing, more gadgets worth inventing, etc., the investment case for the industry can get better as more money flows in.

Whether or not this is sustainable actually ties in with some deep questions about the economy: if costs go down at scale, then reflexivity should apply in more places. If costs go up, it doesn't. You can make an argument for either, depending on scale and timelines.

At the level of a single factory, if the factory goes from running one shift a day to two, it's going to be paying more overtime—the incremental hours are more expensive. If demand goes up even more, it has to buy additional equipment, and maybe extra storage space. There's a range of outcomes where marginal costs decline (e.g. paying for full shifts and going from six hours a day of work to eight) and a range where they rise.

Some costs go way down as we scale, most famously with computers. But there are still physical constraints; we've already dug up most of the easily-accessible metals in the world, so now it’s more expensive to dig for additional iron and copper. But we've also gotten better at it over time. (The question of how to think about costs during times of growth is a topic for another episode of Capital Gains.)

Similarly, software costs tend to go down at scale, since bits get cheaper to deliver over time,but the cost of distributing that software tends to rise; you'll have a very low customer acquisition cost indeed if your marketing plan is to text all of your friends and tell them to check out the thing you just built, but unless you happen to have eight billion friends, you'll eventually have to spend money to grow. And you'll often find that the cheapest marketing channels are also the ones that get saturated first—this is one reason that companies whose original marketing plan is some kind of scrappy, low-cost guerilla marketing approach eventually find themselves doing TV ad buys and sponsoring stadiums.

Reflexivity is worth understanding for two reasons.

First, it's a useful mental exercise for avoiding our tendency to instantly dismiss overhyped industries. Yes, AI startups are raising a ton of money, and no, most of them aren't generating much revenue, and very few are showing good margins right now. But that money can kickstart a feedback loop that makes them more profitable in the future.

And second, there are some phenomena that only make sense as reflexive ones: financial craziness can provide the activation energy necessary to shift some industry from hypothetically valuable but unproven into viable and growing. And that's something you don't want to miss.

Interested in going deeper? The Diff has applied the idea of reflexivity in a few posts over the years:

And tomorrow, paying subscribers will get a piece on how current investment and product trends in AI are an example of a reflexive phenomenon. You can sign up here.

1. These kinds of conclusions are often deflating, but they don't imply that an investor is unskilled. What we can say instead is that Soros managed to identify through experience what later researchers would find by running regressions on voluminous historical data, and that even imperfect implementation of these discoveries produced a strong track record.

Share Capital Gains

Subscribed readers can participate in our referral program! If you're not already subscribed, click that Get Started button below and we'll email you your link; if you are subscribed, be sure to check out the email version of this edition.

Help us get to know you

We'd like your feedback so we know what to cover, so if you haven't already, please fill out this one-minute form.

Join the conversation

or to participate.