Transcription
For the second year in a row, the best performing stock market in the world is South Korea. Last year, the Cosby rose 76%, its best year in over four decades, and this year at its June peak, it was up a further 112%. For comparison, the S&P 500 is up less than 10% year to date, and the MCI world is up around 20%. So, Korea is beating the rest of the world two years running by a margin normally reserved for accounting fraud.
If you'd put your money into Korean stocks 18 months ago, you would by any reasonable measure be delighted, unless you've been watching the news over the last month, in which case you'd assume that you were ruined. Because the best performing stock market in the world is also right now home to one of the most punishing bare markets in the world. Since peaking in June, the Cosby has fallen about 27%. It is somehow both of these things at once, the best and the worst place on earth to own stocks, which shouldn't really be possible. And yet here we are.
To understand how a market can be simultaneously the greatest bull market in the world and a full-blown crash, we need to go back a little. When South Korea's president Lee Jung ran for office, his big headline campaign promise was Cosby 5000. His central platform was quite literally a target level for the national stock index. He's a former trader, which possibly explains this. At the time, the index was trading around half that level, and the pledge was dismissed as cheap political theater, the equivalent of a candidate promising that under his leadership, the weather would improve.
Then on the 19th of June 2026, the Cosby didn't just hit the president's target, it blasted straight through it and peaked at over 9,000 points, nearly double what he promised. So he overd delivered, which is not a word one gets to use about elected officials very often, and then it all fell apart. Since that peak the market has lost a quarter of its value and entered a bare market, which brings us back to where we started, down 25% from the peak and still up almost 60% on the year.
The Cosby is currently more volatile than it was in the 1997 Asian financial crisis and the global financial crisis in 2008. According to the FT, Korean market volatility has been so extreme that the stock exchange paused trading 37 times this year compared with three times in all of last year.
If you listen to the news, you'd believe that the Korean market has become a speculative casino run by irrational overleveraged day traders pumping up local theme stocks on internet hype. This explanation is wrong, or at least it gets the interesting part completely backwards. Because unlike the Western memetock episode of 2021 built on a bankrupt cinema chain and a dying video game shop, the companies at the center of the Korean surge are highly profitable worldclass industrial giants. Samsung Electronics and SKH Highix, two firms that between them control much of the global supply of high bandwidth memory chips, the physical hardware that the entire AI boom runs on. And their earnings are real. In the first quarter of 2026, Samsung's operating profit rose 756% yearonear to 57.2 trillion one. SK Highix saw revenue rise 198% and operating profit rise 405%. These are not companies with a photo of a gold mine that they found on Pinterest. They make the things that everybody wants to buy right now.
The extreme price action here is not the usual story of worthless companies being bit up by fools. It's something more unstable and more interesting. A combination of extreme index concentration, some very aggressively leveraged products, and a retail trading culture with a risk appetite that makes Wall Street bets look like a cautious public pension board.
Before we dig into the Korean retail trading culture, let me tell you about this week's video sponsor, SY. Anyone who travels knows the feeling of landing in a new country, turning on their phone, and worrying about the roaming charges that you could be hit with while you're still walking to baggage claim. SLY solves that problem. SY is an eim app that gives you affordable data in more than 200 destinations around the world. And the best part is that you only have to install the app once. After that, it's easy. Every time you go somewhere new, you open SY and pick a plan. There's no swapping of physical SIM cards, no queuing at airport kiosks, no hunting down a local provider in a language that you don't speak. You sort it all out before you leave home, choosing a plan that fits how much data you actually need. When you land, your phone connects automatically to a local network. SY offers also global and regional plans. For example, Asian or European plans. Because you're on a local data plan rather than roaming, there are no nasty surprises waiting on your bill when you get back home. You can order a car, use your GPS, or call ahead to your hotel the second you step off the plane. Saline now offers instant app-based US phone numbers with one tap activation. This gives you a number that works everywhere you go with no contract, no physical SIM, and no single location lockin, including voice calls and texts. Get an exclusive 15% discount on sale data plans by downloading the SY app using the link in the description or using this QR code and put the discount code boil in at checkout.
So the odds, if you look at who's actually trading in Seoul, the dominant force isn't foreign institutions or domestic pension funds. It's an enormous hyperactive retail army known locally as the ants. It's roughly the Korean equivalent of the Wall Street Bets crowd calling themselves apes. Except there are great many more ants and for a long time they were winning. Out of a population of around 51 million, South Korea has about 14 million retail investors, nearly 30% of the population. And they account for roughly half of all trading volume in Korea. This compares to around 20% of trading volume coming from retail investors in the United States. Foreign investors make up around 31% of Korean volume, and domestic institutions, the pension funds, the asset managers, the professionals, a mere 18%. So the smallest player in the Korean market by some distance is the group that invests for a living. In the US, institutional investors make up 80% of equity trading volume.
Over the last 6 months, the professionals have been leaving. Foreign capital has pulled about $95 billion out, but the ants have stepped in and committed an additional $80 billion. So, as the foreign money heads for the exit, the retail army has been standing underneath, arms outstretched, promising to catch everything on the way down. And for most of last year, they did. We'll come back to how that works out when the thing falling gets heavier.
Now, it's easy for Western commentators to write these traders off as reckless gamblers, but if you look at what young Koreans are actually facing, their behavior is at least internally consistent. South Korea has famously rigid expectations about the order you're meant to do things in. The prestigious job, the apartment, the marriage, then the children. The trouble is that the average Korean man now marries at around 34 years old. And in Seoul, only about a quarter of people in their 30s own their own home. So the apartment, which is supposed to come before the wedding, increasingly arrives after it, if at all. When you've been priced out of the housing market entirely, putting 10% of your salary into a savings account doesn't feel like a slow path to a home. It feels more like trying to inflate a hot air balloon with a bicycle pump.
My friend Dimmitri Cafenus has a name for the mindset this produces. Financial nihilism. Once the ordinary ladder to the middle class has been pulled up out of reach, gambling on markets stops feeling like the reckless option and starts feeling like the only option. If the sensible patient strategy has a payoff of approximately zero, the reckless one doesn't have to look good. It just has to look better than zero. And a lot of things clear that bar.
The academic Owen Lamont, a portfolio manager who's thought at Harvard, Princeton, Chicago, and Yale, has been watching this from the other direction. A while back, he wrote that the US market was Koreafying, slowly turning into the retail driven theme stock chasing market Korea has had for years. More recently, he upgraded the diagnosis and started calling it the Squid Game market, where ordinary people take enormous risks, hoping to become overnight millionaires with broadly the survival odds that the title implies.
And because the ants hold this much of the market, they've developed their own subculture built around what they call theme stocks. A much older concept than meme stocks. And for the ants, the theme rather than the actual business is the entire investment case. The most famous example came in 2012 and involved a company called DIC Corp. Di Corp. makes semiconductor testing equipment. It had released no news of any kind and nothing about its business had changed. But the firm's co-CEO was a man named Juan Ho Park, whose son is the South Korean rapper Sai. That July, Sai released Gangnum Style and for a while it became the biggest thing on the planet. Korean retail investors poured into his father's firm. The stock rose about 800% in 3 months, eventually trading at 60 times earnings, while its direct competitors traded at nine. A semiconductor testing company went from a market value of $38 million to $334 million because the chief executive son was doing a horse dance on YouTube. I like to imagine Mr. Park watching that video for the first time with a certain amount of embarrassment, his grown son doing a horse dance for the internet, and then watching his share price go vertical and slowly making his peace with it. Perhaps his son hadn't brought shame on the family after all. Perhaps his son was a genius who'd understood before the rest of us that this is now how the world works.
What makes today different and potentially more dangerous is that the ants are no longer pumping up obscure testing firms. They've pointed the exact same leveraged hyperaggressive playbook at two of the largest and most systemically important technology companies on Earth. And to understand how a retail army can destabilize companies of this size, you have to start with the concentration built into the index itself.
The Cosby wasn't always like this. For most of the last decade, it drifted sideways, made up of fairly unglamorous companies, car makers, ship builders, arms manufacturers, makers of televisions and fridges. Many were being undercut by Chinese rivals and many were run by the sprawling opaque familycontrolled conglomerates Korea is known for. It was not a thrilling index. It was an index you owned because you felt you ought to. Americans have spent the last couple of years worrying about concentration in the S&P 500. At its peak, Nvidia made up around 7% of the index, and this was discussed as a genuine national vulnerability. In Korea, at the June high, the two stocks Samsung and SKH Highix reached a nearly 60% waiting in the Cosby, up from around 40% 18 months earlier. The Cosby isn't really a diversified basket of a nation's industry. It's a leverage bet on two chip stocks wearing the country as a costume.
The concentration got so extreme it started to trip over other people's rule books. Many American funds are legally required to stay diversified. Rules that cap how much of a fund can sit in a handful of dominant names so that a single stock going wrong can't hurt investors too much. Goldman Sachs worked out that if Samsung and SKH Highix's combined weight climbed just one more percentage point, foreign funds bound by those rules would be forced to sell around $2 billion of Korean shares. Not because they changed their minds, but because the index had become too topheavy for them to legally hold. The market had, in a sense, become too successful to own.
And it was into this arrangement, an entire national index balanced on two names, that Korea's financial industry decided to introduce leverage. In late May 2026, brokerages launched 16 single stock leveraged ETFs tied to Samsung and SKH Highix, each promising retail investors twice the daily return of the underlying shares. These joined an even larger offshore product, a Hong Kong listed leveraged SKH Highix fund that in 9 months had swollen to around $13 billion, according to Bloomberg, the biggest single stock leveraged ETF anywhere in the world. Bear in mind, these are shares that already routinely move 8 or 10% in a day. The ants looked at that volatility and decided it wasn't quite enough. They needed a version with more juice.
Now, to see why regulators started to sweat, you need to understand how a leveraged ETF actually works. I promise it's worth it. The whole story hinges on this one mechanism. If you own an ordinary share and it falls 10%, you lose money, but you don't have to do anything. The position just sits there, worth a bit less, but judging you. A leveraged ETF is less relaxing because it promises to deliver exactly twice the stock's return every single day. It has to reset its exposure at the end of each session to keep that two times ratio ready for tomorrow. And the direction of that reset is the whole problem. When the stock rises, the fund has to add more exposure at the close. It buys more stock. When the stock falls, the fund has to shed exposure. It sells, pushing shares into a market that's already dropping right at the closing bell. So, it buys strength and sells weakness mechanically every day with no opinion about whether that's a good idea or not. On a derivatives desk, this is a short gamma dynamic, which sounds like a medical diagnosis. The planer way to picture it is as audio feedback. Bring a microphone too close to a speaker it's plugged into, and the mic picks up its own output, sends it back through the amp, the speaker throws it out louder, the mic catches that, and within a second or two, the whole room is one rising shriek. A leveraged ETF is a microphone held up to the stock it tracks. Every move it's supposed to follow, it instead amplifies, and then it's forced to amplify the amplification.
According to Bloomberg, on turbulent days, the rebalancing from these products has swelled to as much as 2/3 of all trading in SKH Highix. A remarkable share of the flow for a company worth well over a trillion dollars. So, the tail is not wagging the dog. The tail has hoisted the dog off the ground and is swinging it in a circle to see what happens. You could measure the force of it during this month's crash. Goldman Sachs estimated that SKH Highix's doubledigit drop on Monday obliged the leveraged funds to sell roughly $5 billion of the stock at the close just to rebalance. A single automated adjustment that came to about 18% of all trading in SKH High shares and futures that day. Nobody chose to sell that $5 billion of stock. The clock simply reached the hour at which the machine sells.
That predictability has rewired how professionals behave. In Hong Kong, market makers now spend the early afternoon working out exactly how much the funds will be forced to buy or sell at the close. They get in front of it, ride the wave, and step out before the closing bell. One trader described it to Bloomberg as easy money, a polite way of saying the retail investors force trades have become a daily scheduled contribution to whoever bothered to do the arithmetic.
Underneath all of this runs a layer of plumbing most people never see. The firms issuing these ETFs don't actually hold the extra shares. They rent their leverage from investment banks through total return swaps. The bank agrees to pay the fund twice the stock's daily return and hedges its own side by trading the real shares, which is a large part of what drives those end of day flows. But the banks carry a nightmare of their own called gap risk. The danger is that the stock falls so far, so fast overnight that the fund is wiped out and can't settle, leaving the bank holding the loss. To ensure against this, the banks buy protection from hedge funds using exotic derivatives called cle. The swap is the leverage. The cle is the fire insurance the people selling the leverage take out in case the building they've just filled with fireworks needs to be evacuated. And you can read the anxiety directly in the price of that insurance. As the market has lurched around, the annual cost for banks to buy clees has climbed from roughly 3% in March to more than 10% by mid year. The banks have started rationing how much of this exposure they'll carry and warning the fund issuers that they may soon stop minting new shares altogether, the institutional way of announcing that the grown-ups would now like to leave.
By now, the distortions have gotten extreme enough to push the regulators into public contrition. The head of the Financial Supervisory Service, the body that had waved these products through in the first place, offered a line worth pausing on. Looking back, I regret not doing everything I could to stop it. It's not every day a regulator reviews his own recent work and files a formal complaint about it. The government has now frozen all new listings of single stock leveraged ETFs and announced that from the 5th of August, any retail investor who wants to trade them will first have to sit through mandatory riskmanagement training and put down at least $30 million one, a bit over $20,000 to open an account. Which is to say, having built the casino, installed the tables, and let the crowd in, the authorities have decided that what the situation really needs is a short educational film and a cover charge.
When the machine behind the leveraged ETFs starts selling automatically, it doesn't stop to ask who's on the other side of the trade. And this month, as the Cosby fell nearly 9% in a single session, the people on the other side were almost entirely the retail ants. The scale of what the rebalancing left behind is hard to take in. According to a widely cited estimate from Goldman Sachs trading desk, more than 1.2 million retail leveraged accounts were hit with margin calls during the crash. That's roughly one in every 30 working age adults in the entire country. Margin called in a matter of days. Of those, somewhere between 360,000 and 360,000 people were wiped out completely, liquidated automatically by their broker systems, most of it at the close. And because some of the drops were so sudden, a number of those positions gapped down through their maintenance levels before anything could be sold. So that when the shares were finally dumped, the proceeds didn't even cover the borrowed money. Those investors didn't just lose their savings, they woke up owing their broker money. Over the same stretch, retail brokerage deposits across Korea fell by almost 30 trillion Juan, back to their lowest level since February.
Now, it's important to note that these traders weren't fools who had been sold an obvious fraud, and they hadn't misread the big picture. They were on the right side of one of the great bull markets in modern history, holding shares in ferociously profitable companies at the center of the entire AI buildout. They had the trend exactly right. They just got the bet size wrong.
My friend Victor Hagani ran a famous experiment that perfectly explains how ruinous that one mistake can be, even for people who really ought to know better. He recruited 60 win subjects, gave them $25 each, and sat them in front of a simple coin flipping game. They could bet as much or as little as they wanted, as many times as they wanted for 30 minutes. And they were told in writing, in bold, that the coin was rigged. It would come up heads 60% of the time. Now, a coin that you know lands heads 60% of the time that you're allowed to bet on repeatedly is not a gamble. It's a gift. Bet sensibly and it's pretty much guaranteed that you'll walk away richer. And these people weren't people off the street either. They were economic students and young finance professionals. The exact people trained to see this. This should have been the easiest money any of them ever made. It turned out not to be. Even with the odds tilted firmly in their favor, 28% of them went completely bankrupt inside half an hour. Only about one in five reached the maximum payout. The average player walked away with $91, which sounds fine until you realize that the game was practically handing them $250. They lost on a game they could not lose for one main reason, and it's something Wall Street has lived off of for a hundred years. Even when handed an edge, people don't bet calmly. They bet too big. And then a perfectly normal run of bad luck, the kind of probability guarantees will turn up, weighted coin or not, arrives and clears out their whole stake before the edge ever gets a chance to work. That's the entire lesson of Korea in one coin game.
Leverage doesn't improve the thing you own. It doesn't make Samsung a better company or the AI boom more real. All it does is shorten the amount of time you're allowed to be wrong. Apply two times daily leverage to a wildly cyclical chip stock and you're not doubling your returns. You're setting yourself up so that an ordinary routine dip empties your account long before the long-term story you were right about ever arrives. And it landed hardest on exactly the people we started with. Of the accounts fully wiped out, something like 62% belong to investors in their 20s and 30s. The same young Koreans who'd been priced out of a house decided that the patient road led nowhere and reached for leverage as the fast way in. They were right about where the market was going. They just didn't survive the trip.
If you step back from the retail wipeout and look at the wider Korean economy, you run into a macroeconomic puzzle that's been tormenting another group of investors, foreign exchange traders, all year long. Normally, a country running an enormous trade surplus has a strong currency. Money floods in faster than it flows out, and the currency rises. And Korea's surplus right now is enormous. It's selling the shovels for the entire global AI gold rush, posting a record current account surplus. By every line in the textbook, the one should be one of the strongest currencies on Earth. Instead, it's one of the weakest in Asia. The one recently pushed past 1550 to the dollar, its feeblest level since the depths of the 2009 financial crisis. A record surplus and a currency trading like the country's in crisis. So, where is all of the money going?
The economist Brad Setszer has a name for the answer, Drram Dollars. He compares it to the petro dollar recycling in the 1970s when oil producers sold crude for dollars and parked the proceeds back in American assets rather than bringing them home. The same thing is now happening with memory chips. When Samsung and SKH Heinik sell billions of dollars of chips to American tech companies, they're paid in dollars, but instead of shipping those dollars home and converting them into one, which would push the currency up, they increasingly leave them offshore to fund their own overseas expansion. The result is a phantom surplus real on the balance sheet, but a great deal of it never actually arrives in Korea. So, it never turns into demand for the one.
And at the same time, the ants, who aren't busy detonating their accounts on leveraged ETFs at home, are doing something that pushes the one down further. They're buying America. Korean retail investors have become some of the most enthusiastic buyers of US tech stocks anywhere in the world, including with a certain poetry, Nvidia, the company their own chip makers supply. Every time a saver in soul buys a US share, they first have to sell one and buy dollars to do it. So the country is in effect exporting its savings and importing pressure on its own exchange rate that leaves the Bank of Korea with a nasty problem. A weak one makes imported energy dramatically more expensive, which feeds straight into domestic inflation, which is why last week the central bank was forced to raise interest rates directly into a falling stock market because it had no choice.
The obvious escape would be to open the currency up, let it trade freely offshore, and let global markets sort out the imbalance. But Korea won't do it. And the reason is a scar. In 1997, during the Asian financial crisis, known locally as the IMF trauma, the one lost half of its value in months, and the country needed an IMF bailout to avoid default. That memory runs deep enough that Seoul still refuses to allow a fully convertible, freely traded offshore one. And that single decision is the main reason MSCI, the company that decides which markets count as developed, still files South Korea, the best performing major market in the world, under emerging, not because it isn't rich or sophisticated enough, but because a fund manager in New York or London can't freely trade the one at 2 in the morning.
So, Korea finds itself in a strange trap. A worldbeating technology sector generating billions in surpluses that never quite come home. A currency trading as if there's a crisis and a population busily converting its savings into dollars to buy shares in the American companies its own factories are helping to build.
If you want to know how all of this ends, you'll have to stop looking at the traders in Soul and look at the people actually buying the chips. Samsung and SKH Highix look invincible right now and their earnings are real. But that success rests almost entirely on the spending decisions of a very small group of American companies, the Hyperscalers, Amazon, Google, Meta, and Microsoft. Between them, they're pouring hundreds of billions of dollars a year into AI infrastructure, and their appetite for memory chips is what turned two Korean firms into trillion dollar companies. So stripped of the national flag, the entire South Korean stock market has become a leverage bet on the server budgets of about four American corporations.
The consensus is that this spending simply continues forever because AI is the new industrial revolution. But the market is beginning to twitch. This week, Taiwan Semiconductor, the company that actually manufactures Nvidia's chips and about as close to the beating heart of the AI boom as you can get, reported that quarterly profit had jumped by around 77% to a record. It beat estimates. It raised its guidance. It told the market in effect that the boom was entirely intact, and the stock went down. Now, the reason it went down is the interesting part. It wasn't that the numbers were bad, they were spectacular. It's that alongside those numbers, TSMC also announced that it would have to spend a great deal more on capex and it warned that margins would come under pressure, and investors looking at a company posting record profits decided to worry about the cost of staying ahead.
When a business this good gets sold off on the price of running to stand still, it tells you that the market has started asking a question it spent three years refusing to ask. What does all of this actually earn? Because staying on this treadmill requires the hyperscalers to keep spending at a ferocious pace, and they have been. The four biggest have already laid out around $376 billion in capital spending, with the wider industry on track for something closer to 725 billion. For 3 years though, AI has lived a heavily subsidized life. The frontier labs, the open AIs and anthropics of the world, charged flat monthly fees, often well below what it actually costs to run the models. It's a business that so far consumes far more capital than it produces. And if you offer people near unlimited access to a billion dollar supercomput for $20 a month, they'll cheerfully use it to write birthday poems for their cats. But the investors funding all of this have started asking for their money back.
To stop the bleeding, AI companies have begun shifting from flat fees to usagebased pricing, effectively ending the subsidy. And the result is roughly what you'd expect from installing a water meter in a house that used to get its water for free. People use less, and companies suddenly staring at metered AI bills have started shifting work onto cheaper models, including open-source Chinese ones. The uncomfortable question hanging over the whole edifice is this. What if AI turns out to behave less like high margin software and more like electricity? A commodity where the models are largely interchangeable and nobody has much pricing power. In that world, the value doesn't vanish. It just doesn't go to the people who spent the money. It flows to everyone using the tools as competition drives the price towards zero. The labs that built the models and the companies that funded the infrastructure would have created something genuinely transformative and captured almost none of the returns.
If the hyperscalers begin to suspect that this is what's happening, they won't put out a press release. They'll simply reduce their spending. They'll push the next data center back a quarter or two just to see how demand develops. And if that happens, the first tremor won't be felt in Silicon Valley. It'll be felt in soul. Memory chips aren't a stable subscription style software business. They're much closer to an industrial commodity. And the memory industry is famously viciously cyclical. They're long booms followed by brutal busts. Prices don't drift down gently when demand softens. They tend to fall off a cliff. And a cliff is a bad place to be standing when your national stock index is two chip stocks wide and strapped to billions of dollars of automated double and triple levered retail ETFs. Because those levered funds don't have opinions and they don't wait. The same machinery that amplified returns every step of the way up runs exactly as hard in reverse. Forced to sell into the fall at the close mechanically whether anyone wants them to or not. The way up was a choice that millions of people made. The way down, if it comes, won't ask anyone's permission.
So, where does that leave the people at the center of all of this? For most of this year, being an ant paid off spectacularly. The market went up and up, and the people who piled in with borrowed money made more than everyone else. In Hong Kong, a 26-year-old who'd put his savings into the leveraged SKH fund told Bloomberg with no apparent irony that he had already had the car and could now afford the Patek Philippe. That was the mood. The coin was weighted in your favor. Everyone could see it was weighted in your favor, and the only question anyone was asking in the cafes of soul was whether it was too late to bet more. Then the machinery went into reverse, and roughly 1.2 million of them got the margin call.
The president, the man who campaigned on getting the Cosby to 5,000 and then presided over it sailing past 9,000, appeared this week to say that the market was, and I'm quoting him here, quite unstable, and that after such a large rise in such a short time, it would need time and fluctuation to stabilize, which is the sound of a man who ran on the number going up, gently explaining that the number could also come down.
The line that I keep coming back to is from an ordinary retail investor quoted in the Financial Times. He's 37. He didn't even use the leverage. He just bought Samsung and SKH Highix with his own money back in February. He watched a paper profit of 60 million1 sink to 20. And asked if he'd sell, he said no. He'd hold on to the end of the year because even after everything, he still expected these shares to do better than leaving the money in the bank. And he might be right. That's what makes this whole story so hard to look away from.
If you found this video interesting, you should watch my video on private credit next. Don't forget to check out our sponsor SY using the link in the description. Have a great day and see you in the next video. Bye.