In July the Korean stock market had its third-worst month in recorded history. On the last trading day of that same month, it had its single best day ever.
Both of those sentences are true, and they happened eleven days apart.
I did not lose money in either direction, and I want to be honest about why: not because I saw it coming, and not because I am disciplined. I own none of the products at the centre of this story because of how my accounts are built. There was never a moment where I evaluated leverage and turned it down. It simply was not on the menu.
But something happened in those eleven days that is worth understanding no matter where you invest, because it explains why a lot of people who were right about the recovery still did not get it back.
What actually happened
On May 27, 2026, Korea listed something new: single-stock leveraged exchange-traded funds tracking Samsung Electronics and SK Hynix at twice their daily move. Fourteen of these products launched, and money went in fast — reporting puts retail inflows in the trillions of won within weeks.
Then the semiconductor rally that had carried the market all year turned over. On July 28 the KOSPI fell 10.84% to close at 6,023.66, triggering a market-wide circuit breaker. The next day it triggered another one. Back-to-back circuit breakers had never happened in the index’s history.
Regulators moved quickly. The minimum cash deposit for single-stock leveraged ETFs and ETNs was tripled from 10 million won to 30 million won, and a 20% cap on total leveraged holdings was imposed immediately rather than in August as originally planned.
And then, on July 31, the KOSPI rose 17.91% in a single day — up 1,001.89 points to close at 6,595.45. Largest one-day percentage gain in the index’s history, and the largest point gain. Foreign investors bought 7.25 trillion won of stock. Samsung Electronics rose more than 26%. SK Hynix closed limit-up.
That last day changed the month’s arithmetic completely. Before it, the KOSPI was heading for the worst month ever recorded. After it, July finished down 22.19% — brutal, but behind October 1997 (−27.25%) and October 2008 (−23.13%).
The number that tells the whole story
Here is where it stops being a market story and becomes an arithmetic one.
Over roughly the same stretch, Samsung Electronics shares fell about 15.2%. If a 2x product simply doubled that, you would expect about −30%. The leveraged ETFs tracking Samsung averaged −40.2%.
SK Hynix fell about 18.4%. Doubled, that is about −37%. Its leveraged ETFs averaged −49.4%.
| Underlying | Stock | 2× would be | Actual ETF | Gap |
|---|---|---|---|---|
| Samsung Electronics | −15.2% | −30.4% | −40.2% | −9.8pp |
| SK Hynix | −18.4% | −36.8% | −49.4% | −12.6pp |
That last column is not a market move. No company got worse. Nothing additional was lost in the underlying shares. Those percentage points were destroyed by the arithmetic of resetting leverage every single day in a market that was swinging violently.
It has a name: volatility decay. Across the fourteen products, the average one-month decline was about 47%.
Why a flat stock can still cost you money
The mechanism is simple enough to check on paper, and once you have seen it you cannot unsee it.
A leveraged ETF promises twice the daily return. To keep that promise it resets its exposure every day. That daily reset is the whole problem.
Take a stock that falls 20% one day and rises 25% the next. Check the maths: 1 × 0.80 × 1.25 = 1.00. The stock is exactly where it started. A round trip, no damage.
Now run those same two days through a 2x version. The 20% fall becomes 40%. The 25% rise becomes 50%. So: 1 × 0.60 × 1.50 = 0.90.
The leveraged product is down 10% while the stock it tracks is perfectly flat.
Nothing went wrong there. No crash, no bad earnings, no fraud. Two days of ordinary movement, doubled and compounded, and a tenth of the money is gone. Now imagine a month of circuit breakers and record rallies, which is a machine for generating exactly that pattern over and over.
Which is why the record rally did not save them
This is the part that stayed with me.
On July 31 the index rose 17.91% in one day. For anyone holding an ordinary index fund, that was an enormous recovery of what the month had taken. For someone holding a 2x product that had already fallen 40% or 50%, it was not.
The reason is arithmetic that most people never have to think about: a percentage fall and the percentage rise needed to undo it are not the same number. Lose 50% and you need +100% to get back to even. Lose 75% and you need +300%.
So a leveraged holder down 49% needed roughly a 96% gain to break even. A single day of +17.91% in the index, doubled to about +36% in the product, is a large move — and still nowhere close.
The index came most of the way back in one session. The people who had bought leverage to bet on exactly that recovery did not.
The same tool, two different markets
Here is the question I actually wanted to answer, and it is not “is leverage dangerous.”
The United States has had leveraged and inverse ETFs for close to two decades. TQQQ tracks the Nasdaq 100 at 3x. NVDL tracks Nvidia at 2x. There are hundreds of these products, they trade in enormous size, and they have not repeatedly detonated a national stock index.
So why did the same category of product behave so differently in Korea?
Not because the products are defective. They did precisely what their documents said they would do — deliver twice the daily move. The difference is in everything around the product.
Age. US leveraged ETFs have been through the 2008 crisis, 2020, 2022. An entire generation of investors, journalists and regulators learned what daily rebalancing does to a multi-week holding period, mostly by watching it happen. Korea’s single-stock leveraged ETFs were nine weeks old when the crash arrived.
Warnings. The SEC’s own investor bulletin on leveraged and inverse ETFs states plainly that these funds are designed to meet their objective on a daily basis, and that holding them longer “can differ significantly from their stated daily performance objectives.” FINRA says the same thing. US prospectuses carry explicit language that these funds are designed for short-term, typically single-day, use and are not intended to be held for extended periods because of compounding effects. That is not a legal formality. It is the exact mechanism described above, printed in advance.
Concentration. This is the one I think matters most. TQQQ tracks a hundred companies. Korea’s products tracked two — and those two are, by a wide margin, the largest components of the index itself. A leveraged product on a diversified index is a bet on an average. A leveraged product on Samsung and SK Hynix, inside a market where Samsung and SK Hynix dominate the index, is something closer to a leveraged bet on the market wearing the costume of a single stock.
And when such a product has to rebalance in a falling market, it must sell more of exactly what is already falling — the same two names that move the index. That is a feedback loop. It did not cause the crash; the semiconductor rally unwinding did that. But it was an accelerant, and it was pouring on the part of the fire that was already hottest.
The same week, the same stock, the opposite end of the market
While Korean retail investors were being forced out of two-times ETFs, something almost identical was happening at the other extreme of financial sophistication — in the same stock, in the same week.
Leopold Aschenbrenner is a former OpenAI researcher who wrote the essay series that arguably defined the case for rapid AI acceleration. He is in his twenties. Silicon Valley took him seriously enough that his hedge fund, Situational Awareness, held roughly $45 billion at the start of July. Reported leverage ran as high as 400%.
His largest positions included SK Hynix — the same company underneath many of the Korean leveraged ETFs — along with CoreWeave, Micron, SanDisk and Nebius. Every one of them fell more than 35% during the month. His short positions in software names moved against him at the same time, so both sides of the book lost together.
Margin calls came from Bank of America, Goldman Sachs and JPMorgan. He was forced to unwind all of his public stock positions, selling the bulk to Ken Griffin’s Citadel at a discount. The fund fell to roughly $10 billion in a matter of days.
I find the symmetry difficult to look away from. On one end, an ordinary investor in Korea with a ten-million-won deposit and a two-times ETF. On the other, a fund with forty-five billion dollars, backed by some of the most sophisticated money in the world, run by the person who wrote the definitive argument for the very trend he was betting on.
The mechanism that removed both of them was the same. The ETF had to sell because its mandate required rebalancing every day. The fund had to sell because its brokers demanded collateral. Different trigger, identical outcome: when the asset falls, leverage forces you to sell more of the thing that is already falling, at the worst possible moment, whether you want to or not.
And this is the part that should stop anyone from feeling superior about Korean retail investors: Aschenbrenner was not wrong about AI. His thesis may well turn out to be correct. But being right about the destination does not help if leverage removes you from the vehicle before you arrive. He did not get to find out whether he was right. Neither did the person holding a 2x Samsung ETF who intended to wait for the recovery.
The failure in both cases was not a lack of intelligence, or research, or conviction. It was structure.
What I keep coming back to
The honest reading of this is not “leverage is evil.” A leveraged ETF used the way its documents describe — entered deliberately, held a day or two, closed on purpose — does what it says. Professional traders use these tools and build the decay into their plans on purpose. The decay is not a hidden trap. It is disclosed, predictable arithmetic.
The failure was a mismatch. A short-horizon instrument met a buy-and-hold instinct. Someone who bought a 2x Samsung product in June intending to hold it “until things recover” was using a tool exactly opposite to its design. The tool worked. The use case was wrong.
And I want to resist the temptation to feel clever here, because I did not avoid this through insight. I avoided it because my money goes into broad index funds on a fixed schedule inside retirement accounts, and that structure never presented me with the option. I did not out-think the risk. I was never standing where it could reach me.
I wrote last week about the machine that actually moves my money and how little of it can be predicted. This is the same lesson arriving from a different direction. That distinction matters more than it sounds. Being right in the moment requires you to be right in every moment. Being structurally out of the way only requires one good decision, made once, a long time ago.
Frequently asked questions
What is volatility decay?
Leveraged ETFs reset their exposure daily. Over a single day, 2x leverage works exactly as advertised. Over many days of up-and-down movement, compounding those doubled daily returns produces a drag that can leave you with a loss even when the underlying asset is unchanged over the full period.
If the stock recovers fully, does the leveraged ETF recover fully?
Usually not. A stock that falls 20% and then rises 25% is back to even. A 2x version of that same stock over those two days ends down about 10%. The deeper the fall, the worse the problem: recovering from a 50% loss requires a 100% gain.
How far did the Korean market actually fall in July 2026?
The KOSPI finished July down 22.19%. That is the largest monthly decline since October 2008 (−23.13%) and October 1997 (−27.25%), but it is not a record — the index rose 17.91% on July 31, the largest single-day gain in its history, which substantially reduced the monthly figure.
Are leveraged ETFs always a bad investment?
Not inherently. They are built and disclosed as short-term trading instruments and behave as designed when used that way. The risk appears when they are held for extended periods through choppy markets — a mismatch between the tool and the use, rather than a flaw in the tool.
How can I tell whether a fund is leveraged?
In Korea, names typically include “레버리지” or a 2X reference. In the US, look for “2x,” “3x,” “Ultra,” or “UltraPro.” If you are unsure, the fund’s official fact sheet states its daily target multiplier explicitly — that one number tells you how it will behave.
Does this change anything about how I invest?
No. I make the same contribution on the same day into the same broad index funds, which is what I did through July and what I will do again this month. If anything, watching this made the plainest possible structure look more defensible, not less.
I am not a financial adviser, and nothing here is investment advice. I am an engineer writing down what I am learning as I go. Please make your own decisions, and speak to a qualified professional about your own situation.