Korea Listed 2x Samsung ETFs in May. Ten Weeks Later They Were Down 39% While Samsung Fell 12%.

On May 27, 2026, South Korea listed its first leveraged exchange-traded funds tied to individual stocks. Sixteen of them, from eight asset managers, plus two exchange-traded notes, all offering twice the daily move of Samsung Electronics or SK Hynix.¹ Nine weeks later, on July 31, Samsung Electronics had fallen 12.1% from the session before those products listed. A 2x daily-reset product tracking it was down 39.0%.

That gap is what the arithmetic of daily rebalancing produced on the actual price path, computed from closing prices. The naive expectation, twice a 12.1% decline, would have been 24.2%. The extra 14.8 points came from resetting the leverage every session through a stock whose realized volatility ran at 117% a year. For SK Hynix the same calculation gives 16.3% for the stock and 49.9% for the 2x product.

Meanwhile a second investor, holding the same 2x exposure through a margin loan rather than a fund, faced a different problem entirely. That position could not decay, but it could be closed. At a 25% maintenance requirement, an SK Hynix margin position hit its limit on July 30. The next session, SK Hynix closed up 29.95%, locked at Korea’s daily price limit, as the KOSPI posted the largest one-day gain in its history.

The boom was real, and that is what makes it instructive

Nothing here requires the AI buildout to have been imaginary or the Korean chipmakers to have been bad businesses. SK Hynix announced second quarter results on July 29, 2026: revenue of ₩79.3 trillion, operating profit of ₩60.5 trillion, and an operating margin of 76%, all records. Operating profit was up about 557% from a year earlier, roughly a sixfold increase.³

The stock fell nearly 20% intraday and closed down 9.6%. This is the first thing that confuses people about the episode, and it has a boring explanation covered in Do Valuations Still Matter?: a share price reflects expectations of all future cash flows, so the relevant comparison is not profit against last year but profit against what the price already assumed. A sixfold increase can disappoint.

Cyclical businesses make this worse in a specific way. Memory is a commodity with a long history of capacity cycles. When earnings are near a cyclical peak, a low forward price-to-earnings multiple is ambiguous: it can mean the shares are cheap, or it can mean the market doubts the denominator. Both readings fit a single-digit multiple, and only one of them is good news.

Two companies became the market, and nothing capped them

Korea’s Financial Services Commission published the concentration series in July. Samsung Electronics and SK Hynix together accounted for 34% of KOSPI market capitalisation at the end of 2025, 41% at the end of April 2026, 49% by May 26, and 52% by July 15.² It peaked higher still, above 54% in mid-June.

An investor holding a KOSPI index fund at that point owned something the label did not describe. The overlapping bets were: two companies, one industry, one product cycle, one global capital-spending narrative, one country, one currency, and a shared sensitivity to Chinese competition. Ticker count had nothing to do with it, a point The Tech Bro Portfolio makes about US funds.

There was no mechanical brake on this, and that was a decision rather than an oversight. The Korea Exchange proposed a 30% single-stock cap for the KOSPI 200 in early 2020, when Samsung’s weight reached 33.5%, then dropped the plan in April 2020, publishing capped versions for overseas use while leaving the domestic index uncapped.⁴

The contrast with what a US investor actually buys is sharp. The iShares MSCI South Korea ETF tracks the MSCI Korea 25/50 index, which caps any single issuer at 25% and the sum of all issuers above 5% at 50%, in order to satisfy US regulated investment company diversification rules. As of March 31, 2026, Samsung sat at 22.61% and SK Hynix at 19% inside that capped index.⁵ A US fund buyer held a version of Korea that a tax rule had already diversified. A domestic index buyer did not.

There is precedent. Nokia’s weight in the Helsinki exchange ranged from 9% to 72% over the period Lally and Swidler studied, and they show that a dominant constituent distorts the statistical properties of the whole index, with the stock’s own beta peaking when it is around 32% of the market and cost-of-equity estimates differing by more than 1,000 basis points depending on the treatment.⁶

The leverage rule pointed straight at the concentration

Korea’s decision to permit single-stock leveraged products was announced as policy in January 2026 and enacted by enforcement decree on April 21, effective April 28. Leverage was capped at ±2x. Eligibility was restricted to stocks in the top 10% by market capitalisation, top 5% by trading volume, investment grade, and top 1% by derivatives volume.⁷

Those criteria are sensible in isolation: they select large, liquid, heavily traded names. Applied to Korea in early 2026, only two companies qualified. The stocks retail investors were newly permitted to lever were the same two stocks that already made up half the index. Concentration and leverage were pointed at the same place by the design of the rule.

The regulator published the decay arithmetic before anyone bought

The day before listing, the FSC issued an investor notice explaining what these products do. It warned that they invest in a single stock and are therefore exposed to one firm’s earnings and one industry’s cycle; that gains and losses expand to ±2x the underlying’s move; and that as a stock repeatedly rises and falls, invested capital is eroded. It gave a worked example: a 30% gain followed by a 30% loss leaves an unlevered holder down 9% and a 2x holder down 36%. It noted the maximum single-day loss was 60%, given Korea’s ±30% price limit.⁸

Access was gated: two hours of mandatory pre-training and a ₩10 million basic deposit. Products were barred from plain “ETF” branding and required to display “single stock” and “leverage” prominently.¹

The aggregate value of the sixteen leveraged funds went from ₩4.4 trillion at listing to ₩11.9 trillion by July 15.² Disclosure is not the same instrument as restraint, and this is a large natural experiment demonstrating the difference.

The behavioural research anticipates the pattern. Greenwood and Shleifer examined six independent measures of investor return expectations from 1963 to 2011 and found them strongly positively correlated with past returns and with the level of the market, and strongly negatively correlated with model-based expected returns.⁹ Investors read a higher price as evidence of more opportunity. Separately, stocks with lottery-like payoffs attract retail demand and go on to earn poor risk-adjusted returns.¹⁰

What a daily reset does to a stock moving 120% a year

A 2x daily-reset fund promises twice the underlying’s return each day. Over more than one day the outcome depends on the path, because the position is rebalanced back to 2x every session, which mechanically adds exposure after gains and cuts it after losses. Avellaneda and Zhang derived the exact relationship: the divergence between a leveraged fund and a simple multiple of the underlying is driven by realized variance, not by holding period.¹¹ The mechanism is worked through in detail in the HEDGEFUNDIE guide.

The Korean case is instructive because the underlying was a single stock rather than an index. Measured from May 26, the session before listing, to July 31, realized annualized volatility was 117% for Samsung and 138% for SK Hynix. Broad-index leveraged products, which is what almost every explainer discusses, sit on underlyings closer to 20%. Variance drag scales with variance.

Here is what that produced, computed from closing prices over the product’s first nine weeks of existence.

Two points of context keep this honest. First, Korea’s stocks were not the wild ones. The FSC’s own table, covering May 26 to July 10, put SanDisk at 131%, Micron at 123% and Kioxia at 118% against SK Hynix’s 113% and Samsung’s 96%.² The volatility was a memory-sector phenomenon, and the Korean names were the calmest of the group. The leverage is what converted it into permanent loss.

Second, this is a general property rather than a Korean one. Bessembinder studied US levered single-stock ETFs from inception through mid-2025 and found they underperformed their leverage benchmarks by about 0.79% a month, roughly a third of it from daily rebalancing and the rest from fees and frictions.¹² Korea produced a far larger gap over a far shorter window because realized volatility was several times the US average.

The same exposure on margin fails a different way

A leveraged fund cannot be margin called. The most an investor can lose is the amount invested. Borrowed money is different: it introduces a threshold at which someone else decides you are finished.

Running a 2x margin position over the same closing prices, with a 25% maintenance requirement, the SK Hynix position breached its threshold on July 30, with the stock 35.6% below its starting point. A 35% house requirement, which brokers commonly apply to volatile or concentrated positions, would have closed both Samsung and SK Hynix positions two days earlier, on July 28. The mechanics of maintenance requirements and forced sales are covered in Personal Leverage.

The computed date is not a coincidence. Korean brokerages’ forced liquidations of retail margin positions reached ₩103.8 billion on July 30, about 7.5 times the ₩13.9 billion recorded on July 28.¹³ Across the ten largest brokerages, the number of accounts hit by forced liquidation rose from 14,670 in May to 21,615 in June, and the value liquidated rose from ₩56.5 billion in January to ₩393.5 billion in June. Among investors aged 70 and over, the value rose 791%.¹⁴

This is the mechanism that turns a decline into a rout, and it has been measured directly elsewhere. Using account-level data from China’s 2015 crash, Bian, He, Shue and Zhou found that investors sell heavily as their accounts approach their leverage limits, and that stocks disproportionately held by constrained accounts suffered abnormal price declines that subsequently reversed. Stocks in the top decile of fire-sale exposure underperformed the bottom decile by roughly five percentage points within 10 to 15 trading days, then recovered over the following month.¹⁵ The selling was about collateral, not about the companies.

Research on 1929 reaches a compatible conclusion by a different route: modelling the 1929 market, Borowiecki, Dzielinski and Tepper find that tightening margin requirements through that year, combined with price declines in September and early October, left enough investors constrained to tip the market into instability.¹⁶

A price limit means a 2x product is not 2x when it matters

Korea applies a ±30% daily price limit to individual stocks, and to ETFs and ETNs, measured against the previous close.¹⁷ On July 31, SK Hynix closed up 29.95%. It was pinned at the ceiling.

A 2x product tracking it would have needed to rise 59.9% that day. Its own market price could not rise more than 30%. The structural point generalises: in a market with symmetric price limits, a leveraged product cannot deliver its stated multiple on precisely the days when the underlying moves most, in either direction. The tracking promise is weakest exactly when it is being tested.

Run the path yourself

The lab below carries the real Korean closing prices, and three synthetic paths for comparison. It shows both structures at once: what a daily-reset fund did, and the session on which a margin position at the same leverage would have been closed.

Try the steady uptrend preset. At 2x the daily-reset product beats twice the cumulative return, because rebalancing into a rising market compounds in the holder’s favour. The claim that leveraged funds always bleed value is wrong, and repeating it makes the real risk easier to dismiss. What is true is narrower: the gap is driven by realized variance, so a volatile path costs you and a smooth one does not.

Retail sold the record rebound

On July 31 the KOSPI rose 1,001.89 points to 6,595.45, a gain of 17.91%. It was the largest single-day advance in the index’s history in both point and percentage terms; the previous record was 11.95% on October 30, 2008. Samsung closed up 26.81% and SK Hynix up 29.95%.¹⁸

Foreign investors bought a net ₩7.25 trillion that day and individual investors sold a net ₩8 trillion, both single-day records.¹⁸ Some of that selling was chosen. Some of it was not: an investor liquidated on July 30 had no position left to hold on July 31. This is the practical content of the distinction between volatility and leverage. Volatility is the risk that prices move against you. Leverage is the risk that someone else decides when you sell.

The rebound did not repair the month. The KOSPI ended July down 22.2%, its steepest monthly fall since October 2008, and 27.6% below its June 22 record close of 9,114.55.¹⁹ It is worth being precise here because headlines written on July 30 called it the worst month in KOSPI history; after the final session it was not, and October 1997 still holds that record.

What a diversified investor actually experienced

For a reader outside Korea, what did all of this do to an ordinary portfolio? July 2026 total returns, in US dollars:

A globally diversified investor lost 0.7% in the month Korea lost 22.2%. Korea has an advanced economy and two of the strongest memory manufacturers on earth, and it is a low single-digit share of global market capitalisation. Holding it at that weight, rather than at conviction weight, is what produced the top row of that table.

The two emerging-market rows are worth dwelling on. Both funds are sold as emerging-market exposure. They were 4.7 percentage points apart in a single month, because FTSE classifies Korea as a developed market and MSCI classifies it as emerging. The emerging-markets guide covers that split; July 2026 is what it costs.

Four questions investors collapse into one

Most of the confusion in this episode comes from treating one judgement as four. They are separate, and a yes to the first says nothing about the rest.

1. Is this a good business? Samsung and SK Hynix are formidable companies with genuine technological advantages in high-bandwidth memory. This question was answered yes, and answering it correctly protected nobody.
2. What is already in the price? A sixfold profit increase met a decline because expectations had run further. This is the question that decides your return.
3. How much of my portfolio should this be? Independent of the first two. A great business at a fair price still should not be half of anything.
4. What structure am I using to hold it? Spot, fund, margin and daily-reset leverage have different failure modes on the identical price path, as the two computed columns above show.

When a country or sector bet is reasonable, and how to size it

None of this argues for avoiding Korea. Most globally diversified investors already own Samsung and SK Hynix at roughly their share of world market capitalisation, which is the allocation that requires no forecast.

A deliberate overweight can be defensible when an investor treats it as an active bet rather than diversification, measures the overlap they already hold through international and emerging-market funds, writes down a maximum weight in advance, can tolerate years of underperformance, and rebalances on a schedule instead of adding because the position rose. Position Sizing covers the sizing arithmetic.

For speculative positions specifically, a separate sleeve of roughly 1% to 5% of investable assets is a workable guardrail. That range is a planning constraint rather than an optimal allocation: it exists so that a total loss is survivable. The test is whether a position going to zero would change your emergency reserves, your near-term goals, your retirement date, or your ability to stay invested through the next decline. If it would, the position is too large.

One trap deserves naming. A semiconductor engineer in Korea might hold salary and job security tied to the memory cycle, employer shares, a domestic pension, local property, a KOSPI index fund, and a leveraged Samsung product. Six lines on six statements, one underlying bet. The same pattern applies to US technology workers, as the human-capital guide sets out.

Who these products are actually for

Daily-reset single-stock leveraged products have a legitimate but narrow constituency: traders expressing a defined view over one or a few sessions, with explicit entry, exit and maximum-loss rules, who monitor the position actively and can absorb a near-total loss. FINRA’s long-standing guidance is that leveraged and inverse ETFs that reset daily “typically are unsuitable for retail investors who plan to hold them for longer than one trading session, particularly in volatile markets.”²⁰

US regulators have been specific about the single-stock version. SEC Commissioner Caroline Crenshaw wrote in 2022 that with these products “investors’ returns over a longer period of time might be significantly lower than they would expect based on the performance of the underlying stock,” that these effects “are likely to be especially pronounced in volatile markets,” and that it “would likely be challenging for an investment professional to recommend such a product to a retail investor while also honoring his or her fiduciary obligations or obligations under Regulation Best Interest.”²¹ The SEC’s Investor Advisory Committee later reported that as of January 31, 2023, retail accounts made up 92% of holders across 26 of the most popular single-stock ETFs, and described the products as a way for retail investors to “access leverage without margin accounts or options trading permission.”²²

They are a poor fit for:

• Anyone holding for more than a few sessions.
• Retirement savings, or money needed within a few years.
• Investors already concentrated through employment, employer stock or a sector fund.
• Anyone buying because the underlying recently rose.
• Anyone relying on a low forward multiple without normalising cyclical earnings.
• Anyone who cannot explain the daily reset in their own words.
• Anyone trying to recover a previous loss quickly.

Frequently asked questions

Do leveraged ETFs always lose money over time?

No, and the claim is worth retiring. On a path that trends steadily in the fund’s direction, daily rebalancing produces more than the naive multiple. The lab above shows a case at 2x where the daily-reset product beats twice the cumulative return. The accurate statement is that the divergence is driven by realized variance: choppy paths cost you and smooth ones do not.

Was the Korean crash caused by the leveraged ETFs?

The evidence does not support that as a clean causal claim, and the FSC published data cutting against it: over the same weeks, SanDisk, Micron and Kioxia were all more volatile than SK Hynix and Samsung despite having no Korean leveraged products attached. What the leverage did was determine how much investors lost from a repricing that was under way across the global memory complex, and forced-liquidation data shows it amplified the selling.

Should I buy Korean stocks after a 22% monthly decline?

That is a different question from everything above, and this guide takes no view on it. What can be said is that a decline does not by itself make an asset cheap, that memory earnings are cyclical so trailing multiples mislead near a peak, and that if you already hold a global equity fund you own Korea at market weight without doing anything.

Is a 2x fund the same as borrowing to buy twice as much?

No, and the difference is the point of this guide. The margin position keeps constant share exposure and can be liquidated when equity falls below the maintenance requirement. The fund rebalances to constant leverage daily, cannot be liquidated, and loses value on volatile paths even when the stock ends flat. On the identical Korean prices, one finished down 39% and the other was closed out on July 30.

How do I tell whether my index fund is concentrated?

Look through to holdings rather than at the fund name. Check the top ten weights, the single largest issuer, and whether several of your funds hold the same companies. Summitward’s Portfolio X-Ray does this across your holdings; the general framework is in Concentration Risk.

Could this happen in the US market?

Single-stock leveraged ETFs have traded in the US since 2022, and they arrived under Rule 6c-11 without a specific Commission vote or public comment period, which Commissioner Crenshaw noted at the time.²¹ US index concentration is lower than Korea’s was, though not low. The transferable lesson is about structure rather than geography: concentrated index, levered product, and forced selling combine the same way wherever they occur.

Key takeaways

• Separate the business from the price. SK Hynix reported a sixfold profit increase and fell 9.6% the same day, because the price already assumed more.
• Look through to the holdings. Two companies reached more than half of the KOSPI, and the exchange had considered a single-stock cap and dropped it.
• Realized variance opens the compounding gap. At 117% and 138% annualized volatility, 2x products lost 39% and 50% while their underlyings lost 12% and 16%.
• Leverage adds a failure mode that volatility does not. A fund holder rides it out at a loss. A margin holder was closed out on July 30, the session before the largest one-day gain in KOSPI history.
• Disclosure is not restraint. The regulator published the exact decay arithmetic before launch, and holdings still grew from ₩4.4 trillion to ₩11.9 trillion.
• Diversification did its job. A total world fund fell 0.7% in the month Korea fell 22.2%.
• Size positions for being wrong. The test is whether a total loss would change your plan, not how good the upside looks.

Related guides

• Personal Leverage for maintenance requirements, the drawdown that triggers a call, and a margin stress test on your own account.
• HEDGEFUNDIE’s Excellent Adventure for the daily-reset mechanism in depth, and live evidence on index leveraged funds.
• Concentration Risk for measuring how much of your wealth rides on one outcome.
• Emerging Markets Allocation for why FTSE and MSCI disagree about Korea and what that does to your portfolio.
• When a Value Fund Becomes a Memory-Chip Fund for the same cycle producing the same concentration inside a US fund.
• But What About Japan? for what a single national market can do over decades.
• Compensated vs. Uncompensated Risk for whether a risk is one you get paid to bear, and a speculation sandbox.

Sources and method

1. Korea Financial Services Commission, investor notice on single-stock leveraged products and launch announcement, May 26, 2026. Sixteen ETFs from eight asset managers (fourteen long, two inverse) plus two ETNs from Mirae Asset Securities, listed May 27, 2026; two hours of mandatory pre-training, a ₩10 million basic deposit, and naming restrictions. korea.kr
2. Financial Services Commission, Ministry of Economy and Finance, Financial Supervisory Service and Bank of Korea, joint measures on single-stock leveraged products, July 16, 2026. Source of the concentration series (34% at end-2025, 41% end-April, 49% on May 26, 52% on July 15), the growth in aggregate listed value from ₩4.4tn to ₩11.9tn, and the volatility table for May 26 to July 10 (SanDisk 131%, Micron 123%, Kioxia 118%, SK Hynix 113%, Samsung 96%, annualized from daily returns of the underlying stocks). The same release suspended new listings and banned promotional marketing, thirteen days before the July 29 emergency meeting. fsc.go.kr
3. SK hynix, second quarter 2026 results, announced July 29, 2026: revenue ₩79.3187tn, operating profit ₩60.5426tn, operating margin 76%, against year-earlier operating profit of ₩9.2129tn. news.skhynix.com
4. Korea Exchange proposed a 30% single-stock cap for the KOSPI 200 in early 2020, when Samsung Electronics’ weight reached 33.5%, and abandoned it in April 2020 in favour of capped variants for overseas use. Korea Herald, April 2, 2020
5. MSCI Korea 25/50 Index caps any single issuer at 25% and the aggregate of issuers above 5% at 50%, to support US regulated investment company diversification requirements. Samsung Electronics 22.61% and SK Hynix 19% as of March 31, 2026. This is the index tracked by the iShares MSCI South Korea ETF (EWY). msci.com
6. Lally, Martin, and Steve Swidler. “Betas, market weights and the cost of capital: The example of Nokia and small cap stocks on the Helsinki Stock Exchange.” International Review of Financial Analysis 17, no. 5 (2008): 805–819. Studies a period in which Nokia’s market weight ranged from 9% to 72%.
7. Korea Financial Services Commission, Capital Markets Act enforcement decree amendment approved April 21, 2026, effective April 28, 2026. Leverage capped at ±2x; eligibility limited to stocks in the top 10% by market capitalisation, top 5% by trading volume, investment grade, and top 1% by derivatives volume. The underlying policy was announced January 30, 2026. fsc.go.kr
8. Korea Financial Services Commission, “Points to note when investing in single-stock leveraged products (ETF/ETN),” May 26, 2026. Source of the lack-of-diversification and loss-amplification warnings, the statement that capital is eroded as a stock repeatedly rises and falls, the worked +30% then −30% example giving −9% unlevered against −36% at 2x, and the maximum 60% single-day loss. fsc.go.kr
9. Greenwood, Robin, and Andrei Shleifer. “Expectations of Returns and Expected Returns.” The Review of Financial Studies 27, no. 3 (2014): 714–746.
10. Bali, Turan G., Nusret Cakici, and Robert F. Whitelaw. “Maxing out: Stocks as lotteries and the cross-section of expected returns.” Journal of Financial Economics 99, no. 2 (2011): 427–446. Kumar, Alok. “Who Gambles in the Stock Market?” The Journal of Finance 64, no. 4 (2009): 1889–1933.
11. Avellaneda, Marco, and Stanley Zhang. “Path-Dependence of Leveraged ETF Returns.” SIAM Journal on Financial Mathematics 1, no. 1 (2010): 586–603. Derives the exact relationship linking a leveraged fund’s return to the multiple of the underlying’s return and its realized variance.
12. Bessembinder, Hendrik. “Returns to Constant Leverage Strategies: General Principles and Application to Levered Single-Stock ETFs.” SSRN working paper 5369417. Reports that long levered single-stock ETFs underperformed their leverage benchmarks by about 0.79% a month, of which roughly 0.26 points came from daily rebalancing and 0.53 points from fees and frictions, over a sample running from fund inception to mid-2025. Caveats: this is an unrefereed preprint, and SSRN blocks automated retrieval, so these figures are as consistently reported across independent secondary summaries rather than read from the paper directly. Nothing in this guide depends on them; the Korean figures are computed here.
13. Forced liquidations of retail margin positions at Korean brokerages reached ₩103.8bn on July 30, 2026, roughly 7.5 times the ₩13.9bn recorded on July 28. Seoul Economic Daily, August 1, 2026
14. Account-level forced-liquidation data for the ten largest Korean brokerages, obtained by National Assembly member Kim Sang-hoon: 14,670 accounts in May 2026 rising to 21,615 in June (+47.3%); value liquidated ₩56.5bn in January rising to ₩393.5bn in June (+596%); investors aged 70 and over up 791%. Herald Business, July 28, 2026. Note that these are forced liquidations of positions, not account closures; several larger figures circulating on aggregator sites could not be traced to a primary source and are not used here.
15. Bian, Jiangze, Zhiguo He, Kelly Shue, and Hao Zhou. “Leverage-Induced Fire Sales and Stock Market Crashes.” NBER Working Paper 25040 (2018). Account-level data on brokerage- and shadow-financed margin accounts during China’s 2015 crash. Now subsumed into Bian, Da, He, Lou, Shue and Zhou, “The Drivers and Implications of Retail Margin Trading,” The Journal of Finance, doi 10.1111/jofi.70049. The five-point decile spread cited here is from the NBER working paper. nber.org
16. Borowiecki, Karol Jan, Michal Dzielinski, and Alexander Tepper. “The great margin call: The role of leverage in the 1929 Wall Street crash.” The Economic History Review 76, no. 3 (2023): 807–826. A model-based argument that constrained investors tipped the market into instability, rather than account-level evidence, which does not exist for 1929.
17. Korea Exchange applies a ±30% daily price limit to stocks, depositary receipts, ETFs and ETNs on the KOSPI and KOSDAQ markets, measured from the previous close. Market-wide circuit breakers trigger at 8%, 15% and 20% index declines; a sidecar suspends program orders when KOSPI 200 futures move 5%. Circuit breakers were activated on both July 28 and July 29, 2026, the first consecutive-day activation in exchange history.
18. KOSPI closed July 31, 2026 at 6,595.45, up 1,001.89 points or 17.91%, the largest one-day gain in index history in both point and percentage terms; the previous record was 11.95% on October 30, 2008. Samsung Electronics +26.81%, SK hynix +29.95%. Foreign net buying ₩7.25tn and individual net selling above ₩8tn, both single-day records. Korea Herald, July 31, 2026
19. The KOSPI fell 22.2% in July 2026, its steepest monthly decline since October 2008. The all-time record monthly decline remains October 1997. Korea Times, August 2, 2026
20. FINRA Regulatory Notice 09-31, “FINRA Reminds Firms of Sales Practice Obligations Relating to Leveraged and Inverse Exchange-Traded Funds,” June 11, 2009. finra.org
21. Crenshaw, Caroline A. “Statement on Single-Stock ETFs.” US Securities and Exchange Commission, July 11, 2022. Also the source for the observation that single-stock ETFs reached the market under Rule 6c-11 without a specific Commission vote or public notice and comment. sec.gov
22. SEC Investor Advisory Committee, “Recommendation of the Investor Advisory Committee: Single Stock and Leveraged ETFs,” approved June 22, 2023. sec.gov
23. Computed figures and method. All price-derived numbers in this guide, including the daily-reset simulations, the margin-call dates, realized volatilities and the fund comparison table, were computed from daily closing prices for 005930.KS, 000660.KS, the KOSPI index and the listed US ETFs, with a data cutoff of the close on July 31, 2026, the last session before publication. The daily-reset series applies exactly twice each session’s underlying return with no fees, which understates the loss a real fund would post. The margin simulation assumes a constant share count, a single maintenance threshold, liquidation of the whole position at the closing price, no intraday calls and no loan interest. Realized volatility is the annualized standard deviation of daily log returns from May 26, 2026. Independent press reporting of the daily KOSPI series matches these calculations.

Editor’s note

Educational content, not investment advice, and not a recommendation on any security, fund or market. Data cutoff is the close on July 31, 2026; Korean markets were closed on August 1 and 2 and the next session was August 3, so nothing here reflects trading after that date. Simulated product returns are arithmetic exercises on historical prices rather than the results of any actual fund, and no Korean leveraged ETF or ETN is named, modelled or evaluated. Several figures widely circulated during this episode could not be traced to a primary source and are deliberately absent, including large estimates of the number of investors margin called and of aggregate retail losses; where sources disagreed, the more conservative and better-documented figure was used. Korean regulatory measures described as proposals, including a per-investor portfolio cap and a variable leverage ratio, were under study rather than in force at the cutoff date. Citations verified against primary sources on August 2, 2026.