Deleveraging in the Korean Stock Market: How it Happened and Where It Stands Now

This wave of global deleveraging started in the Korean stock market. The US market began on July 1, but looking back, the big bearish candlestick on the Korean market on June 23 was the one that truly pulled back the curtain. Clarifying its process is far more useful than debating "whether it has fallen enough."

I. Before June 23, all conditions were already in place

To understand the subsequent crash, we must first look at the market structure formed from May 27 to June 22.

First, single-stock two-times leveraged products further concentrated funds into just two stocks. On May 27, single-stock two-times leveraged and reverse ETFs with Samsung Electronics and SK Hynix as underlying assets were listed. By June 19, individual investors had cumulatively net bought approximately 8.2 trillion KRW in long leveraged ETFs (about 4.6 trillion for Hynix and 3.7 trillion for Samsung), while net buying in reverse ETFs was only about 0.3 trillion KRW.

The key point is that funds did not enter from cash, but rather shifted from more diversified semiconductor ETFs and KOSPI index ETFs to single-stock leveraged products. Investors moved from diversified industry exposure to concentrated exposure in just two individual stocks, with volatility amplified two-fold.

More subtly, there is the self-reinforcing nature of these products: The Korea Capital Market Institute estimated that the scale of Hynix-related leveraged ETFs increased by about 4.31 trillion KRW between June 10 and 19, of which about 3.6 trillion was not new subscriptions, but net value expansion driven by the rise in the underlying asset's price. In other words, even without new investors entering, the rise itself automatically creates greater demand for subsequent rebalancing.

Second, the two stocks have approached "half of the KOSPI". The market capitalization weight of Samsung and Hynix in the KOSPI rose from 34% at the end of 2025 to 49% on May 26, and further to 52% on July 15.

This is not traditional debt leverage, but it constitutes extremely strong index structural leverage: If these two stocks fall by 10%, even if the rest of the companies remain unchanged, they can directly drag down the KOSPI by about 5%.

Third, regulatory statements on June 22 became a turning point for confidence. The head of the Financial Supervisory Service of Korea publicly admitted that the approval process for related products was "too rushed," and measures to stabilize the market are being studied.

The meaning of this statement is not "immediate trading ban," but rather: the expectation of policy support for product expansion was broken. The room for product expansion by brokerages and asset managers came under question, and foreign investors began to worry that regulation would change the liquidity structure. For the first time, the market seriously assessed the risks of the negative feedback loop from two-times leveraged ETFs.

At that time, the total scale of retail leveraged investments in Korea had reached about 60 trillion KRW.

II. The Amplifier: Why Leveraged ETFs Must Aid Both Declines and Rallies

Two-times long ETFs must restore their two-times target exposure daily, which dictates that their behavior is mechanical and indiscriminate.

Assume the ETF's initial net asset value (NAV) is A, holding an exposure of 2A. After the underlying asset falls by 10%, the NAV drops to approximately 0.8A, and the market value of the original exposure becomes about 1.8A, while the new target exposure should be 1.6A—therefore, it must sell approximately 0.2A. Conversely, when the underlying asset rises, it must buy additional shares.

The scale of rebalancing is roughly proportional to "previous day's AUM × daily price change," and adjustments occur in the same direction across both spot and futures markets.

The conclusion is: When prices fall, the ETF must sell; when they rebound, the ETF must buy. The market will not gradually converge; instead, both ups and downs are amplified.

III. The Process: Prices Crash First, Debt Decreases Later

On June 23, the KOSPI fell 9.99% in a single day, with both Samsung and Hynix dropping over 12%, triggering a 20-minute comprehensive trading halt.

However, a very dangerous phenomenon occurred on this day: The amount of forced liquidations rose from about 19.9 billion KRW the previous day to 42.427 billion, while unsettled receivables actually increased by 181.6 billion, reaching 1.4792 trillion KRW, and credit financing balances remained near 38 trillion KRW.

The crash on the first day did not lead to widespread debt repayment by investors; instead, some continued to use short-term credit funds to average down during the decline. This was deleveraging of prices, not balance sheet deleveraging.

On June 24-25, the KOSPI rose 3.26% and 5.42% respectively, coming within about 2% of its high at one point. But beneath the surface rebound, two opposite things happened simultaneously: Public statistics showed forced liquidations of about 110.793 billion KRW on June 24, yet on the same day, credit financing balances actually increased by about 539.2 billion, reaching a record high of 38.6328 trillion KRW.

Old accounts were being liquidated, while new accounts were borrowing more money to bottom-fish. The peak in total market financing balances occurred on June 24, not on June 22 before the crash. This was not a healthy rebound after completing deleveraging, but more like re-leveraging midway through the deleveraging process.

From June 26 to 30, shareholding structures began to shift. Foreign capital net outflow from the Korean stock market in the first half of the year was about 70.8 billion USD, with about 12.63 billion in June alone—comprising about 7.5 billion in mutual funds, 4.35 billion in pension funds, and 1.87 billion in hedge funds.

Not all of these sell-offs were due to bearishness on Korea; more importantly, the excessive gains in Korean and Taiwanese chip stocks caused the weights of Samsung, Hynix, and TSMC in global funds to expand rapidly. Passive and active funds needed to control concentration in a single country, sector, or stock.

Meanwhile, Korean individual investors net bought about 42.4 trillion KRW in the KOSPI in June. Foreign capital and pension funds reduced risk, while Korean retail investors took over these positions using cash, financing, and leveraged ETFs—shifting risk from the balance sheets of global institutions to those of Korean residents.

On July 2, the KOSPI plummeted 7.89%, with Hynix falling 14.6% and Samsung falling 9.1%. The catalyst at the industrial level was global semiconductor profit-taking, doubts about the sustainability of AI capital expenditure, and the possibility that memory price growth rates might be peaking; but what truly amplified the drop to nearly 8% was market structure—foreign capital selling heavyweight stocks, two-times ETFs being forced to reduce exposure, market makers hedging synchronously, declining margin ratios in financing accounts, and programmatic fund position reductions.

July 7 was the second emotional turning point. Samsung released preliminary guidance indicating that Q2 operating profit might grow year-on-year by about 19 times, yet the stock price still fell 6.9% that day, briefly exceeding 10% intraday.

This indicates the market entered a phase where "good news cannot push up stock prices": It wasn't that earnings were poor, but rather that expectations had previously been too high. Investors began to worry that current earnings represented a cyclical peak, and good news was used to realize profits.

Even more alarming was the data from that day: The index had fallen about 16% from its high, but KOSPI credit financing balances remained at about 29.7 trillion KRW, only slightly lower than the late-June peak of 29.8 trillion—debt had hardly decreased.

On July 8, the KOSPI fell 5.35%, marking a decline of over 20% from relative highs, officially entering a bear market.

July 13 was the most typical instance of waterfall-style deleveraging: The KOSPI fell 8.95%, Hynix fell 15.37%, foreign capital net sold 1.73 trillion, institutional capital net sold 2.20 trillion, while individuals net bought 3.88 trillion in reverse. The two-times leveraged ETF for Hynix listed in Hong Kong fell over 30% that day, with same-direction position reductions further amplifying the downward trend of the underlying stock.

The sell-off structure can be summarized as: Active risk reduction by foreign and institutional investors + Passive exposure reduction by leveraged ETFs + Margin calls on financing accounts + Brokerage forced liquidations + Programmatic stop-losses.

On July 16, deleveraging became institutionalized. The KOSPI fell 6.37%, and the Bank of Korea raised the benchmark interest rate from 2.50% to 2.75%. The Financial Services Commission announced a restriction plan for single-stock leveraged products on the same day—suspending new product listings, banning marketing, tightening standards for LP price deviations, and raising the minimum base deposit from 10 million KRW to 30 million KRW, requiring cash.

From then on, deleveraging shifted from a spontaneous market behavior to a stage led by regulation, supported by interest rates.

IV. The Most Critical Number in This Deleveraging Wave

The KOSPI fell from 9,114.55 points on June 22 to 6,820.60 points on July 16, a cumulative 25.17%; meanwhile, total market credit financing balances dropped from a peak of about 38.63 trillion KRW on June 24 to about 34.37 trillion, a decrease of approximately 11%.

The price decline was about 2.3 times the debt reduction.

This sentence summarizes the nature of the entire market movement: What happened first was rapid deleveraging at the price and product levels, while debt reduction on investors' balance sheets lagged far behind.

This also explains why it fell for so long—every rebound involved re-leveraging, and every decline required clearing positions all over again.

V. Where Has It Reached?

Two reports from JP Morgan, eight days apart, clearly outlined the progress.

The July 21 report: The KOSPI had retraced about 28%–29% from its high. The scale of relevant Korean leveraged ETFs had dropped from 50 billion USD to 26 billion USD, estimating that this chain had completed about 75%. Stock long/short hedge fund deleveraging was more than halfway done, but positions remained above normal ranges. The core issue at that time was "how much passive selling pressure remained."

The July 29 report: Index retracements expanded to nearly 40%, but leveraged ETF scales had dropped to 17 billion USD, below the "acceptable range" of about 18 billion. The stock long/short ratio in JPM Prime samples dropped from 5.7x to 3.2x. The conclusion was that leveraged ETF clearance was complete, and hedge funds had completed about 90%.

Between these two reports, a seemingly contradictory change was revealed: The index was lower, yet the market structure was healthier. The low point on July 21 still contained significant feedback loops of "must continue selling because prices fell"; the low point on July 29 was closer to a valuation issue after completing position resets.

Several corroborating indicators pointed in the same direction: Bottom-fishing funds did not immediately rebuild leverage; the abnormal premium of VKOSPI relative to VIX began to decline; open interest in single-stock futures continued to fall.

Two other commonly misinterpreted points also deserve clarification:

Retail financing was never the primary fire source. Korean individual investor financing balances were about 20 billion USD, accounting for only 0.5% of total stock market capitalization. There was no aggressive expansion this year, and retail investors had cash, income, and overseas assets to meet margin requirements, giving them autonomy in disposal timing. Leveraged ETFs mechanically reduce positions on down days—what truly determines systemic impact is not nominal balances, but whether there exists "indiscriminate synchronous selling."

110 billion USD in foreign capital outflow: Concentration matters more than total volume. Annual net foreign capital outflow exceeded 110 billion USD, but about 90% came from just two stocks: Samsung and Hynix. As stock prices fell, their weights in the MSCI Emerging Markets Index dropped from 9.5% and 8.3% to 6.5% and 4.5%—the market lowered the pressure of authorization constraints through price adjustments itself. Therefore, a more accurate description is not "foreign capital fleeing Korea," but rather "foreign capital significantly reducing exposure to overly concentrated memory heavyweight stocks." The breadth of advancing vs. declining stocks did not deteriorate like the index did. Value and defensive sectors began to catch up, indicating this round of losses was concentrated mainly in a few heavyweight stocks, not a synchronized collapse in full-market earnings expectations.

VI. From "Who Still Must Sell" to "Can Earnings Take Over?"

Deleveraging nearing its end does not mean the index will immediately bottom out. It only means the driving force behind the decline has shifted from forced selling to macro events, earnings realization, and risk appetite.

Currently, the KOSPI's forward P/E ratio is about 5x, and estimated free cash flow multiples are also around 5x, termed "crisis-level" valuations. However, cyclical stocks often naturally appear cheap at earnings peaks, low multiples themselves do not prove the market is wrong.

The real bet lies in the duration of earnings. Current market prices largely imply: Memory prices will return to pre-AI cycle levels by early 2027. According to JP Morgan's sensitivity analysis, if the average price level in 2026 persists for another year, the valuation of key memory enterprises could increase by approximately 150 billion USD.

Therefore, the biggest source of valuation repair is not further price increases in the current period, but proving that high prices do not need to disappear so quickly.

Existing evidence has not confirmed a rapid downturn—memory spot prices are still rising, and Q3 contract prices continue to climb, albeit with slowing month-on-month growth rates.

And deleveraging nearing its end reduces forced selling "for cash flow or panic," but does not eliminate event risk. The next few days require observing the combined reaction of the Korean and US markets to the freshly released, slightly better-than-expected Microsoft earnings report (in the current market environment, one day really isn't enough to watch).