SNOWSHAGAL RESEARCH · NO.14Contents
Research cover comparing daily KOSPI returns in 2008 and 2026 at the same scale. Red bars mark days with moves above 5%; in 2008 they clustered in the autumn, while in 2026 they occurred throughout the year.
RESEARCH NO.14NOV 18, 2026 · WEDSPECIAL RESEARCH

WHY IS KOREA'S
STOCK MARKET
SO VOLATILE?

An index dominated by two stocks.
Three layers of amplification.

Each selloff had a different trigger: war, earnings or regulation. This report examines what magnified those shocks. Samsung Electronics and SK Hynix dominated the index. Above them sat three sources of potentially reinforcing flows: domestic leverage, offshore leverage and rules-based trading.

The three-layer framework is ours. Estimates of each layer's size and impact come from Daishin Securities and Yuanta Securities. A causal effect has not been established.

Snowshagal ResearchData through Nov 18, 2026 market close (KST)Main text ~15 min · Including expanded sections ~25 min

BASETWO STOCKSCH 1DOMESTICCH 2OFFSHORECH 3SYSTEMATICMASTERINDEX
Our framework: the concentrated base is covered in Section 02; the three amplification layers in Sections 03–05.
VOLATILITY
60.7%
Annualized volatility of daily KOSPI returns this year
Through Nov 6 · 1.5× the full-year 2008 level (39.2%) · Daishin Securities
±5% DAYS
40days
Days the KOSPI moved more than 5% this year
Just 13 days in all of 2010–2025 · Daishin Securities
LEVERAGE ETF
7.8%
July's estimated leveraged-ETF rebalancing demand as a share of KOSPI and KOSDAQ trading value
Modelled demand, not executed trades · Daishin Securities

THREE QUESTIONS · THREE ANSWERS

  1. 1Why the 5–10% daily swings?Two memory stocks made up about half the index and, like overseas memory peers, posted 80–100% annualized volatility. Even excluding both stocks, estimated volatility was still about 44%.01 The swings · 02 The two stocks →
  2. 2Who turned up the volume?No single culprit. Leveraged ETFs trade with the day's direction; margin accounts and TRS positions may face forced selling; systematic funds sell as volatility rises. Their actual price impact remains an estimate.03–06 The three layers and data gaps →
  3. 3Will markets settle down? What should we watch?Domestic leverage has visibly shrunk. With little data on offshore leverage or systematic flows, a sustained decline in volatility is the clearest signal.07 Regulation · END Dashboard →
INTRODUCTION

Same good news.
Different market reactions.

Hover over underlined terms for definitions, or tap on mobile. Red dotted underlines mark key concepts. Use the + panels for additional analysis.

Nvidia beat market expectations in results released early on Nov 18 Korea time. Taiwan's TAIEX and Japan's Nikkei 225 each gained about 0.3% that day. The KOSPI rose 2.0%—roughly seven times as much.

FIG IN-1First session after Nvidia earnings: one catalyst, different moves
Indices
KOSPIUp 3.3% intraday
+1.99%
KOSDAQ
+1.96%
Nikkei 225
+0.28%
Taiwan TAIEX
+0.26%
Major semiconductor stocks
SK Hynix
+4.19%
Samsung Electronics
+2.31%
TSMC
+0.23%
Close-to-close change on Nov 18, 2026.
Sources: Yahoo Finance closing prices1; intraday high: Segye Ilbo2
  • Earnings Q3 revenue was $115.8bn, up 103% y/y; Q4 guidance was $130bn ±3%. Q4 includes an extra fiscal week and excludes China data-center revenue.3 Both beat consensus ($109.1bn for Q3; $125.3bn for Q4). Shares rose more than 4% after hours.4
  • Context The first results under an interim CEO after the founder stepped down on Nov 12.4
  • Flows Foreign investors sold a net ₩330bn of KOSPI stocks; institutions bought more than ₩500bn.5
  • Caveat This is a single session after two sharp KOSPI declines. Index composition also differs across markets. The data establish a larger Korean move, not its cause.

The pattern can reverse. On Nov 16, the KOSPI fell 3.6% while Taiwan's TAIEX lost 6.0%.1 On Nov 17, the KOSPI fell another 2.6%, closing below 5,000 for the first time in nine months.6, 7 Daily comparisons change with news and flows.

The full-year picture is different: KOSPI volatility was twice Taiwan's (Section 01). Moves above 10% occurred on March 4 (−12.06%), July 28 (−10.84%) and July 31 (+17.91%).8

What this report examines

Think of an audio amplifier. War, earnings and regulation supply the signal. The higher the gain, the louder the output. Several mechanisms could have amplified Korean market moves this year.

  • The base Two memory-chip stocks accounted for about half the index (Section 02).
  • Three layers Domestic leverage (03), offshore leverage (04) and systematic trading (05) can each generate extra trades, but at different times.
  • Limits and policy We examine the data gaps (06), policy responses (07) and indicators that might signal calmer markets (END).

Investment Insight NO.32 No.32 asked 'Who sold?' This report asks 'Why did the same news move Korea so much more?'

01METER · THE SCALE

1.5× the 2008 level.
Volatility in numbers.

KOSPI swings this year exceeded those of the financial crisis—and most overseas markets.

  • 60.7% annualized volatility Calculated from daily KOSPI returns, the annualized volatility through Nov 6 was 3.7× the 2005–2025 median of 16.3%, and 1.5× the full-year 2008 figure of 39.2%.9 In daily terms, the standard deviation was about 3.8%, versus roughly 1% in a typical year and 2.5% in 2008. Moves of 3–4% became commonplace.
  • 40 days above 5% There were only 13 such days across the entire 2010–2025 period. July alone had ten this year.9
  • Twice Taiwan and Japan Measured over the same period, KOSPI volatility was twice the TAIEX and Nikkei 225, and even exceeded the U.S. Philadelphia Semiconductor Index.1
  • Research corroborates it The Korea Capital Market Institute measured first-half daily KOSPI volatility at 3.6%, up from 1.4% a year earlier. Across 36 major markets, the average barely moved, from 1.1% to 1.2%. This was not a global volatility shock of equal magnitude.10
FIG 01-1Annual KOSPI volatility: 2026 at 1.5× the crisis-year level
16.7
18.2
23.1
39.0
24.6
15.0
26.6
15.4
12.3
10.1
12.7
12.2
9.4
14.4
12.6
28.3
16.4
18.7
15.4
19.6
21.5
60.7
05060708091011121314151617181920212223242526
2026 (to Nov 6)2008Other years2005–2025 median: 16.3
Annualized standard deviation of daily KOSPI returns (%). Yahoo Finance closes yield slightly different decimals from Daishin Securities (39.0% vs. 39.2% for 2008). The 2026 figure covers only 207 sessions through Nov 6.
Sources: Yahoo Finance KOSPI daily closes8; Daishin Securities9
FIG 01-2Cross-market comparison: annualized volatility in 2026
KOSPI40 days with moves above 5%
60.7%
KOSDAQ27 days above 5%
54.3%
PHLX Semiconductor24 days above 5%
49.9%
Nikkei 2255 days above 5%
31.1%
Taiwan TAIEX2 days above 5%
30.0%
Nasdaq 1001 day above 5%
21.8%
S&P500No days above 5%
13.8%
First 2026 trading session to Nov 6; returns in each market's currency.
Source: Yahoo Finance daily index closes1
Expand · The 40 extreme days, trading halts and drawdown speedTwo charts · Circuit breakers, sidecars, VKOSPI and Samsung volatility
FIG 01-3Forty days with moves above 5%, by month
Jan
0
Feb
2 −5.33 +6.8
2
Mar
3 −7.24 −12.15 +9.69 −6.010 +5.318 +5.023 −6.5
7
Apr
1 +8.48 +6.9
2
May
4 +5.16 +6.515 −6.121 +8.4
4
Jun
5 −5.58 −8.39 +8.215 +5.223 −10.025 +5.426 −5.8
7
Jul
2 −7.93 +5.88 −5.313 −8.915 +6.216 −6.424 −5.728 −10.829 −6.031 +17.9
10
Aug
3 −5.119 −5.820 +5.9
3
Sep
0
Oct
13 −5.515 −7.628 +6.6
3
Nov
5 −5.36 −6.0
2
Up days: 17Down days: 23Moves above 10%
Source: Yahoo Finance daily KOSPI closes8 (through Nov 16; also 40 days in Daishin's tally9)
  • Trading halts and the fear gauge Circuit breakers and sidecars temporarily interrupt trading during rapid moves. The options-implied VKOSPI measures expected future volatility. KOSPI sidecars were triggered three times in 2025 and twice in 2024.11
FIG 01-4Trading halts and implied volatility: the dashboard
CIRCUIT BREAKER
16times
KOSPI: 11; KOSDAQ: 5. Equal to the combined total for both markets over 1998–2025: 16 eventsin all
SIDECAR
96times
KOSPI: 58 (28 buy-side, 30 sell-side); KOSDAQ: 38. KOSPI's total was more than 2.2× the 2008 count of 26events
VKOSPI · 2026 average
59.3
2010–2025 average: 18.0. Peak: 96.9 on June 29; 52.6 on Nov 16
Sources: Digital Daily (counts through Nov 10–11)11; Daishin Securities9; Financial News12; KRED (Nov 16 reading)13
  • Speed The KOSPI fell 44.5% over 93 sessions from its June 22 peak to Nov 6—faster than the 1997 decline (109 sessions) or 2008 (240).14
  • Single-stock volatility Samsung Electronics moved more than 7% in a session twice in 2024, but 36 times through Nov 16 this year.15

Different days, different triggers

Daily moves have three stages. This report focuses on the third.

STEP 1Initial shockWar, earnings or policy news
STEP 2Price discoveryTrading that prices in the news
STEP 3 · OUR FOCUSAmplificationAdditional trades triggered by the first move

Six sessions illustrate different initial shocks.

FIG 01-5REPORTED / ESTIMATEDSix sessions: the catalysts and the accompanying flows
Date · KOSPIReported catalystAlso observed
03.04 · −12.06%Escalation in the U.S.–Iran war and fears of an oil spike. U.S. stocks had fallen roughly 1% in the prior session16Foreigners bought a net ₩235.4bn in KOSPI shares; institutions sold ₩579.4bn17. The estimated KOSPI excluding the two giants still fell 12.7%. Single-stock 2× ETFs had not launched
06.23 · −9.99%Profit-taking in chip leaders, failed MSCI developed-market inclusion, and proposed taxation of unrealized gains18Estimated leveraged-ETF rebalancing demand: ₩9.6tn. Of non-chip stocks, 94% fell (Daishin)9
07.28 · −10.84%Closing reports cited about ₩5tn in foreign net sales and ₩2.6tn in program net sales19Of non-chip stocks, 98% fell (Daishin)9
07.31 · +17.91%Better-than-expected cloud results from U.S. megacaps eased AI spending fears. Foreign intraday net buying approached ₩7tn20Estimated leveraged-ETF rebalancing demand: ₩8.4tn (Daishin)9
10.13 · −5.46%New-product news from a Chinese DRAM maker (Yuanta)21Power equipment, nuclear and robotics stocks sold off simultaneously (Section 04)
11.06 · −6.02%AI data-center regulatory concerns after the U.S. midterms (Daishin)14Estimated mechanical selling by U.S. volatility-targeting funds and leveraged ETFs the prior session (Section 05)
The sessions were chosen for varied catalysts, not ranked by return. The last column records concurrent observations or estimates—not measured contributions to the move.
Sources: references in each row; KOSPI returns from Yahoo Finance closes8
The catalysts varied, and the market fell 12% in March—before single-stock 2× ETFs existed. Leverage alone cannot explain the year.
02INPUT · THE BASE

Two stocks were half the index.
Volatility stayed high without them.

The index's composition is the starting point. Concentration is not the whole explanation.

Concentration rivalled Taiwan's

  • KOSPI Samsung Electronics and SK Hynix rose from 35.22% of index weight at the start of the year to 55.67% on June 22.22 They still accounted for about 46% on the morning of the Nov 6 circuit breaker.23
  • Taiwan The TAIEX's two biggest members, TSMC and MediaTek, represented 45.7% on Aug 31.24 Yet TAIEX volatility was only half the KOSPI's this year.
FIG 02-1Two-stock index concentration: Korea versus Taiwan
KOSPI · Start of yearSamsung + SK Hynix
35.22%
KOSPI · June 22Closing-weight peak
55.67%
KOSPI · Nov 6Morning circuit breaker
About 46%
TAIEX · Aug 31TSMC + MediaTek
45.7%
Both figures cover the two largest index members, but dates differ; this is not a same-day comparison.
Source: EToday22, SBS Biz23, StockFeel24

The difference lies in the industries

Memory chips are cyclical commodities whose prices move sharply with supply and demand. Memory makers' earnings and share prices are therefore more volatile than those of contract foundries.

  • The Korean pair Samsung and SK Hynix recorded their highest volatility since 2010.9
  • Not unique to Korea Memory makers worldwide saw large swings. TSMC, a foundry, recorded roughly one-third their volatility (Fig. 02-2).1
  • Korea's difference Its memory leaders were not uniquely volatile. Their high volatility simply accounted for half the index.
FIG 02-2Memory stocks were volatile everywhere: 2026 annualized volatility
SanDiskMemory · U.S.
115.1%
KioxiaMemory · Japan
113.8%
SK HynixMemory · Korea
97.6%
MicronMemory · U.S.
86.2%
Samsung ElectronicsLarge memory business · Korea
82.4%
TSMCFoundry · Taiwan
33.3%
Through Nov 6, returns in local currencies. TSMC is shown as a different business model, not a direct peer.
Sources: Daishin Securities (Samsung and SK Hynix)9; Yahoo Finance (others)1

How much calmer without the two giants?

Index volatility reflects the weighted returns of its constituents. Independent moves offset; correlated moves compound. Samsung and SK Hynix moved closely together (2026 return correlation 0.82).10 ). Half the index therefore moved sharply in the same direction. What happens if both are excluded?

  • Fewer extreme days Stripping out their estimated index contribution reduces the number of days above 5% from 40 to 13–15.
  • Still unusually high The remaining stocks show estimated volatility around 44%. The KOSDAQ, which does not contain either company, recorded 54.3%.1
  • Research estimate The Korea Capital Market Institute roughly attributed one-third of the first-half rise in KOSPI 200 volatility to the two stocks.10
FIG 02-3APPROXIMATIONExcluding the two stocks reduces tail events, but volatility remains high
2026 (through Nov 6)Annualized volatilityDays above 5%Maximum drawdown
KOSPI60.7%40 days−44.5%
KOSPI ex-two stocks (approx.)About 44%13–15 daysAbout −36%
KOSDAQ54.3%27 days−47.5%
Reference: full-year 2008 KOSPI39.2%14 daysNot comparable
The middle row is our approximation, not an official index. The 2008 row covers the full KOSPI, with a different composition. See Sources & Methodology for details and limitations.
Calculated from Yahoo Finance closes1, 8; Daishin Securities (2008 volatility)9
Concentration explains much, but not everything. The next question is what further amplified a memory-led move across the broader market.
03LAYER 1 · DOMESTIC

Buy more on the way up.
Sell more on the way down.

Domestic leverage is the first layer. Leveraged ETFs rebalance with the day's move near the close; margin and CFD accounts may sell when collateral runs short.

Why do leveraged ETFs trade near the close?

  • The rule A 2× ETF targets twice the underlying asset's daily return. After a rise, its exposure falls short of twice its higher NAV; after a decline, exposure exceeds the new target. The fund therefore buys into up days and sells into down days near the close. A 1% move implies rebalancing equal to 2% of NAV. This is known as leveraged-ETF rebalancing.9
  • Example A 2× ETF with ₩10tn in NAV must buy roughly ₩1tn more exposure near the close after a 5% gain in the underlying.
  • Caveat Actual orders differ. Funds can use futures or swaps, while inflows and redemptions change trading needs. Investors do not necessarily borrow to buy the ETF; the leverage is built into the fund's exposure.
FIG 03-1SCHEMATIC2× ETFs add trades in the direction of the day's move
UP DAY · Underlying +5%₩20tnOpening exposure₩21tnAfter the rise₩22tnTarget exposureBuy ₩1tn near the closeDOWN DAY · Underlying −5%₩20tnOpening exposure₩19tnAfter the fall₩18tnTarget exposureSell ₩1tn near the close
Example assumes ₩10tn NAV. NAV rises to ₩11tn after a gain, requiring ₩22tn in exposure; it falls to ₩9tn after a loss, requiring ₩18tn. The estimated closing trade can be calculated from the intraday return.
Source: Daishin Securities (rebalancing = NAV × leverage × [leverage − 1] × daily return)9

How large did it get?

  • Assets NAV in domestic equity leveraged and inverse ETFs quadrupled in six months.9 Single-stock 2× ETFs launched on May 27 and peaked a month later.25
  • Rebalancing demand Daishin estimated demand equivalent to 5–8% of daily trading value in June–July. This uses NAV and daily returns—not actual executions.9
  • Why July was bigger Assets peaked in late June, but sharper index moves kept rebalancing demand higher in July.9
FIG 03-2Leveraged-ETF assets: three separate groups
① 55 Korea-listed leveraged/inverse ETFs tied to domestic equities
End-2025
₩9.95tn
2026.06.22
₩41.3tn
② Korea-listed single-stock 2× ETFs (Samsung and SK Hynix)
June 25 peak
₩17.6tn
11.06
₩2.8tn · 16% of peak
③ U.S.-listed KORU and SKHX (separate market)
June 18 peak
₩4.0tn
The groups differ in coverage and dates; do not add or directly compare them. Group ② comprised ₩11.5tn in SK Hynix and ₩6.1tn in Samsung on June 25. Hong Kong-listed funds are excluded for lack of asset data.
Source: Daishin Securities (KRX and LSEG)9
FIG 03-3ESTIMATEModelled leveraged-ETF rebalancing demand as a share of KOSPI/KOSDAQ trading value
5.1%
7.8%
3.8%
3.1%
4.6%
JunJulAugSepOct
Average daily demand: June ₩3.0tn · July ₩3.4tnPeak single day: June 23 ₩9.6tn · July 31 ₩8.4tn
Estimated required trading, calculated from NAV and returns, divided by market trading value. It is neither observed executions nor a contribution to volatility.
Source: Daishin Securities estimates9

Did those trades move prices? Researchers disagree

Trading demand is not the same as price impact. Heavy buying at the close may have little effect if sellers are plentiful. The evidence supports different interpretations.

Amplification case · Daishin Securities
  • Daishin examined SK Hynix's closing call auction. During unrestricted trading in single-stock 2× ETFs, down-day declines deepened near the close, while up-day prices slipped in the auction.
  • Daishin argues this asymmetry could amplify downside moves, but acknowledges that the evidence is weak. Differences between periods were not statistically significant (t-statistics −1.3, −1.2).9
Inconclusive case · Research institutes and academics
  • The Korea Capital Market Institute said in July that 32 sessions since the launch of single-stock leveraged/inverse ETFs were too few for a verdict. It attributed much of the rise in volatility from May 27 to a clash of foreign/institutional selling against individual/broker buying.10
  • At a Nov 10 symposium, Seoul National University's Institute for Financial and Economic Research argued that volatility was already elevated before the funds launched and memory-stock shocks mattered more. Retail ETF buying and high-frequency trading (HFT) liquidity provision may have absorbed much of the down-day rebalancing pressure.26
Expand · Closing-auction evidence and caveatsOne figure · Up/down days, sample sizes and statistical limits

Using LSEG one-minute data, Daishin split SK Hynix closing-auction activity into three periods: before single-stock 2× ETFs, unrestricted trading, and post-restriction.9

FIG 03-4Closing-auction asymmetry—and its limitations
SK HynixBefore launch
~05.26
Unrestricted
05.27~07.30
After buy restriction
07.31~
Up-day auction returnNot reported−0.23%t = −2.0 · Negative on 70% of up daysNot reported
Auction return on days up more than 2% intraday−0.14%−0.29%t = −2.9 · Difference vs. pre-launch t = −1.3 (not significant)Not reported
Auction share of trading on up days7.8%11.1%t = 2.09.5%
Late-session return on down days−0.27%−0.62%Difference vs. pre-launch t = −1.2 (not significant)+0.03%
Up/down days are defined by the 2:50 p.m. price relative to the prior close. Samples contain roughly 20 sessions per period. Shaded rows show differences from pre-launch that were not statistically significant.
Source: Daishin Securities (LSEG one-minute bars, Nov 10, 2025–Nov 6, 2026)9
  • Up days During unrestricted trading, shares gained another 0.18% in the last 30 minutes but fell 0.23% in the auction (t = −2.0); the auction's volume share was higher than before launch. Daishin suggests sellers absorbed the rebalancing purchases.
  • Down days Late-session losses were deeper than pre-launch, followed by a further 1.07% drop by the next open. That pattern disappeared after buying restrictions.
CaveatPrices also slipped in the auction before ETF launch on days up more than 2% intraday (Fig. 03-4). Period differences were not significant. Larger estimated rebalancing needs did not predict bigger auction declines. Each period had about 20 sessions, and the seller identity is unknown. Samsung, index leveraged ETFs and KOSPI 200 futures showed no clear period effects.9

The Korea Capital Market Institute viewed brokers' purchases during the flow standoff as largely ETF creations. Trading volume alone does not prove amplification; it recommended monitoring net flows, assets and intraday ETF premiums/discounts.10 The Seoul National University presentation favored policing spoof orders over restricting trades.26

Margin loans and contracts for difference

Retail investors use two main forms of leverage: borrowing against shares through margin loans, or taking price exposure through contracts for difference (CFD). Both can trigger forced selling when collateral falls short.

Retail leverageAt peakLater reading
Margin-loan balance₩38.6tn (June 24)₩25.8tn (Nov 12, −33%)
Long CFD exposure in both stocks at 12 brokers₩709.1bn (end-June)₩388.2bn (end-July, −45%)

Sources: Yuanta Securities27; reports citing KOFIA28; reports citing documents submitted to the National Assembly29

  • Forced-selling mechanism When collateral is insufficient, brokers liquidate positions on a preset schedule. That is forced liquidation. If many accounts are sold on a plunge day, losses may deepen. In November the government temporarily waived maintenance margin requirements (Section 07).
  • Who trades CFDs? The Financial Services Commission reported that 96.5% of actual CFD investors were individuals in 2023.30 Total outstanding balances, including margin, reached a record ₩3.4177tn on April 23.31
Expand · Faster retail tradingDay trading, automated orders and revised cash-deposit requirements
  • Intraday churn Yuanta noted sharp simultaneous buying and selling by retail and foreign investors in individual stocks, alongside more same-day repayment of margin-financed purchases.21
  • Automated orders Trading through broker Open API interfaces also rose. At one broker, average daily Open API turnover jumped from ₩11bn in 2023 and ₩30.2bn in 2025 to ₩630bn in Aug 2026, Yuanta estimated from company disclosures.21
  • After the deposit-rule change From July 31, new purchases of single-stock leveraged ETFs required a minimum cash deposit of ₩30mn—a minimum account deposit.32 Section 06 looks at subsequent activity. Concurrent index swings and changes in foreign flows make it hard to isolate the rule's impact.
The mechanics are clear and potential volumes calculable. Their actual price impact remains disputed and statistically weak.
04LAYER 2 · OFFSHORE

Offshore leverage:
Losses can force broad deleveraging.

Some hedge funds use investment-bank derivatives for leveraged Korean equity exposure instead of owning stocks outright. This is the least observable of our three layers.

Equity exposure through contracts

  • The contract A hedge fund enters a total-return swap (TRS) with a prime broker. It receives share-price gains and pays losses without owning the underlying stocks. A relatively small margin deposit can support several times that exposure.
  • Whose name appears in the data? The prime broker may buy shares to hedge its exposure, although futures and other instruments are possible. Korean trading records then show the prime broker—not the hedge fund—as the foreign investor.21
  • Unknown share Not all offshore hedge-fund positions use TRS, and their share of Korean investment is not public.
FIG 04-1SCHEMATICIllustrative TRS chain: economic exposure via contract, reported trading under 'foreign investors'
Offshore hedge fund~20% margin supports 5× exposureTRS contract · Share total returnsPrime brokerBank · often London-basedOften hedges by buying sharesKorean exchangeCash-equity tradesRecorded asFOREIGNU.K. accounts, etc.IF PRICES FALL① Hedge fund faces a margin call② Without more collateral, TRS exposure is cut③ Prime broker unwinds its share hedge④ Korean data record foreign net selling
Sources: Yuanta Securities21; Daishin Securities9

RECAP · NO.32₩214tn sold by foreigners. Who was behind it?

Foreign investors sold ₩214tn of Korean listed equities in Jan–Sep (FSS, settlement basis). Accounts classified as 'U.K.' accounted for 46%. Yet IMF-reported Korean equity holdings of U.K. residents were only 0.53× the size of those accounts. Daishin interpreted the gap as possible hedge-fund trading routed through London prime brokers.9 The series have different reporting entities and coverage, so the gap does not prove the money came from non-U.K. investors. Monthly flows and methodology are detailed in Investment Insight No.32.

The math of 5× leverage

In the example cited by Yuanta from the U.K. Prudential Regulation Authority, a 20% margin supports 5× share exposure. A 12% stock decline consumes 60% of that initial margin.21

FIG 04-2ILLUSTRATIVE CALCULATIONExposure of 100 backed by margin of 20: after a 12% drop
TRS notional exposure: 100Face value of contract with prime brokerInitial margin: 2020% of exposure = 5× leverageShares −12% → contract loss of 12Margin left: 8 · Loss: 12Margin 20 → 8 (−60%)Exposure 88; margin 8 (cover: ~9%)Add collateral or reduce exposure
Source: Yuanta Securities (citing the U.K. PRA example)21

Forced selling can spread across a portfolio

TRS and CFD risks are often managed at the account or position level. A margin shortfall in one stock can force sales of other holdings. Yuanta says this can transmit a stock-specific shock to unrelated shares.21

FIG 04-3SCHEMATICA loss in one stock may trigger sales in others
① One stock plungesMemory stock down >10% in a day② Margin shortfallCalculated across account or position③ Other holdings soldIncluding non-chip shares④ More stocks declineShock spreads through the market⑤ Index falls furtherMore accounts face margin callsNextforcedsale
Sources: Diagram based on Yuanta Securities21and Daishin Securities9explanations

Circumstantial evidence, not proof

  • Other stocks fell too Across the ten heaviest foreign-selling sessions from June 22 to Nov 6, almost all net selling targeted semiconductors (just 1.3% outside the sector). Yet 62% of non-chip stocks among the top 300 by market value declined. Daishin suggests margin calls on leveraged investors may have spread selling beyond chips.9
  • Oct 12–13 The KOSPI lost 3.7% and 5.5%. A Chinese DRAM product announcement was the reported catalyst, but power equipment, nuclear and robotics stocks also plunged. Yuanta judged it likelythat forced CFD/TRS liquidation contributed, without establishing causality.21
  • London turnover Many prime brokers operate through London entities. Trading value attributed to U.K. foreign accounts peaked in June, at more than 60% of all foreign turnover.21 Heavy activity does not prove that forced deleveraging drove the selloff.
Expand · London-linked trading and margin rulesOne figure · U.K.-linked turnover, HFT, dynamic margin and selloff data
FIG 04-4U.K.-linked turnover and foreign HFT commissions
U.K. foreign accounts: KOSPI/KOSDAQ value traded (buys + sells)
Jun61.0% of foreign turnover
$555.4bn
Jul
$422.6bn
Aug60.6% of foreign turnover
$282.0bn
HFT commissions from foreign clients at four brokers
H2 2025 monthly avg.
₩3.2bn
Jun 2026
₩15.1bn
Jul
~₩11bn
Aug (through 24th)
₩6.4bn
June U.K.-linked turnover: $261.6bn bought and $293.8bn sold. HFT commissions combine four of five brokers, excluding the largest.
Source: Yuanta Securities (citing FSS and National Assembly filings)21
  • Algorithmic orders Prime brokers offer TRS clients DMA and algorithmic execution. Five brokers earned ₩179.3bn in HFT commissions between Jan 2 and Aug 24; foreign-client revenue peaked in June.21 The counterargument—that HFT supplies liquidity—is covered in Section 03.
  • Flows shifting into Asia Yuanta suggests global quant trading may have migrated to Korea, Japan and Taiwan as China tightened DMA and HFT rules after a February 2024 quant-market slump.21
  • Margin requirements can jump faster Following the 2021 Archegos collapse, prime brokers began adjusting margin to underlying volatility and crowded positioning, known as dynamic margining. Yuanta expects margin calls to cluster on particular days, then selling pressure to fade quickly once positions are sharply reduced.21
  • Numbers from the selloff Yuanta cited foreign net sales of ₩2.02tn and ₩5.15tn on Oct 12 and 13, respectively.21
The size and pace of selling—and the sectors that fell together—are observable. No official statistics measure the forced liquidations on those dates.
05LAYER 3 · SYSTEMATIC

Sell as volatility rises.
The rules require it.

The third layer is rules-based trading. These funds need not borrow: their mandates can require them to cut equity exposure as volatility climbs.

  • The mechanism Volatility-targeting funds cut equity allocations as market volatility rises to keep portfolio risk near a target. A fund targeting 10% annualized volatility, for example, might halve its equity weight if volatility doubles. Unlike the first two layers, it can sell even on an up day.
  • Long-term funds can follow suit Pension and asset-allocation funds with risk budgets may also cut equities as volatility rises. Daishin suggests that short-term volatility forced longer-term funds to retreat, potentially deepening declines.9
  • Only U.S. market estimates are available Daishin attributed part of the Nov 5 U.S. selloff to mechanical selling by volatility-targeting funds and leveraged ETFs.14 All figures cited were U.S. market model estimates. Barclays projected that volatility-targeting funds might sell up to $135bn in U.S. equity futures over 24–48 hours.33 Nomura estimated roughly $47bn in leveraged-ETF selling, again for U.S. markets.14 Neither figure is confirmed execution volume. The subsequent selling in Korean equities is unknown.
  • Not all funds trade at once Lookback windows and rebalancing cycles differ by strategy.
  • It works in reverse too The same rules can trigger buying when volatility subsides. Systematic flows may become supportive in calmer markets.
FIG 05-1SCHEMATICThe feedback loop in rules-based trading
Volatility risesVOL ▲Equity weights fallVol target · risk budgetSynchronized sellingDeeper declinesFEEDBACKAs volatility fallsThe same rules trigger buying
Source: Diagram based on Daishin Securities9, 14analysis
Expand · Options-market signals and 'Why Korea?'One figure · Put/call skew, Yuanta's three conditions and their limits
  • What the options market says Yuanta reads the put/call skew as evidence that the market's dominant fear changed around July and has stayed that way.34
FIG 05-2SCHEMATICYuanta's interpretation: the dominant options risk changed
May–Jun · Call-heavy'FOMO' volatilityFear of missing the rallydrove call buyingVKOSPI exceeded 90even as the KOSPI roseSince Jul · Put-heavy'Fear' volatilityHedging against lossesdrove put buyingVKOSPI spikeswhen the index falls
Illustrative interpretation, not a chart of measured values. Lines show direction only.
Source: Yuanta Securities34
  • Why Korea? Yuanta described Korea as the only major market outside the U.S. combining open access to AI/semiconductor leaders, deep trading and single-stock futures/options for aggressive short exposure.34
  • A limitation Taiwan also lists TSMC single-stock futures and options. The 'only market' claim would require a comparison of liquidity and foreign access, which this report does not provide.
The mechanism is plausible, but this layer has the least Korean data. U.S. estimates cannot establish actual selling in Korea.
06MUTE · DATA GAPS

The label says 'foreign'.
The trader behind it is hidden.

Investor data identify who bought and sold. They do not show whether a trade was leveraged or whose capital ultimately stood behind it.

FIG 06-1What each layer reveals—and conceals
LayerObservableNot observable
Base · Two-stock concentrationDaily index weightsNone
Layer 1 · Domestic leverageMargin/CFD balances (KOFIA, daily) · Leveraged-ETF NAV (KRX/managers)Open API and high-turnover trading (no regular series)
Layer 2 · Offshore leverageForeign net flows (KRX, daily) · Country flows (FSS, monthly)Leverage, beneficial owners and outstanding TRS exposure
Layer 3 · Systematic tradingBroker-cited model estimates for U.S. sellingKorean equity exposure and flows
Fast trading · HFTBroker commission revenue (National Assembly filings)Volume and counterparties (no regular series)
Sources: Financial Services Commission35; Yuanta Securities21; Daishin Securities9

Four blind spots

A headline saying 'foreigners sold ₩1tn' cannot distinguish discretionary selling from forced liquidation. Four gaps make the three layers difficult to measure.

  • Was leverage involved? An offshore hedge fund's TRS trades can properly appear as foreign transactions. But the data do not disclose the leverage or whether a prime broker bought to hedge.
  • Retail-like capital After the July 31 deposit rule, trading by retail and foreign investors in single-stock leveraged ETFs both fell sharply. Yuanta sees the timing as evidence that some foreign flows may reflect retail-like short-term strategies.21
  • Country labels Official sources do not clarify how FSS 'U.K.' classifications relate to the account holder's nationality or the ultimate beneficial owner.
  • Market depth To demonstrate amplification, researchers would need order-book depth during shocks and the price impact of comparable-sized trades. This report lacks order-level and execution data.
Expand · How CFDs appear in investor statisticsBefore and after the September 2023 reform—and a discrepancy in broker data
  • Through Aug 2023 Even when the beneficial investor was retail, CFD trades routed via Korean brokers were recorded as 'institutional'; trades via foreign brokers as 'foreign'.30
  • From Sep 1, 2023 Equity trades arising from CFDs are recorded by the exchange under the underlying investor category (retail, institutional or foreign). Total CFD balances are published daily by KOFIA.35
  • Conflicting descriptions Yuanta wrote on Oct 14 that domestic CFDs routed through overseas prime-brokerage services (PBS) appear as foreign transactions.21 That differs from the September 2023 reform. This report follows the FSC notice. How offshore-routed CFDs are now classified requires separate clarification from the KRX or KOFIA.
Only domestic leverage is measured daily with reasonable precision. Our finding is therefore limited: plausible amplification mechanisms exist, and the circumstantial evidence is consistent with them.
07LIMITER · REGULATION

Korea targeted domestic leverage.
Other markets chose different tools.

The 2026 Korean response is organized by layer and date. Overseas measures are grouped by purpose, without judging their merits.

FIG 07-1Korea's measures by layer (2026)
LayerKorean response
Base · Two-stock concentrationNone
Layer 1 · Domestic leverageNew listings paused; ₩30mn cash deposit with T+2 settlement; proposed 20-unit lot size; proposed individual caps; temporary maintenance-margin relief. Higher deposits also apply to Korean retail buyers of overseas-listed single-stock leveraged products.
Layer 2 · Offshore leverageNo Korean measure directly targeting offshore hedge-fund TRS or prime-broker positions
Layer 3 · Systematic tradingNo direct measure; sidecar review under consideration
Source: Timeline below

Korea: 2026 timeline

  1. 05.27
    Single-stock 2× ETFs listEight managers launched 16 products tied to Samsung Electronics and SK Hynix.25
  2. 07.16
    Joint regulatory packageNew listings temporarily halted; advertising banned; minimum deposits raised from ₩10mn to ₩30mn in cash; proposed trading lot raised from one to 20 units; liquidity-provider pricing deviation limit cut from 3% to 2%; three hours of mandatory education.36, 37Financial Services Commission and other agencies
  3. 07.29
    Emergency market reviewProposed retail cap on single-stock leveraged products (at most 20% of total investment); proposed legal authority to adjust permitted leverage in extreme markets. No short-selling ban.38, 39Ministry of Finance and Economy · BOK · FSC · FSS
  4. 07.31
    ₩30mn minimum cash deposit takes effectImplementation was brought forward (announced July 24). Proceeds from stock sales count toward the deposit only after settlement on T+2.32Financial Services Commission
  5. Oct
    20-unit trading lot: details deferred to NovemberReports said regulators considered allowing sub-20-unit holdings to be sold at the after-hours closing price for a transition period.40 The measure was not in force as of Oct 6,41and no effective date had been reported by Nov 18.
  6. 11.06
    Margin-loan maintenance ratio requirement waived for three monthsEffective Nov 9 for three months, aiming to prevent excessive forced sales from worsening volatility. This did not ban liquidation; it gave brokers discretion to defer it.42, 43Ministry of Finance and Economy
  7. 11.16
    Sidecar reviewAuthorities are examining whether sidecars dampen or worsen volatility, with changes under consideration (reported).44Financial Services Commission

Researchers have also proposed reforms. The Korea Capital Market Institute recommended wider use of capped-weight indices, volatility interruptions calibrated to stock liquidity, block-trading platforms for institutions, and a deeper domestic long-term investor base. These are research proposals, not adopted policy.10

Short selling was not banned

Past temporary restrictions ran from Oct 2008 to May 2009, Aug–Nov 2011, and Mar 2020–Apr 2021. A comprehensive ban ran from Nov 2023 to Mar 2025.39 There was no ban in the latest selloff. Views are divided.

The case for a ban
  • Hana Securities viewed it as a practical way to prevent further declines while single-stock leveraged positions remained large.
  • Critics argue short selling adds pressure and increases volatility.
The case for caution
  • A ban could impede MSCI developed-market inclusion. MSCI downgraded Korea's short-selling assessment after the 2023 ban.
  • Whether bans reduce volatility is also disputed.
For context, KOSPI returns during previous bans were −3.6% in 2008–09, +5.89% in 2011 and +77.7% in 2020–21. Short selling accounted for about 7% of KOSPI turnover on May 26, 5% on June 25 and 6.4% on Aug 4.39

International approaches

FIG 07-2Overseas measures, grouped by regulatory goal
Objective · JurisdictionActionDate
Leveraged products · U.S. SECRule 18f-4, adopted in 2020, effectively capped new leveraged ETFs at 2× (some legacy 3× products were grandfathered). In Dec 2025 the SEC sent comment letters to nine issuers seeking up to 5× funds, asking them to revise targets or withdraw filings.45, 462020.10 · 2025.12
Leveraged products · Hong Kong SFCPermitted variable-leverage structures that adjust daily within existing ±2× limits for leveraged/inverse funds whose management needs vary sharply. Hong Kong also lists single-stock leveraged products linked to Samsung and SK Hynix.472026.07.24
Short-selling limits / stability · Taiwan FSC and Ministry of FinanceCut the daily short-sale order cap for borrowed shares from 30% to 3% of average volume over the previous 30 sessions, and raised minimum short-sale collateral from 90% to 130% (effective Nov 9).48 Hana Securities said this was the sixth such restriction since 2008.49 The National Financial Stabilization Fund was also authorized to intervene that day; authorization does not establish actual purchases.502026.11.06
Prime-broker oversight · U.K. PRAReportedly examined whether U.K.-regulated banks were overexposed to Asian equities, especially a handful of AI-related stocks (Financial Times report; no formal statement).51Jul 2026 (reported)
Market-manipulation enforcement · India SEBIIssued an interim order alleging that Jane Street entities manipulated index options around expiries between Jan 2023 and Mar 2025. SEBI ordered a deposit of ₹48.44bn (4,843.57 crore) in alleged unlawful gains; trading resumed after deposit on July 21. The interim finding was not a final determination. No evidence connects similar conduct to Korea.52, 532025.07.03
Expand · India's SEBI order and the trading mechanism

First, a caveat: Daishin said public data cannot establish whether any specific manager pursued such strategies in Korea.9 India illustrates how derivatives can dominate cash-market exposure. It does not explain Korean losses.

On Jan 17, 2024, SEBI estimated BANKNIFTY index-options turnover at 353× the underlying stocks' cash turnover on a notional-equivalent basis. The group bought ₹43.7bn net in underlying shares and futures in the morning, while adding ₹321.1bn in bearish options positions—7.3× the cash/futures purchases. It lost ₹620mn in cash/futures but earned ₹6.62bn on options expiring that day.9

Korea's response has focused on visible domestic leverage. It has not directly targeted offshore hedge-fund leverage or systematic strategies.
END · MASTER

Is volatility easing?
Six gauges to watch.

Read from top to bottom: first check whether volatility falls, then whether domestic leverage stays subdued, and finally confirm overseas flows using lagged monthly data. The right column gives signals to watch, not forecasts.

VKOSPIExchange · Daily · Fastest-moving
Nov 6 65.2 → Nov 13 50.2 → Nov 16 52.6 · 2026 average: 59.3
Calmer signalMoves toward 2010–2025 average (18.0)
Days above 5%Index prices · Daily · Outcome measure
2026 total: 40 days · None after Nov 6 (largest: +4.39% on Nov 10)
Calmer signalNo new extreme days
Single-stock leveraged ETFsKRX / issuer disclosures · Daily
NAV ₩2.8tn(Nov 6; 16% of peak) · Estimated Oct rebalancing demand was 4.6%
Calmer signalNAV does not rebound rapidly
Margin loans / CFD balancesKOFIA · Daily, one- to two-day lag
Margin loans ₩25.8tn(Nov 12; 33% below peak)
Calmer signalNo renewed buildup during rallies
U.K.-linked foreign turnoverFSS · Monthly, lag >1 month
Aug $282.0bn · 60.6% of foreign turnover
Calmer signalBoth value and share decline
Two-stock index weightKRX · Daily
Nov 6 morning About 46% · 55.67% on June 22
Calmer signalOther sectors rise, reducing concentration
CUE SHEET · CALENDAR

Dates when mechanical flows or rule changes may matter

Date (KST)EventWhy it matters
During NovDetailed plan for 20-unit lots in single-stock leveraged ETFs (proposed)Effective date and treatment of fractional lots
Late NovFSS October foreign portfolio investment reportCountry-level flows; U.K.-linked accounts
Mon, Nov 30MSCI November rebalance effective at closeIndex-tracking flows concentrate near the close
Thu, Dec 10KOSPI 200 futures and options expiryDerivatives unwinds can cluster at the close
Mid-DecFSS November foreign portfolio investment reportU.K.-linked flows after November's plunge
Early Feb 2027Three-month margin-maintenance waiver expires (began Nov 9)Watch for clustered forced liquidations afterward
TBDOutcome of sidecar-rule reviewPotential changes to trading interruptions
KNOWN · ESTIMATE · UNKNOWN

Established, estimated, still unknown

KNOWNEstablished facts

  • 60.7% volatility and 40 days with moves above 5% (Section 01)
  • Two stocks near half of KOSPI; memory names volatile worldwide (Section 02)
  • Domestic leveraged positions have fallen sharply from peak (Section 03)
  • Korean measures focused on domestic leverage; no short-selling ban (Section 07)

ESTIMATEBroker and research estimates

  • Two stocks explained roughly one-third of the increase in KOSPI 200 volatility (KCMI)
  • ETF rebalancing amplified downside (Daishin); evidence remains inconclusive (KCMI/academics)
  • Forced CFD/TRS selling likely contributed to the Oct 12–13 plunge (Yuanta)
  • Rules-based funds and leveraged ETFs compounded the Nov 5 U.S. selloff (Daishin)

UNKNOWNStill unknown

  • The contribution of each amplification layer
  • How much foreign hedge-fund trading was leveraged
  • How order-book depth deteriorated during shocks
  • Actual systematic selling in Korean equities
OUTRO · CONCLUSION

Amplification worked both ways.
Its size remains unproven.

Korean volatility was not confined to declines. U.S. megacap earnings sparked the July 31 rally, but estimated leveraged-ETF rebalancing demand also exceeded ₩8tn that day. The Nov 18 rebound likewise dwarfed neighbouring markets' response to the same news. The shocks came from outside, but the market itself had mechanisms capable of amplifying them. How much amplification they caused is still unknown.

Further reading
Glossary
Leveraged-ETF rebalancing
Trading near the close to reset a 2× ETF's daily exposure. Buy after an up day, sell after a down day. Estimated amount = NAV × leverage × (leverage − 1) × daily return. Actual orders depend on futures, swaps and fund flows.
Prime broker
An investment-bank unit providing hedge funds with financing, stock lending, trading and settlement. Many operate through London entities.
TRS/CFD
Total-return swaps (TRS) and contracts for difference (CFDs) give price exposure without owning shares. Small margin deposits can support leveraged positions. Dealers often hedge by buying the underlying shares, but may use other instruments.
Margin calls and forced liquidation
Brokers sell collateral positions when margin runs short, often at a prescribed time. Portfolio-wide margining can spread selling from one stock to others.
Volatility-targeting fund
A rules-based strategy that adjusts equity exposure to keep portfolio volatility near a target: selling when volatility rises and buying when it falls.
Annualized volatility
Daily return standard deviation scaled to a year, usually daily standard deviation × √252.
Circuit breaker
A market-wide trading halt after an 8% or 15% decline (20 minutes), with the session ended after a 20% decline.
Sidecar
Pauses program-trading orders for five minutes if KOSPI 200 futures move 5% or KOSDAQ 150 futures 6% for at least one minute.
VKOSPI
An index of annualized 30-day expected volatility implied by KOSPI 200 options; often called Korea's fear gauge.
Closing call auction
Orders are collected from 3:20 to 3:30 p.m. and matched at one price, which sets the session close.
Correlation coefficient
A measure of how closely two return series move together: +1 means perfectly aligned, 0 means uncorrelated, −1 means perfectly opposite.
t-statistic
A measure used to assess whether a sample mean or difference could be due to chance. An absolute t near 2 is often a rough significance threshold.
Margin loan
Borrowing from a broker to buy shares, with the purchased stock serving as collateral.
Minimum account deposit
Cash required in the account before opening a position in certain high-risk products.
Open API
A broker-provided programming interface that allows investors to place orders automatically.
Dynamic margining
Margin requirements that vary with underlying volatility and position concentration rather than a fixed percentage; recalculated using market prices.
DMA
Direct Market Access (DMA) lets investors route orders to exchanges through a broker's infrastructure with minimal intervention.
High-frequency trading (HFT)
Rapid algorithmic order submission, modification and cancellation using dedicated trading systems.
Country classification in FSS statistics
The FSS publishes foreign portfolio flows by country. Positions classified as U.K. accounts are far larger than Korean equity holdings reported by U.K. residents, suggesting that the named account and ultimate risk-bearer can differ. Daishin interprets the classification as the location of settlement accounts, but that is its interpretation; official treatment and beneficial-owner identities remain unconfirmed.
Put/call skew
Difference between implied volatilities of put and call options. Relatively expensive puts suggest demand for downside protection; expensive calls may indicate demand to chase gains.
Liquidity-provider premium/discount limit
The gap between an ETF's market price and its NAV. Liquidity providers (LPs) are required to keep the discrepancy within a specified range.
Margin-loan collateral ratio
Collateral value divided by the loan amount. Brokers normally must maintain at least 140%; falling below that may trigger a collateral call or forced sale.
Sources, methodology and data differences

Definitions All dates are Korea time. Index returns are versus the prior close. 'Days above 5%' means absolute close-to-close changes of at least 5%. Annualized volatility = standard deviation of daily returns × √252.

Verification scope Broker figures are reproduced from their reports and have not all been independently recalculated. We did reproduce Daishin's 60.7% volatility and 40 extreme days from Yahoo Finance closes. Catalysts on major trading days follow contemporaneous reporting.

Cross-market comparison (Fig. 01-2) Yahoo Finance daily closes from the first trading day of 2026 to Nov 6, in local currencies and exchange calendars. This is not a full-year reading.

KOSPI excluding two stocks (Fig. 02-3) This is an approximation, not an official index. Starting with reported two-stock weights at the beginning of the year (35.22%) and June 22 (55.67%), we rolled the weights forward using daily stock and index prices and removed their estimated contribution to KOSPI returns. The resulting weights roughly match reports for July 15 (~52%) and Nov 6 (46–48%). Our estimated June 23 (−7.0%) and July 28 (−7.7%) ex-two-stock returns resemble Daishin's average declines for non-chip shares (−7.0%, −7.6%), but different methods prevent treating this as validation. Depending on starting weights, volatility ranges from 43.5% to 43.9% and the number of 5% days from 13 to 15. Preferred shares, new listings and capital raises were not fully adjusted for; actual error may exceed these ranges. Full-index 2008 KOSPI figures are not directly comparable.

Discrepancies between sources

  • Nvidia consensus Reuters quoted LSEG consensus of $109.1bn in Q3 revenue and $125.3bn for Q4. Other media/compilers put Q3 at $108.7–109.3bn. After-hours moves were reported as 4–5%, depending on the timestamp.
  • Nov 18 KOSPI flows Reports placed foreign net sales at ₩327.6–333.9bn and institutional net buying at ₩504.7–549.2bn, depending on outlet and time. The text rounds these figures.
  • CFD investor classification Yuanta (Oct 14) said domestic CFDs routed through overseas PBS appear as foreign trades. The FSC said that, from Sep 1, 2023, CFD trading would reflect the underlying investor's category. We follow the FSC statement.
  • Single-stock leveraged products Daishin's NAV series falls from ₩17.6tn on June 25 to ₩2.8tn on Nov 6 (ETFs; Hong Kong excluded). Saturday Economy reported market capitalization for 16 products falling from ~₩16.5tn on June 25 to ₩6–7tn in early October. Coverage and dates differ.
  • Margin loans Daishin's weekly report put balances 25.9% below peak on Nov 6; Yuanta measured ₩26.5tn on Nov 10 (−31.3%); news citing KOFIA reported ₩25.7728tn on Nov 12. The dates differ.
  • Peak CFD balance SBS Biz called ₩3.4662tn on Dec 30, 2025 an all-time high; Financial News called ₩3.4177tn (including margin) on April 23, 2026 a record. Treatment of deposits may explain the difference.31, 54
  • June 23 foreign net selling Korea Economic Daily cited roughly ₩6tn in KOSPI sales, versus ₩3.7tn in Daishin's table. Market coverage or settlement-versus-trade accounting may differ, so no amount appears in the body.
  • Two-stock volatility Daishin reported Samsung 82.4% and SK Hynix 97.6%; Yahoo Finance closes imply 83.4% and 98.4%. TSMC, Micron, Kioxia and SanDisk are based on Yahoo Finance closes in local currencies from the first 2026 session through Nov 6.
  • VKOSPI Daishin reported VKOSPI at 65.2 on Nov 6, Financial News at 65.17 at the close and SBS Biz at 64.74 intraday. The latter is not a closing value. Nov 13 (50.15) and Nov 16 (52.64) are KRED daily closes.13
  • Oct 12–13 foreign net selling Yuanta's −₩2.02tn and −₩5.15tn figures do not specify coverage or whether they are based on settlement or trade date.
  • TSMC weight StockFeel put it at 41.478% on Aug 31; Hana estimated about 43% on another date. Korean weights (June 22/Nov 6) and Taiwan's (Aug 31) are not same-date comparisons.
  • 20 ETF units versus 20 shares The FSC uses the ETF term 'units'; news reports say 'shares'. Both refer to the same rule.
  • 2008 volatility Daishin's 2008 figure is 39.2% versus 39.0% calculated from Yahoo Finance closes. Both give 60.7% for 2026 through Nov 6.

Sources (in order of first citation)

  1. Yahoo Finance, daily closes for KOSDAQ, Taiwan TAIEX, Nikkei 225, Philadelphia Semiconductor, Nasdaq 100, S&P 500, SK Hynix, TSMC, Micron, Kioxia and SanDisk; accessed Nov 18, 2026 · finance.yahoo.com
  2. Segye Ilbo, 'KOSPI Reclaims 5,100 on Nvidia Results; SK Hynix Surges More Than 6%', Nov 18, 2026 · segye.com
  3. NVIDIA (press release attached to SEC Form 8-K), 'NVIDIA Announces Financial Results for Third Quarter Fiscal 2027', Nov 17, 2026 U.S. time · sec.gov
  4. Reuters (via MarketScreener), 'Nvidia delivers strong results as investors weigh post-Huang future', Nov 17, 2026 U.S. time · marketscreener.com
  5. Supple (citing MoneyToday), 'Institutions Lift KOSPI Above 5,000; Samsung and SK Hynix Close Up 2% and 4%', Nov 18, 2026 · supple.kr
  6. Finance Today, 'KOSPI Sinks 3.59% to Near 5,100 as AI Spending Uncertainty Hits Chip Stocks', Nov 16, 2026 · fntoday.co.kr
  7. Financial News, 'KOSPI Breaks Below 5,000 for First Time in Nine Months as AI Sentiment Sours', Nov 17, 2026 · fnnews.com
  8. Yahoo Finance, KOSPI Composite Index (^KS11) daily closes; accessed Nov 18, 2026 · finance.yahoo.com
  9. Daishin Securities, 'Foreign Selling, Volatility Conditions and Scenarios for Renewed Inflows (Quant)', Nov 10, 2026
  10. Korea Capital Market Institute, 'Drivers of Rising Stock Market Volatility (Issue Report 26-12)', July 27, 2026 · kcmi.re.kr
  11. Digital Daily, 'How Often Have Sidecars and Circuit Breakers Hit Korea This Year?', Nov 14, 2026 · ddaily.co.kr
  12. Financial News, 'KOSPI Swings 1.5 Times More Than in the Crisis; Anxiety Lingers as Fear Gauge Eases', Nov 11, 2026 · fnnews.com
  13. KRED, daily closes for VKOSPI 200; accessed Nov 18, 2026 · kred.dev
  14. Daishin Securities, 'Earnings Rose, Prices Fell: An Undershooting Phase (Weekly Market Outlook)', Nov 9, 2026
  15. Yahoo Finance, Samsung Electronics (005930.KS) daily closes; accessed Nov 18, 2026 · finance.yahoo.com
  16. MoneyToday, 'Why Did Only Korea Crash? New York −1% vs. KOSPI −12%: Three Reasons', Mar 5, 2026 · mt.co.kr
  17. Newsis, 'Middle East Escalation Shakes Korean Retail Investors', Mar 4, 2026 · newsis.com
  18. Korea Economic Daily, 'KOSPI Plunges 10% as Bad News Piles Up', June 23, 2026 · hankyung.com
  19. Seoul Shinmun, 'KOSPI Ends Down 10.84% at 6,023.66 as Sidecar and Circuit Breaker Trigger', July 28, 2026 · seoul.co.kr
  20. MoneyToday, 'Foreigners Buy ₩7tn in Two Hours as KOSPI Rockets 17%', July 31, 2026 · mt.co.kr
  21. Yuanta Securities, 'HFT in Asian TRS, Domestic CFDs, DMA and Open API Scalping (Quant)', Oct 14, 2026
  22. EToday, 'Samsung and SK Hynix Reach 56% of KOSPI as Chip Supercycle Drives the Index', June 23, 2026 · v.daum.net
  23. SBS Biz, 'Samsung and SK Hynix Combined Market Value Drops Below ₩2,000tn; Record-Fast Circuit Breaker', Nov 6, 2026 · biz.sbs.co.kr
  24. StockFeel 股感, 「權值股是什麼?權值股有哪些?權值股漲一元大盤漲幾元?」, 2026.09.18 · stockfeel.com.tw
  25. MoneyToday, 'Samsung and SK Hynix Leveraged ETFs to List May 27', May 23, 2026 · mt.co.kr
  26. Supple (citing SBS Biz), 'Academic Symposium Links KOSPI Volatility to Korea's Chip-Heavy Market', Nov 10, 2026 · supple.kr
  27. Yuanta Securities, 'KOSPI After the Plunge: Declining Margin Debt (Equity Weekly)', Nov 13, 2026
  28. Jiksseol, 'Margin Debt Falls ₩8tn in a Month; Forced Sales and Buying Power Both Shrink' (citing KOFIA), Nov 17, 2026 · ziksir.com
  29. Asia Today, 'Samsung and SK Hynix CFD Positions Avoid Forced-Liquidation Fears', Oct 15, 2026 · asiatoday.co.kr
  30. Financial Services Commission, 'Major Changes to CFD Regulation' (press release), May 30, 2023 · fsc.go.kr
  31. Financial News, 'Brokers Tighten Access Amid Surge in High-Risk Leveraged Bets', Apr 27, 2026 · fnnews.com
  32. Financial Services Commission, 'Earlier Implementation of Higher Deposits for Single-Stock Leveraged ETFs and ETNs' (press release), July 24, 2026 · fsc.go.kr
  33. TokenPost, 「Barclays Model Projects $135 Billion in Potential U.S. Futures Sales」, 2026.11.08 · tokenpost.com
  34. Yuanta Securities, 'Emergency Strategy: Hold Rather Than Panic-Sell; Buy Rather Than Wait (House View)', Nov 9, 2026
  35. Financial Services Commission, 'CFD Transparency and Investor Protection Measures Take Effect September 1' (guidance), Aug 31, 2023 · fsc.go.kr
  36. Financial Services Commission, 'Joint Measures for Single-Stock Leveraged ETFs and ETNs' (press release), July 16, 2026 · fsc.go.kr
  37. Newsis, 'Samsung and SK Hynix Leveraged Funds Face 20-Unit Lots and ₩30mn Deposits', July 16, 2026 · newsis.com
  38. MTN, 'Government Proposes Individual Limits on Single-Stock Leverage After Market Plunge', July 29, 2026 · news.mtn.co.kr
  39. BizWatch, 'Why Korea Has Not Reinstated a Short-Selling Ban Despite Market Turmoil', Aug 5, 2026 · news.bizwatch.co.kr
  40. Dailian, 'Single-Stock Leveraged ETFs to Trade in Lots of 20; After-Hours Sales of Smaller Holdings Considered', Oct 6, 2026 · dailian.co.kr
  41. Saturday Economy, 'Single-Stock Leveraged Trading Drops Before 20-Unit Rule Takes Effect', Oct 6, 2026 · sateconomy.co.kr
  42. Ministry of Finance and Economy (Korea Policy Briefing), 'Three-Month Waiver of Brokerage Margin-Maintenance Requirements', Nov 6, 2026 · korea.kr
  43. Infostock Daily, 'Korea Waives Margin-Maintenance Rules for Three Months to Limit Forced Sales', Nov 6, 2026 · infostockdaily.co.kr
  44. Supple (citing MoneyToday), 'Sidecars Triggered 96 Times This Year as Regulators Review Effectiveness', Nov 16, 2026 · supple.kr
  45. U.S. SEC, 'Use of Derivatives by Registered Investment Companies and Business Development Companies' (Rule 18f-4 adoption), Oct 28, 2020 · sec.gov
  46. etf.com, 「SEC Says No to 5x ETFs」, 2025.12.03 · etf.com
  47. The Standard, 「SFC adjusts regulatory framework for leveraged and inverse products to ensure orderly market trading」, 2026.07.24 · thestandard.com.hk
  48. The Edge Malaysia (Bloomberg), 'Taiwan imposes short-selling limit after stock market rout', Nov 6, 2026 · theedgemalaysia.com
  49. Hana Securities, 'Taiwan's Market Rebound: A Turning Point?' (China Thematic Strategy), Nov 10, 2026
  50. Focus Taiwan (CNA), 'Stabilization fund ready for market intervention after Taiex's plunge', Nov 6, 2026 · focustaiwan.tw
  51. Hedgeweek, 'BoE reviews prime brokers exposure to AI-driven Asian equity trades' (citing Financial Times), July 30, 2026 · hedgeweek.com
  52. SEBI, 'Interim Order in the Matter of Index Manipulation by Jane Street Group', July 3, 2025 · sebi.gov.in
  53. The Week(PTI), 「Sebi lifts trading ban on Jane Street after Rs 4,843 cr deposit in escrow account」, 2025.07.21 · theweek.in
  54. SBS Biz, 'Risky Leveraged Investments Grow to Record High', Jan 14, 2026 · biz.sbs.co.kr

This report is informational, not a recommendation to buy or sell any security or product. Estimates and opinions attributed to brokers are their own. Regulatory measures are described, not endorsed.