The 31% rise in Samsung Electronics daily range volatility and 58% rise in SK Hynix warrant a fresh look at position sizes. Both followed the listing of Korea's first single-stock leveraged ETFs, offering 2x and inverse exposure to one stock. Lin, Nguyen and Wang measure the changes carefully. They cannot establish that the ETFs caused them, and acknowledge that limit. The paper gives a desk reason to revisit volatility budgets on these names. It gives no basis for a signal.

The daily reset reaches individual stocks

Until April 2026, Korean rules required ETF indices to contain at least ten securities, with any single holding capped at 30%. The rule change let qualifying products concentrate on one stock and target up to twice its daily return. Samsung Electronics and SK Hynix were the first eligible names. Sixteen leveraged and inverse products on them listed on May 27, 2026.

The authors focus on the daily reset, which adjusts exposure to maintain the fund's return target. After an up day, a long 2x fund must add exposure; after a down day, it must cut. Those same-direction trades now reach a single stock.

Using Fn DataGuide daily data from January 2, 2025 to July 16, 2026, the authors run a separate event study for each firm. Welch t-tests compare each outcome's mean across the 40 trading days before listing with its mean across the 36 days after. Day 0 belongs to the post-listing window. The outcomes span:

The volume data combine the Korea Exchange (KRX) and Nextrade (NXT). The alternative venue accounted for about 28.8% of domestic trading value in its first year. For a descriptive comparison, the authors apply the same tests to the earlier Hong Kong 2x listings on these stocks, on May 28, 2025 and October 16, 2025. The two KODEX 2x products follow their targets closely: their slopes are 2.004 and 2.043, with an R2 of 0.997 over 35 days.

The authors state the central finding plainly: "volatility is the most consistent post-listing change across the two stocks."

Higher volatility after May 2026, without lasting repricing

Samsung's consolidated range volatility increased from 3.952% to 5.173% (t=2.46). For SK Hynix, it increased from 3.841% to 6.050% (t=4.85). The KRX-only ranges show the same pattern. Daily downside loss also increased, from 1.071% to 2.812% and from 1.094% to 3.074%.

The abnormal returns fade. SK Hynix records a 7.22% cumulative abnormal return (CAR) over days 0 to 1 (t=2.47). Its CAR then reaches -2.98% by day 5 and 11.14% by day 35, neither statistically significant. Samsung's 35-day CAR is 1.59% (t=0.19). As the authors acknowledge, a benchmark with large weights in both stocks may attenuate these estimates.

My objection starts with the t-stats supporting the volatility result. The KOSPI 200 averaged 1.175% a day before listing and -0.298% afterward. SK Hynix moved from 2.130% a day to -0.007%. A 40-day melt-up followed by a declining market, with the KOSPI 200 losing -0.298% a day, will lift measures constructed from losing days. This applies to downside return, defined as max(-r, 0), and to the 21-day downside deviation.

Range volatility presents a harder objection. A change to a higher-volatility market could explain its increase, but that explanation rests on analogy, making the objection weaker. The authors flag contemporaneous semiconductor news as a possible influence on the SK Hynix price. They also mention the July 2026 regulatory measures, solely to indicate that the setting matters. Both fall within the post-listing window. The comparison still pits a rally against its aftermath.

Did greater SK Hynix turnover buy cheaper fills?

SK Hynix turnover increased from 1.174% to 1.469% (t=3.40), while Amihud illiquidity declined from 0.403 to 0.256 (t=-2.35). Samsung changed little. Its turnover moved from 0.914% to 0.998% (t=1.34), and Amihud from 0.276 to 0.297 (t=0.47). Extra volume follows volatility; the improvement in price impact survives controls for volatility. Neither finding establishes what happened to closing-auction execution.

Across all four firm-events, the correlation between turnover and range volatility runs from 0.527 to 0.764. Controlling for volatility takes the SK Hynix post-listing turnover coefficient from 0.295 to -0.004. The authors reach the same conclusion: "the turnover response appears closely tied to volatility."

The Amihud improvement is harder to explain away. The coefficient remains -0.238 (t=-2.36) with contemporaneous volatility included, and -0.204 (t=-2.34) under the pre-event adjustment. But Amihud measures absolute daily return per won traded. It reveals nothing about spreads or depth in the closing auction, nor the slippage paid where the rebalance occurs. Zhao (2026), cited by the paper, connects volatility in this same episode to predictable closing rebalancing. Averaging a daily ratio over the whole session cannot establish whether fills near the close improved or deteriorated.

Seventy-six days, two names, one date

The sample also gives me less confidence than the t-stats suggest. Samsung's trailing 21-day return SD increases with t=8.36. Yet for roughly the first 20 post-listing days, the rolling window still includes returns from before listing. Overlapping windows also breach Welch's independence assumption.

Abnormal volatility subtracts a constant estimation-window mean from range volatility. Its differences of 1.222 and 2.210, and t-stats of 2.46 and 4.85, therefore reproduce the range result exactly. The authors' Newey-West re-estimates, using five lags, assign smaller t-stats to the same post-listing differences. For new margin, Samsung's t-stat falls from t=-2.04 to -1.52, and SK Hynix's from -1.87 to -1.14. Samsung individual participation falls from t=-2.58 to -1.81.

Timing creates a further problem: the 40-trading-day pre-window ending May 26 starts around late March and thus includes the April rule change itself.

The Hong Kong comparison makes a universal launch effect less convincing. SK Hynix range volatility increased after its Hong Kong listing, from 2.639% to 3.548% (t=3.21). Samsung's range volatility was unchanged, moving from 1.515% to 1.636% (t=0.87), while its 21-day SD declined from 2.196% to 1.859%. Individual participation in the Korean shares increased by 6.997 and 8.880 points after the offshore listings. Following the domestic listing, Samsung's declined by 3.298 points.

Securities-lending balances fell in all four events (-0.121, -0.333, -0.192, -0.573). A secular trend could also generate declines across all four windows in 2025 and 2026. The paper tests no non-listing windows that could distinguish that explanation from a listing effect.

Revisit the sizing assumptions

For a desk holding these stocks, the figures raise immediate questions about the existing book. SK Hynix post-listing range volatility was about 58% higher, at 6.050% against 3.841%; Samsung's was 31% higher. An inverse-volatility allocation calibrated on the pre-May window would carry SK Hynix at roughly 1.6 times its current-risk weight. Closing-auction cost assumptions need checking too. The daily Amihud improvement cannot establish cheaper execution at the print.

We did not run our own test; we have no Korean data covering the May 2026 event.

Minute data showing post-listing volatility concentrated after about 3:00 p.m., with that pattern absent before listing, would change my view. The authors themselves propose this test and leave it for future work. Zhao (2026), whom they cite, reports related evidence on closing rebalancing during the same episode. From this paper's daily data alone, the 58% increase describes the regime a desk is trading through. The ETF's part in it remains unproven.