🇰🇷 한국어 원문 · Originally written in Korean on Oct 26, 2026. Slides have been re-labeled in English.
Last week, I published a two-part backtest series on frequent “buy-the-dip, sell-the-rip” trading on the KOSDAQ 150 (the flagship index of 150 leading stocks on KOSDAQ, South Korea’s tech-heavy secondary stock exchange).
Those posts were based on theoretical backtest numbers. So I decided to examine how my actual KOSDAQ trading accounts performed in real life, and how closely my live fills matched the backtest results.
NH Investment & Securities (the South Korean brokerage I use for this volatility harvesting strategy) provides an API (Application Programming Interface, a programmatic channel to retrieve account data). Using it, I pulled and analyzed every single trade execution, realized profit/loss, and balance ledger across 23 sub-accounts from my first trade on November 22, 2024 through October 23, 2026.
I do not disclose raw KRW amounts here. Instead, all figures are indexed by setting my initial capital at the start of trading in November 2024 to 100.
First, a Few Key Concepts
Volatility Harvesting (Ladder Trading)
This is a strategy of buying a fluctuating asset in fixed intervals as it drops, and selling each tranche after a small rebound. I opened multiple sub-accounts and bought one account’s worth of KODEX KOSDAQ150—South Korea’s main ETF (Exchange-Traded Fund, an index fund traded on an exchange like a stock) tracking the KOSDAQ 150 index—each time the price fell by a set step, then sold that specific account once it rebounded by a set target. Because it steps down a ladder to buy and climbs back up to sell, I call it “ladder trading.”
Inverse ETF
An ETF that rises when the underlying index falls. In this post, “Inverse” refers to KODEX KOSDAQ150 Futures Inverse, which is designed to gain roughly 1% on a day when the KOSDAQ 150 falls 1%.
How I Indexed Capital to 100
I set the maximum capital ever tied up in this strategy at any single point in time as my base principal of 100. Any capital sitting idle at other times is treated as zero-interest cash. Because idle cash is still counted in the denominator as deployed capital, this is a conservative way to measure returns.
Maximum Drawdown (MDD)
The largest percentage drop from a peak to a subsequent trough. It measures the deepest pain I actually had to endure during the period.
Sharpe Ratio
A measure of how much return a strategy earned per unit of volatility (price swings). Here it is calculated as annualized return divided by annualized volatility; a reading above 1.0 is generally considered strong.
Payoff Ratio (Win/Loss Ratio)
Average profit per winning trade divided by average loss per losing trade. Even with an extremely high win rate, a strategy will lose money if its payoff ratio is too low.
Conclusion: Ladder Made 114, Inverse Dropped to 71, Combined Was 98
Tracking the portfolio from my first trading day in November 2024 (= 100) through October 23, 2026 shows the following:
- KOSDAQ 150 Ladder: 100 → 114 (7.0% annualized). Simply buying and holding KODEX KOSDAQ150 over the same period would have turned 100 into 109.
- Inverse: 100 → 71.
- Combined: 100 → 98. The inverse positions wiped out nearly all the gains from the long ladder.
The ladder itself did its job. It earned slightly more than buy-and-hold (114 vs. 109), while its worst drawdown (-12.5%) was just one-fourth that of buy-and-hold (-49.7%). Its return per unit of volatility (Sharpe ratio) was 0.74, more than double buy-and-hold’s 0.34.
In my Part 2 five-year backtest, the “Balanced” ladder (20 sub-accounts, spacing step sizes by ATR—Average True Range, the average daily price range) posted an annualized return of 7.0% and a Sharpe ratio of 0.52. My live ladder performance (7.0% annualized, Sharpe 0.74) matched or slightly beat the backtest. The backtest numbers were not a mirage.
The problem was the inverse ETF.
Evidence 1: When I Made Money and When I Lost It
I divided the two years into five phases based on my holdings. Each number shows how many points a leg added to (or subtracted from) the combined base principal of 100.
Phase 1 — Ramp-Up (Nov 2024 – Feb 2025)
I started the ladder by gradually adding sub-accounts. Contribution: +2.7 points.
Phase 2 — Ladder Round 1 (Mar – Jun 2025)
Over these four months, the KOSDAQ 150 went virtually nowhere (+0.5%), yet the strategy earned +4.3 points. The index dropped sharply in March and recovered between April and June. This confirmed in live trading that the ladder works best in a choppy, range-bound market that swings up and down without going anywhere.
Phase 3 — Inverse-Heavy (Jul 2025 – Apr 2026)
While the KOSDAQ 150 surged +64%, I accumulated the inverse ETF like a ladder. Every time the index rose, the inverse ETF fell—and every time it fell, I bought more. Finally, over two days on April 24 and April 27, 2026, I liquidated the entire inverse position. That leg alone cost -16.8 points during this phase, completely swamping the +4.3 points earned by the long ladder over the same period.
There was a specific reason I dumped all the inverse shares over those two days. At the time, I saw compelling opportunities in individual stocks—led by semiconductors—and judged that redeploying that tied-up capital into individual stocks would be better than sitting on underwater inverse shares waiting for a market crash. I evaluate whether that call paid off in “Evidence 2” below.
Phase 4 — Sidelines (Apr – Jul 2026)
I held no KOSDAQ ETFs at all. While the KOSDAQ 150 plunged -32%, this strategy contributed 0.0.
Phase 5 — Ladder Round 2 (Jul – Oct 2026)
I resumed running only the long ladder. The KOSDAQ 150 fell -7%, while the strategy gained +3.9 points.
Evidence 2: Where Did the Cash from Selling the Inverse Go?
On April 24 and 27, 2026, I sold all of my inverse shares and recovered the remaining cash, locking in a loss equal to 40% of my total purchase cost on those shares. Setting that recovered cash at 100, I compared three paths over the next two and a half months through July 13 (the day before I resumed buying KOSDAQ ETFs):
- The path I actually took—individual stocks (mainly semiconductors) and ETNs (Exchange-Traded Notes, index-tracking debt notes issued by brokerages): 100 → 190
- If I had kept holding the inverse ETF: 100 → 136. Because the KOSDAQ 150 fell 32% in the meantime, the inverse position would have recovered a fair chunk of its losses.
- If I had put that cash back into the KOSDAQ 150 ladder (modeled as holding the KOSDAQ 150): 100 → 67.
Looking strictly at the outcome, the call was right. The profit I made in individual stocks exceeded the loss I locked in on the inverse ETF. Measured against the combined base principal of 100 from earlier, the inverse leg subtracted -16.8 points, whereas individual stocks and ETNs added +26.1 points over those three months.
To be fair, however, there is an important caveat. I did not trade those individual stocks using only the recovered cash. I combined cash from other sub-accounts plus margin loans (money borrowed from the brokerage), deploying an average of 1.7x—and at peak 3.7x—the recovered cash. As a result, “190” overstates the return on the recovered cash alone. Measured against the average capital actually deployed, the return over those three months was about +53%. Moreover, most of those gains came from just three sell trades in SK Hynix and Samsung Electronics (South Korea’s two flagship memory-chip makers; see Evidence 4). While the decision clearly paid off this time, whether I can repeat that kind of discretionary call in the future is a separate question.
Evidence 3: The Win-Rate Illusion
On the surface, my win rates looked phenomenal:
- In the KOSDAQ 150 ladder, 594 out of 596 sell trades were profitable (99.7%).
- Even in the inverse ETF, 185 out of 206 sell trades were profitable (89.8%).
Yet in the inverse ETF, the average loss on a losing trade was 42 times the average profit on a winning trade. That is a payoff ratio of 0.02. I harvested tiny gains frequently while letting losses pile up until I realized them all at once. Of the 20 largest realized losses across the entire two-year period, 16 came from the April 24 and 27 inverse liquidation, and all 13 sub-accounts that primarily traded the inverse ETF ended in net loss.
If I plug those numbers into the Kelly Criterion—proposed by mathematician John Kelly in his 1956 paper “A New Interpretation of Information Rate” to calculate optimal bet sizing as Win Rate − (1 − Win Rate) ÷ Payoff Ratio—the inverse strategy yields -3.4. A negative Kelly number means: “Do not bet a single cent on this game.”
It is the textbook asymmetric trap described in Nassim Taleb’s Fooled by Randomness: a structure that collects small, frequent pennies while exposed to a rare, catastrophic loss looks steady for a long time—until a single blowup wipes everything out.
Even the ladder’s 99.7% win rate comes with a catch: underwater tranches are simply held. Because the ladder relies on the belief that a broad market index will eventually recover and trend upward, it refuses to sell losing accounts and waits—which naturally inflates the realized win rate. The KOSDAQ 150 shares I still hold today sit 8.4% below my average purchase price. A ladder’s true performance must be judged not by its win rate, but by “realized gains on closed trades + unrealized losses on open tranches.” The 114 figure reported above already includes those unrealized losses.
Evidence 4: The Entire Account Including Individual Stocks
Starting in April 2026, I also traded individual semiconductor stocks inside the same NH brokerage accounts. So I calculated the total account performance separately.
Here I used Time-Weighted Return (TWR). TWR strips out the distortion of cash deposits and withdrawals by chaining each day’s return (daily P&L ÷ prior-day net equity). Because I withdrew a large sum between May and July 2026 and deposited cash back after July, looking at raw account balances would distort the return. Since deposit/withdrawal logs prior to September 2025 were unavailable, I indexed September 11, 2025 at 100.
- Total NH Account: 100 → 148 (MDD -21.6%, Sharpe 1.03)
- KODEX 200 (an ETF tracking the 200 largest blue-chip stocks on South Korea’s main KOSPI market) buy-and-hold: 100 → 195
- KODEX KOSDAQ150 buy-and-hold: 100 → 90
A 48% gain looks respectable at first glance, but simply holding the benchmark KOSPI 200 ETF over the exact same window returned 95%. Furthermore, 90% of my realized gains in individual stocks came from just three sell trades (two in SK Hynix, one in Samsung Electronics). Rather than a repeatable “strategy,” that result reflects a handful of concentrated bets paying off. The sample size is far too small to tell whether it was skill or luck. (One individual stock—PSK Holdings—was traded using the mechanical ladder rule, which I examine separately in Evidence 5.)
Evidence 5: Testing the Ladder on a Market Leader
Among the individual stocks in my account, my trading in PSK Holdings (a South Korean semiconductor equipment maker) was different in nature: I transplanted the exact KOSDAQ 150 ladder directly onto a single stock. My hypothesis was that a high-beta market leader with wide intraday swings might harvest volatility even more effectively than an index ETF.
The mechanics were identical. I used 7 sub-accounts as ladder rungs, scaled into price bands across each rung, and sold whenever a rung rebounded by roughly 0.5% on average. Over a two-week window from October 6 to October 19, 2026, I placed 160 orders and executed 79 sells—and all 79 sells were profitable.
Indexing the peak capital tied up in this specific trade at 100 yields the following:
- PSK Holdings Ladder: 100 → 105 (as of Oct 23, MDD -15.6%)
- Buying and holding PSK Holdings over the same period: 100 → 86
While the stock price fell 14% over those two weeks, the ladder made 5%. However, not all shares were sold: I still hold an open position sitting 15% below my purchase cost. Just like the KOSDAQ 150 ladder, the 79-for-79 win record happened because losing rungs were kept open. And two weeks is far too short a sample to prove skill.
Including this trade in the overall volatility harvesting bucket (against the same base principal of 100 as the KOSDAQ 150 analysis) changes the totals slightly:
- Total Volatility Harvesting (KOSDAQ 150 Ladder + Inverse + PSK Holdings Ladder): 98 → 99
- Long ladders only (excluding Inverse): 114 → 114.4
Because the trading window and capital allocation were small relative to my total principal, it barely moved the needle on the aggregate numbers. Still, it confirmed that the ladder mechanics—making money from swings even while the underlying asset declines—can function the same way on an individual stock.
The remaining homework is which leading stocks to pick, how to screen them, and when to stop. Unlike a broad index ETF, an individual stock can suffer permanent impairment from an industry downturn or even delisting, never to recover. Just as the inverse ETF turned out to be an asset that does not come back no matter how long you wait, a former market leader that loses its leadership can leave a ladder permanently trapped underwater. I plan to study objective rules for selecting leaders (sector momentum, trading value, relative strength) and hard stop conditions, and share those in a separate post.
Two Structural Weaknesses of This Strategy
Looking back on nearly two years of live execution, I felt two structural weaknesses in my bones beyond the raw numbers.
1. Tied-Up Capital Means Missed Opportunities Elsewhere
A ladder fills sub-accounts one by one as prices fall and waits until they rebound. During a prolonged downturn, a large chunk of capital stays locked up for months. During my inverse-heavy phase (July 2025 to April 2026), an average of 36—and up to 49—out of my 100 base principal was tied up in this strategy. Over that exact same stretch, KODEX 200 surged +141%. The tied-up capital missed that entire rally. Opportunity cost never shows up on a strategy’s standalone scorecard. Ultimately, freeing up that trapped capital was the main reason I liquidated the inverse position all at once in April.
2. An Inverse ETF May Never Come Back No Matter How Long You Wait
Even if a broad index ETF like KODEX KOSDAQ150 drops sharply, it has a reasonable chance of recovering when the market cycle turns. That is what makes scaling in and waiting viable. An inverse ETF is fundamentally different. Equity markets tend to drift upward over the long run, and an inverse ETF moves in the opposite direction. Worse, because it resets daily to deliver -1x the index’s daily return, repeated up-and-down swings slowly erode its value even if the index ends flat—a phenomenon known as volatility drag. In fact, since KODEX KOSDAQ150 Futures Inverse was listed in August 2016, KODEX KOSDAQ150 has risen +33%, yet the inverse ETF has fallen not by the ~25% simple math would suggest, but by -75%. The core premise of ladder trading—”if I get stuck, I can just wait for it to come back”—simply does not hold for inverse ETFs.
Wrapping Up
Here is the final scorecard from nearly two years of running volatility harvesting with real money:
- The long KOSDAQ 150 ladder alone turned 100 into 114. It beat buy-and-hold with one-fourth the maximum drawdown, closely matching my backtest expectations.
- However, scaling into an inverse ETF during a bull market turned 100 into 71, dragging the combined result down to 98.
- A two-week test of the ladder on an individual market leader (PSK Holdings) turned 100 into 105 (vs. 86 for buy-and-hold), lifting the combined volatility harvesting result from 98 to 99. How to systematically select and exit individual leaders remains open homework.
- My decision to cut the inverse ETF and redeploy the recovered cash into individual stocks paid off in the end, turning 100 of recovered cash into 190 over three months—though that involved extra cash, margin loans, and highly concentrated gains in just three trades.
In the end, what decided my overall return was not a flaw in the ladder rule itself, but what I did outside the rules. Buying the inverse ETF on a gut feeling that “the market has to pull back soon” ate away my ladder profits, while my discretionary decision to cut that position and rotate into semiconductors earned it back. Above all, I learned two expensive lessons: a ladder locks up capital and creates hidden opportunity cost, and an inverse ETF is never an asset you can ladder into and wait out.
Going forward, I will run this volatility harvesting strategy strictly according to the rules derived from my backtests, and report back in a few months using the exact same indexing method (Day 1 = 100).
Disclaimer: This post is a personal record of my live account performance and money-management rules, and is not a recommendation or financial advice to buy or sell any security. All figures deduct trading commissions and taxes, and exclude margin interest and dividends. Past performance does not guarantee future results.
