The Interval
Problem

Why one index, over one month, produced three irreconcilable stories, and what that implies about how often anyone should look.

On Friday, July 31, 2026, Korea's Kospi Composite rose 17.9 percent. It was the largest single-day gain in the index's history.

That same month, the index fell 22.2 percent. Also a record, in the other direction.

Both statements are true. Both describe the same twenty-two trading sessions. Neither is a technicality or a rounding artifact. The difference between them is not the data. It is the interval over which the data was measured.

That distinction sounds academic until you notice that it governs almost everything an investor believes about their own portfolio.

01 The July sequence

Some context is owed before the arithmetic. The Kospi entered July at 8,476.48, up 96.7 percent on the year, and the index had become a proxy for two companies. Samsung Electronics and SK Hynix carried more than half its market capitalization, and once single-stock leveraged products are counted, the pair regularly exceeded 70 percent of daily turnover. Korea had not so much doubled as its memory sector had, and the benchmark came along.

Leverage came next. Korean exchanges approved two-times leveraged single-stock ETFs on both names in May, and retail money moved in. Assets in leveraged products tied to Korean equities grew from under $10 billion at the start of the year to roughly $50 billion by late June, while outstanding margin loans reached a record 29.2 trillion won in early July. By late July those same products held about $17 billion. Citigroup put cumulative retail losses through the forced unwind at roughly $38.7 billion.

Then the inputs turned. The Bank of Korea raised rates for the first time since 2023, the global AI trade cooled, and on July 28 came reports that China had begun mass production of domestic DUV lithography tools. SK Hynix reported record second-quarter revenue of 79.3 trillion won the following day and still missed consensus, which was 84 trillion.

The decline itself was faster than most coverage suggested. July 2 fell 7.9 percent, and by the close on July 13 the index sat at 6,806.93, having given back 19.7 percent in nine trading sessions. What followed was less a slide than a sequence of shocks: a 10.8 percent drop on July 28, the second-worst session of the year, then 6.0 percent the next day, leaving the index at 5,593.56 on July 30.

Then July 31 gained 17.9 percent.

02 Three frames, one dataset

Look at the month and you see a collapse of 22.2 percent. Look at Friday and you see the best session the index has ever recorded. Look at the year and you see the Kospi up 53.0 percent through July 31.

All three are accurate. All three are computed from identical closing prices. An investor who checked once on June 30 and again on July 31 experienced a catastrophe. An investor who checked once in January and again on July 31 experienced an excellent seven months. They held the same asset for the same period.

03 Why the best day arrives inside the worst month

The obvious reading is that Korea is unusually volatile and therefore unusually prone to this effect. That reading is too comfortable. Korea makes the problem legible because the magnitudes are large. It does not create the problem.

The more useful observation is that the large moves cluster. The worst session of 2026 was March 4, down 12.1 percent. The next session, March 5, gained 9.6 percent. The second-worst session was July 28. The best session in the index's history came three trading days later.

Good days and bad days are not distributed like coin flips across a calendar. They arrive together, out of the same weather system, and they are frequently adjacent.

04 The arithmetic of stepping out

This is where the interval stops being an abstraction and starts costing money, and the arithmetic is worth sitting with.

Widen it and the effect compounds. Without the three best sessions the year returns 9.2 percent. Without the five best it is negative, at minus 6.9 percent, in a market that finished up 53.0 percent, and without the ten best the position is down 31.8 percent.

The arithmetic is not an argument for enduring volatility as a matter of character. It is a structural observation about when the best sessions occur. They occur inside the worst stretches, because that is when prices are dislocated enough to move that far. An investor who sold on July 30 to stop the pain was positioned to miss July 31 by exactly one session.

05 An interval you did not choose

Every investor has a measurement interval, and very few have chosen one.

For most households the interval is set by infrastructure rather than intention. A custodial statement arrives monthly, so the month becomes the unit. An application updates continuously, so the day becomes the unit, or the hour. A headline arrives when something has already happened, which makes the interval whatever the news cycle decides it should be.

None of those intervals were selected against a goal. They were inherited from whatever system happened to be delivering the information. And because the interval determines what gets experienced as a gain or a loss, it quietly shapes behavior. It decides what feels survivable and when someone picks up the phone.

Two things in that table deserve attention, and they point in opposite directions.

The first is the last column. The return over the period is identical on every cadence, because the cadence never touched the money. It only touched what the investor knew.

The second is the second column, which does not fall as the interval lengthens. It rises. The investor checking weekly never received worse news than a 10.6 percent decline. The investor checking monthly opened one statement that was down 22.2 percent, because a month is wide enough to hold the entire collapse in a single reading. Looking less often does not reduce what you feel. It concentrates it. Anyone who has opened a quarterly statement after a bad quarter already knows this.

A plan that does not specify its own measurement interval will have one assigned to it.

The interval is not only a matter of feeling, which is the part that gets written about. It is also the trigger on a set of irreversible decisions. Selling inside a drawdown converts a paper loss into a realized one, and it does so in a specific tax year, against a specific cost basis, with specific consequences for the harvesting that was scheduled for December and for the gain the household was going to recognize next year at a lower rate. The custodian that set the interval is not coordinating any of that. Whatever chose the interval is, in practice, choosing when the household transacts. For most people, nothing chose it.

06 Two honest objections

It does not mean checking less is a strategy. Ignoring a portfolio is not the same as holding one deliberately, and the difference shows up precisely when something needs attention.

It does not mean volatility is harmless. A household drawing income from a portfolio experiences a 22 percent month differently than one contributing to it, and the sequence in which returns arrive is a real risk rather than a psychological one.

What it does mean is narrower. The interval is doing more work than most people realize, and it happens to be the one variable in the exercise an investor actually controls. Nobody chooses the return or the volatility or the day the rebound lands. They choose how often they look, and what they have agreed in advance to do about what they see.