The IHSG's Six-Month Return Was -34.7% for One Investor and -20.9% for Another. Here's Why.
5 min read
Loading...
5 min read
Say two people each held an IHSG-tracking fund for exactly six months in 2026. One started counting on 30 December 2025 and checked their return six months later. The other started counting two months after that, on 27 February 2026, and checked six months after their own start date.
The first investor's six-month return was -34.7%. The second investor's was -20.9%. Same index, same six-month length, a 14-point gap between them. Neither investor did anything differently. The only thing that differed was which day they happened to start counting from. That gap is exactly what the difference between rolling returns and point-to-point returns is about, and it changes how much you should trust any single "your return over the last six months" number you see.
Here is what the Jakarta Composite Index (IHSG) actually did across those months, using confirmed closing levels reported by Databoks/Katadata, ANTARA News, KSEI (Indonesia's central securities depository), Mediaindonesia, IDX Channel, CNN Indonesia and Okezone:
| Date | IHSG close |
|---|---|
| 30 Dec 2025 | 8,646.94 |
| 15 Jan 2026 | 9,075.41 |
| 27 Feb 2026 | 8,235.48 |
| 31 Mar 2026 | 7,048.22 |
| 30 Apr 2026 | 6,956.80 |
| 30 Jun 2026 | 5,643.19 |
| 28 Aug 2026 | 6,518.12 |
The IHSG actually rose about 5% in the first two weeks of 2026 before turning down hard. By 30 June it had fallen to 5,643.19, its low point for the year. It then partly recovered, closing at 6,518.12 on 28 August.
(End value − Start value) ÷ Start value
Investor one's window (30 Dec 2025 to 30 Jun 2026) ends right at the year's trough: (5,643.19 − 8,646.94) ÷ 8,646.94 = -34.7%. Investor two's window (27 Feb 2026 to 28 Aug 2026) starts after the market had already fallen for two months and ends after a partial recovery: (6,518.12 − 8,235.48) ÷ 8,235.48 = -20.9%. Both numbers are correct. Both describe the same six-month holding period. They just don't describe the same six months.
A point-to-point return, which is what both numbers above are, picks exactly one start date and one end date and measures the change between them. It is easy to calculate and it is what most performance summaries show you, but it is entirely at the mercy of where those two dates happen to land relative to a market's ups and downs. Start right before a downturn and end right at the bottom, and the number looks brutal. Start two months later and end after some recovery, and the same length of time looks far milder.
A rolling return fixes this by recalculating the same window length, six months in this example, starting from every possible day in a longer period, not just one. Instead of a single number, you get a whole series: the best six-month return achieved in the period, the worst, and everything between. That series tells you what a six-month holder could actually have experienced depending on timing, rather than handing you one snapshot and letting you assume it is representative.
The gap between -34.7% and -20.9% is not random. It comes down to where each window's endpoints sit relative to the crash. Investor one's window ends on 30 June, the exact date of the IHSG's 2026 low. Investor two's window starts on 27 February, after roughly two months of decline had already happened, so less of the fall is captured inside that window, and it ends in August after part of the recovery. A single point-to-point return has no way to tell you this. It just reports the two endpoints you gave it, whether or not either one happened to land on an extreme.
This is also why a single "how has this performed over the last six months" figure on a fund factsheet or a broker app deserves some skepticism. It is not wrong, but it is one path through a much bumpier road, and a different start date during the same underlying period could have shown you a very different number.
The fix is not to distrust every return figure, it is to look at more than one window before drawing a conclusion. Open your portfolio dashboard and drag its history chart's date range yourself, so you can see how a holding's return looks from several different starting points instead of relying on whichever fixed period a summary happened to pick. Before reacting to any single six-month, one-year or year-to-date number, check whether it changes much if you shift the start date by a month or two. If it does, you are looking at one point on a much wider range of outcomes, not the whole picture.
For a related distortion in how returns get reported, see how arithmetic and geometric mean returns can disagree even when both are calculated correctly, and how your broker's own return figure can differ from your actual XIRR. To see the IHSG's full 2026 trajectory rather than a single window of it, visit the market overview.