Do foreign investors move the KOSPI? 20 years of monthly data
Updated September 2026 · Data January 2006 – August 2026
"Foreigners bought, so the market went up" is the most common sentence in Korean market commentary. Is it true, and more usefully, does foreign buying tell you anything about next month? We took every month from January 2006 to August 2026 — 248 months — and compared net foreign purchases of KOSPI stocks with the index's return in the same month and the following month. The short answer: the same-month link is real and strong, the predictive link is essentially zero, and 2026 has been the great exception to the first rule.
Same month: foreigners and the index move together
From 2006 through 2025, the correlation between monthly net foreign purchases and the monthly KOSPI return was 0.53 — high for financial data. In months when foreigners were net buyers, the KOSPI rose 77% of the time, with an average gain of 3.0%. In months when they were net sellers it fell 65% of the time, with an average loss of 1.8%.
That is not surprising. Foreigners hold close to 40% of the market, trade the largest stocks, and their flows are partly a result of price moves (index funds receive inflows when markets rise). The direction of causality runs both ways. Retail investors are the mirror image: the correlation between retail net buying and the same-month return is −0.16, because Korean individuals reliably buy dips and sell rallies.
Next month: no signal
This is where the popular story falls apart. The correlation between one month's foreign flow and the following month's KOSPI return is −0.06 (2006–2025) — indistinguishable from zero. After a net-buying month the index gained an average of 0.2% in the next month; after a net-selling month it gained 1.0%. If anything, chasing last month's foreign flow was mildly counterproductive. Extending the horizon to three months does not help.
In other words, foreign flow data tells you what is happening, not what will happen. That is still useful — it identifies who is on the other side of a move and whether a rally has foreign sponsorship — but it is not a mechanical trading signal.
The biggest months
Largest net buying (₩ tn)
| Month | Flow | KOSPI same month | Next month |
|---|---|---|---|
| Feb 2024 | +7.9 | +5.8% | +3.9% |
| Sep 2025 | +7.6 | +7.5% | +19.9% |
| Sep 2013 | +7.3 | +3.7% | +1.7% |
| Jan 2023 | +6.4 | +8.4% | −0.5% |
| Jul 2025 | +6.3 | +5.7% | −1.8% |
Largest net selling (₩ tn)
| Month | Flow | KOSPI same month | Next month |
|---|---|---|---|
| Jun 2026 | −58.8 | 0.0% | −22.2% |
| May 2026 | −51.5 | +28.4% | 0.0% |
| Mar 2026 | −40.1 | −19.1% | +30.6% |
| Feb 2026 | −25.5 | +19.5% | −19.1% |
| Nov 2025 | −16.5 | −4.4% | +7.3% |
Notice the scale. The largest buying month in twenty years was ₩7.9 trillion (February 2024, when the government's "value-up" programme was announced). The five largest selling months are all from the last ten months, and the largest — June 2026 — was more than seven times the size of the largest buying month.
2026: the year the rule broke
Foreign investors sold KOSPI stocks in every month of 2026 through August, a cumulative ₩199 trillion in our figures. The KOSPI rose 56% over the same period, from 4,214 to 6,563. The correlation between foreign flow and index return, 0.53 over the prior twenty years, collapses to 0.13 when 2026 is included.
Who bought? Domestic retail investors, a net ₩145 trillion, plus domestic institutions at ₩22 trillion. The rally was concentrated in the two memory-chip makers, which foreigners sold into strength while Korean households bought. It is the largest hand-off between investor groups in the market's history, and it is the reason the foreign ownership ratio rose even as the flows pointed the other way.
Year by year
| Year | Foreign (₩ tn) | Institutions | Retail | KOSPI return |
|---|---|---|---|---|
| 2006 | −9.5 | +8.6 | −1.8 | +4.0% |
| 2007 | −24.0 | +11.2 | +4.7 | +32.3% |
| 2008 | −28.5 | +20.8 | +1.6 | −40.7% |
| 2009 | +29.2 | −24.8 | −1.3 | +49.7% |
| 2010 | +21.3 | −13.4 | −5.0 | +21.9% |
| 2011 | −9.0 | +13.6 | −2.7 | −11.0% |
| 2012 | +17.0 | +3.1 | −14.4 | +9.4% |
| 2013 | +4.3 | +3.8 | −5.4 | +0.7% |
| 2014 | +4.8 | +0.2 | −2.6 | −4.8% |
| 2015 | −3.8 | +0.1 | −0.6 | +2.4% |
| 2016 | +10.5 | −5.8 | −7.1 | +3.3% |
| 2017 | +7.3 | −2.2 | −10.3 | +21.8% |
| 2018 | −6.4 | −3.5 | +8.1 | −17.3% |
| 2019 | +1.0 | +8.9 | −12.1 | +7.7% |
| 2020 | −24.2 | −25.4 | +46.9 | +30.8% |
| 2021 | −24.9 | −39.3 | +65.5 | +3.6% |
| 2022 | −6.9 | −11.3 | +17.2 | −24.9% |
| 2023 | +11.2 | +1.0 | −13.4 | +18.7% |
| 2024 | +1.5 | +2.1 | −6.1 | −9.6% |
| 2025 | −9.4 | +17.8 | −18.9 | +75.6% |
| 2026* | −199.2 | +21.6 | +145.3 | +55.7% |
* 2026 through 2 September. Net purchases of KOSPI common and preferred shares. Institutions = all domestic institutional categories combined. The three columns do not sum to zero because "other corporations" (mainly companies buying back their own shares) are omitted.
The table shows the pattern clearly: foreigners were the marginal buyer in the good years of the 2009–2017 period and retail the marginal buyer in 2018, 2020–22 and 2026. Whether 2026 ends up looking like 2020–21 (a retail-led rally that eventually gave back a large part of its gains) is the question every Korea investor is asking.
Caveats
- Our flow totals aggregate the stocks currently listed, so delisted names are missing from earlier years. Large caps rarely delist, so the distortion is small for the KOSPI total.
- Since March 2025 our figures include trades on the NXT alternative venue as well as the KRX main board. Sources that count only KRX report smaller numbers — for June 2026, roughly ₩49 trillion versus our ₩59 trillion.
- Monthly correlations are not a trading backtest. Transaction costs, timing within the month and the choice of horizon all matter.
For the live numbers, see today's foreign flow rankings, which show the market total and the top stocks by net purchase, and the daily archive.
This article is general information, not investment advice.