Korea Investor Flow Data: Reading the KRX 3-Way Daily Breakdown

by

The single most useful thing to know about Korea’s daily investor-flow table is this: the three headline columns — foreigners, institutions, individuals — are not three comparable actors, and the middle column is the one that lies to you most often. “Institutions” bundles the National Pension Service’s slow, rule-driven rebalancing together with securities firms’ fast, largely mechanical hedging flows, so a headline institutional net-sell figure can mean two opposite things depending on which sub-category produced it. Learn to open the sub-category breakdown before reading the headline, and this data set becomes one of the most informative free tools in Asian markets.

Interest in this table tends to spike whenever flows swing sharply, and early August 2026 was a textbook case: foreign investors net sold roughly 3.3 trillion won of KOSPI-plus-KOSDAQ shares in a single completed session on 2026-08-06, kept selling through 2026-08-10, then flipped to net buying — culminating in back-to-back sessions above 2 trillion won on 2026-08-12 and 2026-08-13. That kind of stretch sends readers searching for how the data actually works. This guide is the durable answer. A companion piece on this site covers foreign flows in depth; here we complete the picture with the domestic side — institutions, their sub-categories, and individuals.

Where the data comes from and how it is published

The Korea Exchange (KRX) publishes, for every trading session, net purchase values broken down by investor type, for KOSPI (the main board), KOSDAQ (the tech-and-growth board), and derivatives separately. Two practical access points matter:

  • KRX data portal (data.krx.co.kr) — the official source, with full sub-category detail and downloadable history. Its on-screen tables commonly label trading values in millions of won. Korean-language flow headlines, by contrast, quote figures in units of 100 million won (romanized eok won; 10,000 eok won = 1 trillion won) — so divide a millions-of-won figure by 100 to reach eok won. This two-unit landscape is the first thing to internalize, because a missed conversion between the portal’s labeled unit and the headline unit is off by a factor of 100.
  • Naver Finance (finance.naver.com) — a free, fast daily investor-trend page showing roughly ten sessions of foreign, institutional, and individual net flows at a glance, likewise commonly labeled in millions of won on screen. Convenient for the headline three-way split, but you must go to KRX for the institutional sub-categories.

One timing caveat: intraday and same-day figures are provisional, compiled as the session’s data is assembled. When comparing days, use completed sessions rather than the current day’s provisional row.

The three cohorts, and what each one’s flow usually signals

  • Foreigners (romanized oegugin) — global institutional money, benchmark-driven and currency-sensitive. Persistent foreign buying or selling is the cohort most correlated with medium-term index direction, which is why it dominates headlines. (Covered fully in our separate foreign-flows guide.)
  • Institutions (romanized gigwan) — the aggregate of seven or so domestic sub-categories, discussed below. The aggregate is close to meaningless without the split.
  • Individuals (romanized gaein) — Korean retail investors. Structurally, retail in Korea acts as the contrarian liquidity provider: buying into sharp declines and selling into rebounds. Retail net buying during a crash is therefore normal market plumbing, not a bullish signal by itself.

Inside “institutions”: the sub-categories that actually matter

On the KRX portal, the institutional column decomposes into sub-categories. The three you should check every time:

Sub-category (romanized) Who it is How to read its flow
Pension funds (yeongigeum) Dominated by the National Pension Service (NPS), plus other public pensions and funds Slow, persistent, rule-driven. Multi-week streaks reflect asset-allocation rebalancing, not a market view. The most “informative” institutional line for medium-term positioning.
Financial investment (geumyung tuja) Securities firms’ proprietary books Largely mechanical: ETF creation/redemption, market-making, index-arbitrage hedging. Big one-day swings here often carry no directional opinion at all.
Investment trusts (tusin) Public mutual funds and asset managers Reflects retail fund subscriptions and redemptions with a lag — effectively indirect retail flow wearing an institutional badge.

The remaining lines — private funds, insurance companies, banks, and other financial institutions — are smaller and episodic. The core discipline: never quote the institutional aggregate without checking whether pension money or proprietary desks produced it. A 500 billion won institutional net sell driven by the financial-investment line during an ETF-heavy session is noise; the same figure driven by pension funds over ten consecutive sessions is a genuine allocation shift.

Why National Pension Service flows are mechanical, and why that is useful

The NPS is one of the world’s largest pension pools, managing assets on the order of a thousand trillion won, with a domestic-equity target that has historically sat in the mid-teens as a share of the total portfolio. It operates under a published annual asset-allocation plan with tolerance bands around each target. The consequence is predictable behavior: when Korean equities rally hard, the domestic-equity weight drifts above target and the fund becomes a systematic seller; when the market crashes, the weight falls below target and the fund becomes a systematic buyer. Pension-fund buying near market lows is therefore not a “smart money bottom call” — it is arithmetic. The useful signal is deviation from that pattern: pension selling into weakness, or buying into strength, suggests an allocation-plan change rather than rebalancing, and that is worth investigating.

Reading individuals: pair the flow with the two liquidity gauges

Retail flow only makes sense alongside two KOFIA-published balance-sheet numbers:

  • Investor deposits (romanized tuja yetakgeum) — idle cash sitting in brokerage accounts, i.e., retail dry powder. Roughly 99.98 trillion won as of 2026-08-12.
  • Margin lending balance (romanized sinyong yungja) — outstanding broker credit used to buy shares, i.e., retail leverage. Roughly 30.4 trillion won as of the same date.

The interaction is what matters. Heavy retail net buying funded while deposits fall and margin balances rise means the cohort is deploying leverage into weakness — historically the fragile configuration, because margin calls can force selling if the decline extends. Retail buying while deposits remain high is far more sustainable. As a rough historical frame, deposits near or above the 100 trillion won mark have accompanied strong retail engagement, while margin balances in the high-20s to low-30s of trillions represent a substantial leverage load whose unwind can amplify any downturn.

Worked example: reading one real number, step by step

Here is an actual figure. On 2026-08-13, foreign investors’ combined KOSPI-plus-KOSDAQ net purchase was +21,102 eok won. How to read it:

  1. Convert the unit. 21,102 eok won divided by 10,000 = roughly 2.11 trillion won of net buying — around USD 1.5 billion at the won’s level of roughly 1,416 per dollar at the 2026-08-13 close.
  2. Size it. In the nine completed sessions around it (listed below), four prints exceed 2 trillion won in absolute size, and this is one of them. Rather than leaning on any absolute rule of thumb, judge a print against its own recent window — and this one sits at the top of that range.
  3. Check the run, not the day. The completed-session series around it: -28,220 (08-03), -3,716 (08-04), +14,464 (08-05), -32,893 (08-06), -8,651 (08-07), -14,909 (08-10), +535 (08-11), +28,354 (08-12), +21,102 (08-13), all in eok won. Read as a sequence, this is a violent regime flip: sustained heavy selling through 08-06 to 08-10, then three straight positive sessions culminating in back-to-back prints above 2 trillion won. One green day inside a red streak means little; three in a row, with the last two both above 2 trillion won, is a trend change worth respecting.
  4. Identify the mirror. Every net buy has a net seller. Open the same table’s other columns: if individuals were the sellers, retail was taking profits into the rebound (typical); if pension funds were selling, mechanical rebalancing was absorbing the inflow. The mirror tells you whether the move had domestic conviction behind it or was purely a foreign-driven repricing.
  5. Cross-check the board split. The figure above combines KOSPI and KOSDAQ. Foreign buying concentrated in KOSPI large caps signals index-level allocation; KOSDAQ-tilted buying signals a sector or thematic bet. KRX lets you view each board separately.

Common mistakes and how to avoid them

  • Reading the institutional aggregate as one actor. Always open the sub-categories. Financial-investment flows are mostly hedging; pension flows are mostly rebalancing; only deviations from those defaults are signals.
  • Extrapolating one day. Use 5-session sums as your minimum unit of analysis. Daily prints whipsaw; weekly sums reveal direction.
  • Mixing board scopes. A KOSPI-only figure and a KOSPI-plus-KOSDAQ figure for the same day can differ by trillions of won. Always confirm which scope a headline number covers before comparing sources.
  • Treating retail crash-buying as a contrarian buy signal. It is the structural default. The informative retail signal is capitulation — retail flipping to heavy net selling after sustained buying — or leverage-funded buying visible in a rising margin balance against falling deposits.
  • Quoting provisional numbers. Same-day figures are provisional and can change before the session’s data is complete. Date-stamp everything and prefer completed sessions.

FAQ

Where can I see the National Pension Service’s flows specifically?

The KRX daily table shows the pension-fund sub-category, which the NPS dominates but does not equal — other public pensions and funds are included. Exact NPS-only positioning appears with a lag in the fund’s own disclosures. For daily reading, treat the pension-fund line as a reasonable NPS proxy while remembering it is an aggregate.

Do the three cohorts’ net flows sum to zero?

Approximately, yes — every trade has a buyer and a seller, so foreigners, institutions, individuals, and the small “other corporations” categories net out to roughly zero each session. This is why the data is best read as a transfer map: the question is never “was there buying?” but “who bought from whom, and which side was acting mechanically?”

Is foreign or institutional flow the better index predictor?

Neither predicts reliably day to day. Historically, persistent multi-week foreign flows have tracked medium-term index direction most closely, pension flows tend to be contrarian by construction (rebalancing), and retail flows are contrarian by behavior. The strongest setups are alignments — for example, foreign buying while pension rebalancing has already run its course — rather than any single column.

How large is “large” for a daily flow?

There is no fixed threshold, and magnitudes vary widely from session to session. In early August 2026, for example, completed-session foreign flows on KOSPI-plus-KOSDAQ ranged between roughly -3.3 trillion and +2.8 trillion won within nine sessions. The most useful yardstick is a figure’s size relative to its own recent window — a print near the top of its window’s range deserves attention regardless of any absolute cutoff.

Sources

Disclaimer: This is an information and analysis publication, not investment advice. See our Methodology for data sources, standards, and our corrections policy.

Related reading