Foreign Flow Watch: Who Is Buying Korean Stocks? (Live Data)

Who is actually buying and selling Korean stocks? This is the permanent home of Foreign Flow Watch — K-Export Stars’ foreign investor flow tracker covering 2,555 KOSPI and KOSDAQ common stocks, refreshed every trading day from official brokerage API data. The tables below always show the latest settled trading day.

📊 Foreign Investor Flows — Latest Trading Day

Loading latest data…

🗞️ Weekly Flow Watch Issues

Every week we go beyond the raw numbers: who sold, who bought, and the why behind it — read directly from Korean-language DART filings and KRX data that rarely make it into English.

• Two Stocks Lost 140% of the KOSPI’s Weekly Decline — Week of August 7, 2026
• Korea Retail Took Profit Into the Record Rally — Week of July 31, 2026
• Foreigners Sold the KOSPI but Bought the Chip Giants — Week of July 24, 2026
• Who Foreign Investors Are Buying in Korean Exporters — Week of July 21, 2026
• → All issues (archive)

New to flow-reading? Start with our primer: How Foreign Flows Move Korean Stocks.

⬇️ Download the Data

Free, updated every trading day, licensed CC BY 4.0. Cite as “K-Export Stars — Foreign Flow Watch (kexportstars.com)”, DOI 10.5281/zenodo.21833374.

These are derived aggregates computed by us — not a redistribution of raw exchange feeds. If you need something cut differently for research or reporting, ask us.

…

블록으로 기존 동일 구간을 통째로 교체.

기존 5줄에서 바뀐 것:
· 유지: Universe · Source · Valuation · Timing · What “foreign” means
· 추가: Retail(이제 수집됨) · FX(환율·시점) · Preferred lines · Index totals(상장폐지) · DOI
근거: Flow Watch #4 에서 우선주 처리가 20세션 수치를 37% 움직였고, 잔여 시총이
상장·폐지를 흡수한다는 점이 논지의 핵심이었는데 문서에 아무 설명이 없었다.
–>

🔬 Methodology

• Universe: 2,555 KOSPI + KOSDAQ common stocks (KRX Open API issue master). SPACs, ETFs, ETNs and preferred lines are excluded from market-wide totals.
• Source: daily investor net-buy volumes (foreign / institutional / individual, in shares) via the Korea Investment & Securities Open API — an authenticated official channel, not scraping. Closes, share counts, market capitalisation and sector indices come from the Korea Exchange Open API; corporate filings from DART.
• Valuation: ₩ figures are approximations — net shares × that day’s settled close, summed across trading days for 5/20-day windows. Valuing a whole week at a single price instead can shift the answer by roughly 20% in a volatile week; we have made that mistake and corrected it publicly.
• Individual (retail) flows: now collected across the full universe, not a sample. Earlier issues stated we did not have a market-wide retail figure — that limit no longer applies, and figures published from 2026-08 onward are market-wide.
• Currency: USD conversions use a single rate per issue, struck at that issue’s most recent settled close and stated in the issue itself. We do not restate figures quoted from earlier issues at a newer rate — comparing across issues means comparing across rates, and we say so rather than silently re-basing.
• Preferred lines: several Korean groups list a second, non-voting share class (Samsung Electronics’ is 005935). Whether it is counted can move a headline number materially — in the week of 3–7 August 2026 it changed Samsung’s 20-session foreign net from −$0.82bn to −$0.51bn, a 37% difference. Market-wide totals exclude preferred lines; where an issue counts them for a single company, it says so in the table.
• Index and market-cap totals: index-level market capitalisation is taken as reported by KRX at each settled close. It absorbs listings, delistings and rights issues over the period, so any “rest of the index” residual we publish is a subtraction rather than a constituent-level roll-up. We label those lines as approximate.
• Timing: collected nightly (02:00 KST) for the previous trading day; each run re-fetches a 30-day window, so gaps self-heal. Settled sessions only — Korean intraday snapshots circulate widely and get republished as if final.
• What “foreign” means and its limits: see our primer — custodial flows ≠ conviction, single-day prints can mislead, and net flows do not identify who traded with whom.
• Archive and citation: the daily files are archived with a DOI on Zenodo — 10.5281/zenodo.21833374 (concept DOI, always the newest release). Cite the version DOI of the release you used if you need a fixed snapshot. Source and history: github.com/james-brand/korea-market-data.

Found something wrong? We log corrections rather than editing silently — see our corrections policy, or open an issue on the data repository.

Disclaimer: informational and educational only — not investment advice. Flow data reflects settled daily prints and may be revised at source. Verify independently before acting.