🔗 Related: KOSPI May 12 selloff / rebound — AI national-dividend / windfall-tax debate · Why Korea Part 4 — $6.7B inflow vs. value-trap setup · Korean Semis May 6 Rally Synthesis · Korea Daily Market Hub · Samsung Electronics Citi TP ₩460,000

The KOSPI opened May 13 at -1.69%, slid toward 7,400, then closed at 7,844.01 (+2.63%) — a fresh all-time high. Foreign investors net-sold ₩3.76tn, and the index made a new high anyway. The KOSDAQ closed -0.20% and local-DB breadth was 955 advancers vs. 1,455 decliners — 500 more stocks fell than rose. “Index hits new high, market feels weak” is the actual signature of the session.

TL;DR

• The biggest oddity is the intraday V-reversal. Open -1.69%, slide toward 7,400, close 7,844.01 (+2.63%) at a new all-time high.
• But the KOSDAQ closed -0.20%, and local-DB breadth was 955 up vs. 1,455 down — this is a compressed large-cap rally, not a broad rally.
• Foreigners net-sold ₩3.76tn but retail (+₩1.89tn) and institutions (+₩1.69tn) absorbed it. Domestic liquidity pulled the index up.
• The leadership axis was semis large-caps + substrates + autos. Conversely, batteries, parts of nuclear / power, gaming, and individual growth stocks lagged.
• The macro backdrop was outright unfriendly, yet the index rallied. Hot US CPI, WTI back above $100, Middle East tension, soft prior-day SOX. Macro lost to flow and large-cap news flow.
• The three things to verify today (May 14) — Samsung Electronics / SK hynix foreign selling stopping, Samsung Electro-Mechanics / Daeduck Electronics institutional buying continuing, KOSDAQ catching up. If any one fails, May 13 was a “large-cap short-cover + flow-reversal rally” rather than the start of a new leg.

1. What happened — the intraday picture

1.1 Time-of-day KOSPI movement

09:00 open: -1.69% (vs. prior close 7,643)
Morning: slide toward \~7,400 (estimated low \~ -2.5%)
Afternoon: rebound despite continued foreign selling, on retail + institutional buying
14:00+: acceleration
15:30 close: 7,844.01 (+2.63%, all-time high)

Intraday range: \~4.3% from low to close

Closing at a new all-time high is the strongest move since the May 6 +6.5% session (when KOSPI cleared 7,000). The critical difference: May 6 saw foreign net buying of ₩3.13tn; May 13 saw foreign net selling of ₩3.76tn.

1.2 May 13 market-level flow

Combined domestic net buying (retail + institutional) ≈ ₩3.58tn vs. foreign net selling ₩3.76tn — almost 1:1 absorption.

1.3 Index-vs-breadth divergence

Local-DB breadth May 13:
Total listed: \~2,827
Advancers: 955 (33.8%)
Decliners: 1,455 (51.5%)
Flat: \~417 (14.7%)

→ 500 more stocks fell than rose
→ Index +2.63% all-time high, but more than half the market was down

This is the operational definition of “compressed large-cap rally.” The top 10-20 names by market cap pulled the index, while most mid- and small-caps stayed out.

2. Leadership stocks — where the money went

2.1 The six names that led

The read is direct:

• Samsung Electronics + SK hynix = index-defense axis. The #1 and #2 market-cap names had to move together for KOSPI to print a new all-time high.
• Samsung Electro-Mechanics + Daeduck Electronics = AI-component / substrate diffusion axis. The names from Korea AI PCB Ecosystem followed semis higher.
• Hyundai Motor + Hyundai Mobis = non-semis large-cap rotation axis. Foreigners did not sell Korea wholesale; they sold parts of memory and rotated some into auto large-caps.

2.2 What didn’t participate — the underside of breadth

Those laggards are what produced KOSDAQ -0.20% and the 1,455 down-tickers.

3. The macro backdrop was unfriendly — why did this rally happen?

3.1 The risk-off inputs going into May 13

In a normal regime, this combination produces a -2% to -3% Korea open. Open was -1.69%. The intraday V-reversal that followed → all-time-high close. Macro variables didn’t set the price in this session.

3.2 Why was macro ignored — three hypotheses

Hypothesis A — self-reinforcing flow. With KOSPI in record-high territory after the May 6 +6.5% session, dry-powder waiting to “buy any pullback” was thick. -1.69% open → bid trigger → foreign supply absorbed.

Hypothesis B — large-cap news flow. Samsung Electronics Citi TP raise to ₩460,000 on May 11; multiple foreign-broker memory-cycle re-rating reports across May 12-13; refreshed SK hynix HBM share work — concentrated upgrade flow in the same week. Single-name news flow outweighed macro.

Hypothesis C — Hyundai Motor 1Q earnings + autonomy momentum. Hyundai Motor +9.91% and Hyundai Mobis +18.43% are hard to explain by simple sector rotation alone. Coverage on autonomous driving / robotaxi, plus a reframing of Hyundai Motor’s US-plant IRA exposure, likely added single-name catalysts.

The most defensible read is that all three operated in parallel. None of the three alone explains the -1.69% → +2.63% (4-point reversal).

4. The ₩3.76tn foreign selling — dissected

4.1 What foreigners actually sold and bought

Foreigners didn’t sell Korea wholesale. Memory mega-caps were sold; AI components (Samsung Electro-Mechanics, Daeduck Electronics) and non-semis large-caps (Hyundai Motor) saw some buying. The absolute net-sell number (-₩3.76tn) was large, but direction wasn’t one-way.

4.2 Linking to Why Korea Part 4

Why Korea Part 4 framed the paradox of $6.7bn YTD ETF inflow to Korea with forward PER still below the 10-year average. “Earnings outran price” was the resolution. May 13’s foreign selling can be read as short-term profit-taking inside that cumulative inflow context.

YTD 2026 Korea ETF net inflow: \~$6.7bn (\~₩9.5tn)
May 13 foreign KOSPI net sell: -₩3.76tn

→ May 13 sell ≈ 40% of cumulative YTD inflow
→ Large as a single-session print, but partial-profit-take in cumulative terms
→ Single-session data is insufficient to declare a trend reversal

The key is whether foreign selling stops or turns to buying today (5/14) and through the rest of the week. May 7 saw a record single-day ₩7.15tn foreign net-sell that flipped back to buying immediately after. Single-session prints don’t define trends.

5. Compressed rally vs. broad rally — what’s the difference?

5.1 Side-by-side

May 13 is clearly a compressed rally. A broad rally requires KOSDAQ participation and foreign buying — neither was present.

5.2 Signals that end a compressed rally

Signal 1: Large-caps stop adding gains, go flat
          → Top-10 names go sideways
          → Compression energy exhausted

Signal 2: KOSDAQ strong recovery
          → 1,176 → 1,200+
          → Capital diffuses to mid/small-caps (= broad-rally transition)

Signal 3: Foreign selling accelerates or stops
          → Acceleration → compressed rally also ends
          → Stop / reversal → sustainable mode

Which of these three surfaces first determines the next 1-2 weeks for KOSPI.

6. Analytical implications — “what’s working works harder”

6.1 Where the money is concentrating

The structure that produced the May 13 close:

The four clusters share: (a) top market-cap, (b) recent foreign-broker target-price upgrades, (c) high earnings visibility.

The names that lagged share the opposite features — (a) small / mid-cap, (b) thin foreign coverage, (c) priced more on expectation than delivery: battery (selected), gaming, biotech, parts of nuclear / power.

6.2 Signals from an analytical perspective

• The 5/13 all-time high is the result of macro-ignored compression flow. Near-term momentum is strong, but the conditions to sustain it are demanding.
• Whether foreign selling is a trend or a single-session print isn’t determinable from one day. 5/14 through 5/16 require additional observation.
• Without KOSDAQ catching up, this can’t be called a “broad rally.” Compressed rallies are inhospitable to chasing large-caps and unhelpful to mid/small-caps.

6.3 Three checks for today (5/14)

1. Does foreign selling on Samsung Electronics / SK hynix stop? Stop = continuation candidate; accelerate = 5/13 was one-off.
2. Does institutional buying on Samsung Electro-Mechanics / Daeduck Electronics persist? The AI-component diffusion axis verification.
3. Does KOSDAQ catch up? Capital diffusion / broad-rally transition signal.

If any of these three fails, May 13 reads as “large-cap short-cover + flow-reversal rally”. All three triggering = trend-shift candidate.

7. Bottom line

May 13’s KOSPI 7,844.01 all-time high was a V-reversal that ignored macro headwinds. Foreign -₩3.76tn was absorbed by retail +₩1.89tn and institutions +₩1.69tn. SK hynix +7.68%, Samsung Electro-Mechanics +7.41%, Daeduck Electronics +11.64%, Hyundai Mobis +18.43% — large-caps did the work. KOSDAQ closed -0.20% with 500 more decliners than advancers.

This is a compressed large-cap rally, not a broad rally. Macro lost to flow and single-name news flow. Structurally, this widens the gap between names that work and names that don’t.

The verification today (5/14) is simple. Does foreign selling stop, do institutions keep buying Samsung Electro-Mechanics / Daeduck Electronics, does KOSDAQ catch up. All three = the start of a trend shift. Any failing = single-session compression rally. The answer arrives within 1-2 sessions.

FAQ

Q: How does the KOSPI hit a new all-time high while foreigners sell ₩3.76tn? A: Retail (+₩1.89tn) and institutions (+₩1.69tn) absorbed it — combined ₩3.58tn vs. foreign ₩3.76tn = near-1:1 offset. The risk is that single-session foreign selling congealing into a sustained trend would be hard for domestics to defend alone. The foreign direction across the rest of this week is the variable that matters.

Q: Why did KOSDAQ close -0.20%? A: Capital concentrated in large-caps. In a compressed top-of-cap rally, mid/small KOSDAQ names don’t tag along. KOSDAQ recovery requires capital diffusion — that needs the KOSDAQ average daily turnover to recover and foreign + institutional buying to align.

Q: Hot CPI, WTI back above $100, Middle East tension — why did markets ignore them? A: Three hypotheses appear to operate in parallel: (1) dry-powder around record highs after the May 6 +6.5% session, (2) concentrated single-name upgrade flow across the same week (Samsung Electronics Citi TP raise, multiple memory re-rating reports), (3) Hyundai Motor / Hyundai Mobis autonomy / robotaxi catalysts. Single-name news flow outweighed macro.

Q: How does May 6 (+6.5%) compare to May 13 (+2.63%)? A: The most important difference is foreign direction. May 6 = foreigners +₩3.13tn; May 13 = foreigners -₩3.76tn. Same strong move, opposite flow signature. May 6 = foreign-driven cohesive advance; May 13 = domestic-absorbed compressed large-cap rally.

Q: What kinds of names do compressed rallies favor? A: Generalizing is hard, but May 13 data shows: top market-cap + foreign-broker coverage + earnings visibility tends to attract the flow. The four clusters (semis flagships, AI substrate / SUMs, auto large-caps, display) are examples. Mid/small-cap names with thin coverage and expectation-based valuation participate later.

Q: Are foreigners selling Korea wholesale? A: No. The May 13 mix shows Samsung Electronics / SK hynix selling alongside Samsung Electro-Mechanics / Hyundai Motor / LG Display buying — partial profit-take in memory + rotation into other large-caps. Against the YTD $6.7bn inflow (see Why Korea Part 4), May 13’s sell was ~40% of cumulative inflow — large for one day, modest as a profit-take in cumulative terms. Insufficient data to declare trend reversal.

Q: What does KOSDAQ need to participate? A: (a) Foreign selling to stop or reverse, (b) KOSDAQ average daily turnover to recover, (c) earnings momentum to diffuse into mid/small-caps. Currently item (1) is unverified, so KOSDAQ recovery may take a few more sessions.

Q: Is this compression a dangerous signal? A: Data is insufficient to call it dangerous. Compressed rallies can be the early signal of a strong trend or a late-stage signal. The distinguishing test is simple — does capital diffuse from large-caps to mid/small-caps over the next 1-2 weeks (trend start), or do large-caps deflate and the rally ends (late-stage).

This article is for research and informational purposes only and does not constitute investment advice. KOSPI / KOSDAQ closes, daily moves, and investor-segment flow are from KRX and media reports (Yonhap, AJU Press). Single-name flows reference local data (screener_kr.db / kr_market_surface_daily) and may differ from official tick-level data. Global variables (US CPI, WTI, Middle East, SOX) come from public reporting. Hypotheses A / B / C are analyst inferences and may differ from actual participant decisions. Whether foreign -₩3.76tn is a trend or a single-session print cannot be settled from one day of data; the May 14-16 pattern adds the needed confirmation. Analysis can be wrong. Data cut: May 13, 2026 KST (Korea close) and May 14, 2026 KST (analysis).

Disclaimer: For research and information purposes only. Not investment advice. Names cited are for analytical illustration; readers should perform their own due diligence and consult licensed advisors before any investment decision.