Silicon2 on Korea Invest Insights

KOSDAQ Series Part 2 — Screening for Quality with ROE: 1,820 → 105 → 35 → 6. The 0.3% That Combines High ROE, Margin, Growth, and Flow

Sun, 10 May 2026 22:30:00 +0900

📚 KOSDAQ Series — Part 2. Part 1: KOSDAQ Complete Guide for Foreign Investors — 1,820 Companies, October 2026 Promotion / Relegation Reform

KOSDAQ’s promotion-relegation reform launches in October 2026. Only 100 of ~1,820 listed companies enter the Premium tier. The criteria are financial soundness, growth, and governance. That’s screen-able now. The wide net (ROE ≥15%, OP growth positive) returns 105 candidates. The tight net (ROE ≥25%, daily turnover ≥₩1bn) returns 35. Layering ROE + operating margin + earnings growth + foreign-and-institutional flow as 4 simultaneous filters compresses to 6 names. 1,820 → 105 → 35 → 6 — the final candidate set is 0.3% of the listed universe. The real alpha in this reform is not owning KOSDAQ broadly. It is the separation between names institutions can buy and names they cannot.

TL;DR

The 2-stage funnel: 105 → 35 → 6. Filter 1 (ROE ≥15%, OP growth positive) clears 105 of KOSDAQ’s ~1,820 names. Filter 2 (ROE ≥25%, ₩1bn daily turnover) cuts to 35. The intersection of high ROE, high margin, growth, and constructive flow narrows to 6 names — 0.3% of the universe.
Three different shapes — growth, margin, value. Silicon2 (ROE 47%, margin 18%, PER 15×) for balanced growth. Classys (ROE 26%, margin 51%, PER 27×) for highest-quality margins. Easy Bio (ROE 29%, PER 7×) for value.
The real alpha is in classification, not the list itself. The 2022 KOSDAQ Global Index proved curation works (+160% vs. broad +65%). It also proved that a “good list” without ETF / NPS / institutional plumbing doesn’t move money. The October reform is designed to fix the second part — Premium ETFs, NPS benchmark inclusion, and the People’s Growth Fund are the three explicit channels.
There are losers from the reform too. If Premium criteria are earnings-based, large-cap-but-loss-making names — Ecopro BM at PER ~596×, ABL Bio at -₩48.6bn, Robotis at PER ~390× — risk exclusion. Falling out of Premium means losing institutional and passive flow, which is a negative alpha vector worth tracking on the short side.
Don’t buy KOSDAQ. Separate it. The investment substance of the reform is identifying which KOSDAQ names institutions will be forced to own and which will lose their pseudo-benchmark status.

1. Why ROE matters — Premium-tier inclusion logic

1.1 The Premium tier’s selection criteria

Part 1 covered the headline framework: Premium (Tier 1) holds ≤100 names selected on financial soundness, growth, and governance. Final criteria are still being calibrated, but the existing KOSDAQ Global Segment requirements offer a useful prior:

Market cap ≥₩500bn or top 7%
Operating profit ≥₩30bn OR revenue ≥₩300bn
Corporate-governance rating ≥B
Financial-health and disclosure-quality requirements

Premium will likely be tighter than this. The design intent is “names that actually earn money,” not “names with large caps alone.”

1.2 Why ROE is the right starting filter

ROE measures how efficiently equity capital generates earnings. High ROE = the business uses small capital to produce large profit.

For Premium-inclusion screening, ROE works because:

High ROE:
→ Business has an earnings base (loss-making companies have meaningless ROE)
→ Capital efficiency — exactly what pension funds underwrite
→ Compatible with both growth and quality screens

Low ROE:
→ Either no earnings, or capital base too large relative to earnings
→ Pension capital can't easily be deployed
→ Lower probability of Premium inclusion

1.3 What the KOSDAQ Global Index already proved

KOSDAQ Global Index (\~50 quality names):
2022.11.21 inception (1,000) → 2026.05.08 (2,602) → +160.2%

Broader benchmarks (same period):
KOSDAQ broad: +65%
KOSDAQ 150: +93%
KOSDAQ Global: +160%

Annualized:
KOSDAQ Global: \~31.8% CAGR
KOSDAQ 150: \~20.9% CAGR
KOSDAQ broad: \~15.6% CAGR

Verification: Global CAGR = (2,602 / 1,000)^(1/3.46) - 1 = 31.8% ✓

The asset class isn’t the problem. Selection is. The Premium reform institutionalizes selection. For investors, the alpha lies in identifying probable-Premium names before the list is published.

2. The 2-stage funnel — wide net to tight net

2.1 Filter 1 — ROE ≥15% + positive OP growth

Start with the widest defensible net.

Filter 1 conditions:
- ROE ≥15%
- OP growth YoY positive
- OP positive (no losses)
- No liquidity / flow / PER constraints applied

(Financial cut: 2026-05-07. Price / flow cut: 2026-05-08.)

Group	Names passing
All Korean listed	173
KOSDAQ	105
KOSPI	68

105 of KOSDAQ’s ~1,820 names clear Filter 1 — about 5.8%. That’s the size of the universe of profitable, growing, high-ROE businesses on the small-cap board.

The top 27 of the 105, sorted by ROE:

Name	Ticker	ROE	OP growth	OP margin	PER	20D price	20D F+I net buy
Sunic System	171090	101.5%	+1,311%	21.6%	5.2×	+7%	-₩39.6bn
Hanpass	408470	94.2%	+58%	12.4%	—	-22%	-₩5.5bn
Genic	123330	59.8%	+150%	19.1%	7.0×	-14%	-₩14.6bn
JTC	950170	57.4%	+117%	15.4%	3.3×	+7%	+₩0.3bn
S-Connect	096630	49.8%	+27%	4.8%	1.3×	+390%	-₩1.1bn
Silicon2	257720	46.9%	+49%	18.4%	14.8×	+9%	+₩58.6bn
Sungho Electronics	043260	46.3%	+21%	3.3%	6.9×	+3%	-₩4.0bn
ITCEN Global	124500	45.9%	+378%	3.2%	15.1×	+20%	-₩6.7bn
BHI	083650	44.4%	+245%	9.8%	24.8×	0%	-₩141.4bn
SM Entertainment	041510	41.6%	+110%	15.6%	8.9×	+6%	-₩0.9bn
Inca Financial Service	211050	39.7%	+10%	9.3%	9.1×	-12%	₩0bn
Alteogen	196170	39.4%	+321%	49.5%	170.7×	-9%	+₩20.6bn
Wooone Construction	046940	36.8%	+2,300%	17.9%	1.3×	+23%	-₩2.1bn
Inzisoft	100030	35.1%	+109%	12.4%	3.1×	+17%	₩0bn
Dong-Ah Eltec	088130	33.6%	+670%	19.4%	1.4×	+41%	-₩12.7bn
KONA I	052400	32.6%	+166%	28.8%	8.9×	+7%	+₩7.5bn
KNJ	272110	30.8%	+57%	26.3%	6.8×	+11%	-₩0.9bn
Easy Bio	353810	29.1%	+39%	9.4%	7.0×	+4%	+₩1.7bn
Global Tax Free	204620	27.8%	+34%	18.8%	9.1×	+20%	+₩6.4bn
Mico	059090	27.2%	+7%	10.3%	15.2×	+72%	+₩29.4bn
PharmaResearch	214450	26.9%	+70%	40.0%	28.4×	+16%	-₩21.9bn
Classys	214150	26.2%	+39%	50.6%	27.1×	+5%	+₩11.2bn
Leeno Industrial	058470	22.4%	+43%	47.5%	30.2×	-4%	-₩80.5bn
Jeju Semiconductor	080220	19.2%	+274%	11.9%	23.0×	+27%	+₩116.0bn
Dawonnexview	323350	20.4%	+208%	15.0%	11.9×	+96%	+₩2.3bn
Vitzro Cell	082920	18.6%	+34%	28.5%	14.0×	+52%	-₩5.0bn
iNet	462980	18.4%	+725%	8.5%	14.7×	-10%	-₩0.1bn

Three things become immediately visible:

One — Filter 1 alone mixes good candidates with traps. S-Connect (+390% over 20 days), Dawonnexview (+96%), Mico (+72%) — already-overheated names. BHI (institutional net sells -₩141.4bn) — money walking out the door. Hanpass (data instability) — newly listed, unreliable record.

Two — high ROE with thin margin is a quality trap. Sungho Electronics (ROE 46.3%, margin 3.3%), ITCEN Global (margin 3.2%), S-Connect (margin 4.8%). Margins this thin disappear at the first environmental change.

Three — Inca Financial Service stands out. ROE 39.7%, PER 9.1×, clean Filter-1 entry. Margin 9.3% and 20D price -12% mean it’s not a market leader. Discussed separately as a “non-AI quality candidate” in §5.

2.2 Filter 2 — ROE ≥25% + ₩1bn daily turnover

Tighten Filter 1’s 105 with two more constraints.

Filter 2 additional conditions:
- ROE ≥25% (up from 15%)
- 20-day average daily turnover ≥₩1bn

This returns 35 names — under 2% of the listed universe.

2.3 Filter 2’s top 10

Name	ROE	Margin	OP growth	PER	PBR
Silicon2	46.9%	18.4%	+49.3%	14.8×	5.5×
Alteogen	39.4%	49.5%	+320.9%	170.7×	54.4×
KONA I	32.6%	28.8%	+166.2%	8.9×	2.5×
KNJ	30.8%	26.3%	+57.4%	6.8×	1.8×
Easy Bio	29.1%	9.4%	+38.9%	7.0×	1.8×
Global Tax Free	27.8%	18.8%	+34.3%	9.1×	2.3×
Mico	27.2%	10.3%	+6.8%	15.2×	3.7×
PharmaResearch	26.9%	40.0%	+70.0%	28.4×	6.8×
Classys	26.2%	50.7%	+39.4%	27.1×	6.4×
VT	25.7%	19.0%	-25.2%	10.2×	2.5×

2.4 Lesson — quality rises with each cut

Funnel by stage:
Total KOSDAQ: 1,820
Filter 1 (ROE ≥15%, OP growth +): 105 (5.8%)
Filter 2 (ROE ≥25%, ₩1bn turnover): 35 (1.9%)
Filter 3 (ROE+margin+growth+flow combined): 6 (0.3%)

1,820 → 105 → 35 → 6. The final investable candidate set is 0.3% of the universe. This is the operational mechanic behind the KOSDAQ Global Index’s 2.5× outperformance — narrow enough selection, applied consistently.

2.5 The five high-ROE traps to avoid

Trap 1 — High ROE with thin margin. Sungho Electronics (ROE 46.3%, margin 3.3%). High ROE looks attractive but earnings quality is low.

Trap 2 — Holding-company accounting. Soulbrain Holdings (ROE 32.1%, PER 1.6×, margin 1.6%). PER 1.6× looks impossibly cheap, but margin 1.6% suggests holdco-accounting structure rather than operating reality.

Trap 3 — Data instability. Hanpass (ROE 94.2%). Newly listed; data history hasn’t stabilized.

Trap 4 — Already overheated. Mico (20D price +71.6%). Flow is constructive but the move is too fast — buying now is chasing.

Trap 5 — Best-in-class quality at unsupportable multiple. Alteogen (ROE 39.4%, margin 49.5%, PER 170.7×). Fundamentally a great business, but the multiple already prices several years of expected growth.

3. The 6 names that pass all four filters

3.1 Silicon2 — most balanced growth

ROE 46.9% / Margin 18.4% / OP growth +49.3% / PER 14.8×
20D F+I net buy: +₩58.6bn
20D price: +9.2%

K-beauty distribution platform — connecting Korean cosmetic brands to overseas channels. The structural advantage is brand-agnostic exposure: with a portfolio of brand mandates rather than a single hero brand, brand-cycle risk is diversified.

ROE 47%, margin 18%, growth +49%, PER 15×, F+I flow strongest in the cohort at +₩58.6bn. Cleanest 4-of-4 simultaneously.

The caveat is that it’s already a known leader — chasing carries setup risk. Watch the 43,000–45,000 won support zone or a confirmed breakout above the prior high.

3.2 Classys — highest-quality margins

ROE 26.2% / Margin 50.7% / OP growth +39.4% / PER 27.1×
20D F+I net buy: +₩11.2bn
20D price: +5.3%

Aesthetic medical-device company. HIFU (high-intensity focused ultrasound) systems plus consumables sold to dermatology clinics globally.

Margin 50.7% is the headline — best-in-class even within the high-ROE cohort. Razor-and-blades structure: device sale + recurring consumable. 2026 estimate ROE 29.4%, PER 19.4× — not particularly stretched for a quality compounder.

Watch around the 55,000-won support zone.

3.3 GlobalTaxFree — cheapest of the high-ROE set

ROE 27.8% / Margin 18.8% / OP growth +34.3% / PER 9.1×
20D F+I net buy: +₩6.4bn
20D price: +20.4%

Tax refund operator for foreign tourists in Korea. As Korea attracts more visitors driven by K-content and K-beauty tourism, the structural revenue pool grows.

ROE 28%, margin 19%, PER 9× — cheapest in the high-ROE cohort. Risk: any softening in tourism / consumption flow can compress the multiple quickly.

3.4 Easy Bio — cheapest high-ROE value play

ROE 29.1% / Margin 9.4% / OP growth +38.9% / PER 7.0×
20D F+I net buy: +₩1.7bn
20D price: +3.5%

Covered in detail across Part 1 and Part 2 of the Easy Bio sequence. 78% of revenue is now feed additives, building a North America M&A platform. Trades at a 50–64% discount to global peers (Phibro 18×, Anpario 15×).

PER 7× with ROE 29% — a rare combination across the entire Korean market. Cheap high-ROE value play, though not a market leader by character. Margin of 9.4% means it isn’t a high-margin business. Premium-inclusion eligibility depends on whether market cap and absolute earnings clear the still-being-finalized criteria.

3.5 KNJ — numbers good, flow not yet engaged

ROE 30.8% / Margin 26.3% / OP growth +57.4% / PER 6.8×
20D F+I net buy: -₩0.9bn
20D price: +10.9%

Semiconductor wet-cleaning / etching equipment. Numbers are excellent — ROE 31%, margin 26%, growth +57%, PER 7×. But foreign + institutional flow at -₩0.9bn means institutions haven’t engaged.

Buy priority is lower until flow inflects. Good numbers without institutional engagement don’t typically move the price meaningfully.

3.6 PharmaResearch — quality good, flow exiting

ROE 26.9% / Margin 40.0% / OP growth +70.0% / PER 28.4×
20D F+I net buy: -₩21.9bn
20D price: +15.8%

Maker of Rejuran (PDRN-based skin regeneration product). Margin 40%, growth +70% — quality is high. But 20D F+I flow at -₩21.9bn means institutions are net sellers. Price went up, institutions sold — a classic “be careful” combination.

4. Where the real alpha lies — at the separation, not the list

4.1 Don’t own KOSDAQ broadly

A “buy KOSDAQ” trade off the reform headline has no alpha. As shown in Part 1, much of the 1,820-name universe is loss-making or theme-driven.

The alpha is in the separation between names institutions can buy and names they cannot:

If included in Premium:
→ Auto-inclusion in dedicated Premium index ETFs
→ Eligible for pension allocations
→ Liquidity uplift, multiple re-rating

If excluded from Premium:
→ Same retail-driven flow regime as before
→ Possible loss of "leader" status
→ Names with cap-only profile (no earnings) may face further derating

4.2 The Premium-reform losers

This is the under-discussed angle. If Premium criteria are earnings-based, large-cap-but-loss-making names that currently sit in KOSDAQ 150 risk Premium exclusion:

Name	Market cap	Estimated 2026E NI	Implied PER	Risk
Ecopro BM	₩23tn+	~tens of ₩bn	~596×	May fail Premium criteria
ABL Bio	#9 by cap	-₩48.6bn (loss)	n/m	May fail earnings criterion
Robotis	₩5tn	~₩10bn	~390×	Cap excessive vs earnings

These aren’t bad businesses. They may have large future growth. But if Premium criteria require current earnings, exclusion would mean loss of institutional and passive-flow access — a negative alpha vector worth tracking on the short side, or at minimum a position-size reduction signal.

4.3 Three tradable angles

Angle 1 — Already-quality names that gain incremental flow. Leeno Industrial, Jusung Engineering, KOSDAQ Global Segment incumbents. Probably included in Premium. But the incremental alpha is limited — quality is already in the multiple.

Leeno’s 1Q26: revenue ₩99.8bn, OP ₩47.3bn, margin 47.4%. Target prices revised up to ₩150,000. High inclusion certainty but a less asymmetric setup than Angle 2.

Angle 2 — Earnings inflection + Premium-inclusion catalyst. Simmtech is the cleanest example. 2025 OP ₩11.9bn → 2026E OP ₩120-150bn — a 10× lift driven by AI server / memory module substrate mix shift. The market currently treats it as a cyclical-substrate recovery name. If 2026 OP exceeds ₩100bn, it becomes a viable Premium-inclusion candidate. Earnings inflection + policy-driven flow = double re-rating.

Angle 3 — Non-AI quality candidates. Inca Financial Service. Insurance distribution (general agency) business: 2025 revenue ₩1.022tn, OP ₩95.2bn, NI ₩71.3bn. 2026 NI target ~₩100bn. The market will likely interpret KOSDAQ reform through an AI / biotech / robotics lens — but if Premium criteria are earnings-and-governance based, profitable non-AI names also qualify. Surprise re-rating optionality.

5. Cross-section — high-ROE candidates × Premium-inclusion probability

5.1 High-ROE names with Premium-inclusion probability

Name	ROE	Premium probability	Reason
Silicon2	46.9%	High	All four (ROE / margin / growth / flow) clean
Classys	26.2%	High	Margin 50%+, stable profitability
Global Tax Free	27.8%	Medium	Cap and earnings size verification needed
Easy Bio	29.1%	Medium → High	Visible 2026 OP at ~₩50bn level should clear
KNJ	30.8%	Medium	Earnings size and cap thresholds need verification
PharmaResearch	26.9%	High	Margin 40%, stable profitability

5.2 Lower-ROE but high Premium probability

ROE under 25% can still clear Premium if absolute earnings and stability are strong:

Leeno Industrial — margin 47%, 1Q26 OP ₩47.3bn. Stable inclusion candidate.
Jusung Engineering — flagship semi-equipment name. Cap and earnings both clear typical thresholds.
SOOP — internet-broadcast platform. Stable profitability.
Simmtech — 2026 OP >₩100bn would clear earnings criterion.
Inca Financial Service — NI ₩70-100bn band. Non-AI quality candidate.

6. Caveats

6.1 Final criteria not yet published

The most important caveat. Final Premium-100 criteria are not yet released. Industry briefings are scheduled for June and finalization for July. This screen is probable-candidate identification, not a confirmed list.

6.2 High ROE has many traps

ROE alone is not enough. Earnings quality (margin), persistence (growth), and market validation (flow) must align. High ROE with 3% margin, holdco accounting, or unstable data history fails the quality test even if ROE numerically clears.

6.3 APR is KOSPI-listed, not KOSDAQ

For the avoidance of doubt — APR (KOSPI 278470) is on KOSPI, not KOSDAQ, and is not a direct beneficiary of the KOSDAQ promotion-relegation reform. Same with Cosmax (KOSPI 192820). Both are well-known Korean beauty names but on the wrong exchange for this reform.

6.4 Premium inclusion ≠ guaranteed price appreciation

If included, passive flow can arrive — but the lesson of the 2022 KOSDAQ Global Index is that the list part is easier than the flow part. Whether ETFs scale, NPS adopts the benchmark, and the People’s Growth Fund actually deploys to KOSDAQ — those determine whether classification translates to multi-year price action. Visible only after October.

7. Practical priority

Priority	Name	Character	Verdict	Trigger
1	Silicon2	High-ROE growth	Pullback buy #1	43,000–45,000 won support OR confirmed new-high breakout
2	Classys	High-margin device	Buy on support	~55,000 won support hold
3	Simmtech	Earnings inflection + Premium catalyst	Buy on confirmation	2Q26 OP path ≥₩50bn confirmed
4	Easy Bio	Cheap high-ROE value	Value candidate	1Q26 OPM ≥9.4% confirmed
5	Leeno Industrial	Stable inclusion candidate	Wait for pullback	15–20% retracement
6	Global Tax Free	Cheap high-ROE	Conditional buy	6,000-won base + sustained turnover

8. Bottom line

The investment substance of the KOSDAQ reform is not “buy KOSDAQ.” It is the separation between names institutions can own and names they cannot.

Premium-100 inclusion isn’t about market cap — it’s about earnings. The wide net (ROE ≥15%, OP growth positive) clears 105 of 1,820 (5.8%). The tight net (ROE ≥25%, ₩1bn turnover) clears 35 (1.9%). The simultaneous-4 (ROE + margin + growth + flow) clears 6 (0.3%). Silicon2 is the most balanced, Classys the highest-margin, Easy Bio the cheapest.

The corollary matters as much as the headline. Names that fail Premium criteria — large-cap-but-loss-making — face institutional and passive-flow loss. That’s a negative alpha angle worth tracking, possibly via reduced position size on cap-heavy thematic names.

Final criteria publish in July. Until then, the candidate list is preliminary. The October launch is when ETF / NPS / People’s Growth Fund deployment begins, and that’s when real flow either validates the curation or proves the 2022 KOSDAQ Global Index lesson all over again. The KOSDAQ Global Index showed selection matters 2.5× more than asset class. The same logic applies to Premium — which names you choose drives everything.

FAQ

Q: Why ROE as the starting filter? A: Premium-tier criteria emphasize “financial soundness and growth.” ROE captures both — it indicates the business has an earnings base and is using capital efficiently. This is exactly the profile pension capital can underwrite. A practical first filter for Premium-eligibility screening.

Q: Why narrow from 105 to just 6 final candidates? A: ROE alone is not enough. Earnings quality (margin), persistence (growth), and market validation (foreign + institutional flow) all need to coincide. The intersection across all four filters is the 0.3% of the universe — and this 0.3% is where the alpha sits.

Q: Why does Alteogen drop out of the final 6 despite having ROE 39%, margin 49%? A: Because PER 170× already prices in multi-year forward growth. “Good business” and “good entry price” are different questions. Alteogen the company is high-quality; Alteogen the entry-level position is not.

Q: Which names are likely to be excluded from Premium? A: Large-cap-but-low-earnings names. Ecopro BM (PER ~596×), ABL Bio (loss-making, -₩48.6bn), Robotis (PER ~390×). Future growth potential aside, current earnings criteria can fail them. Exclusion means losing institutional / passive-flow access — a negative alpha vector.

Q: Why include Simmtech in the watchlist if it’s not in the final 6? A: Simmtech’s 2026 OP forecast ₩120-150bn is a 10× jump from 2025 ₩11.9bn driven by AI server / memory module substrate mix. The market still treats it as cyclical recovery. If OP clears ₩100bn, it becomes Premium-eligible — a “double re-rating” candidate (earnings inflection + policy flow).

Q: Should one chase or wait? A: Most of the final 6 have already moved meaningfully. Each has a separate entry condition: Silicon2 wants 43,000–45,000 won support or confirmed breakout, Classys wants ~55,000 won support hold, GlobalTaxFree wants 6,000-won basing with sustained turnover. Wait for the trigger; the chase has worse risk-reward.

Q: What changes in June and July that matters? A: June industry briefings and the July KOSDAQ 30th-anniversary event are when Premium-100 final criteria publish. Whether the market-cap threshold is ₩500bn vs higher, whether the OP threshold is ₩30bn vs higher, and what governance grade is required — those details flip individual probabilities. Re-score the candidate list once the criteria are public.

This article is for research and informational purposes only and does not constitute investment advice. Screening data drawn from a proprietary database; financial cut at 2026-05-07, price-and-flow cut at 2026-05-08. Filter 1 (ROE ≥15%, OP growth positive) applied no liquidity / flow / PER constraints — wide-net design. Filter 2 (ROE ≥25%, ₩1bn turnover) and Filter 3 (margin + growth + flow added) layered for compression. KOSDAQ promotion-relegation reform details from Maeil Business Newspaper (2026-05-10), KRX KOSDAQ Global Segment operational rules, and FSC press releases. Premium-tier final criteria are not yet finalized — list, criteria, and timeline subject to change. APR (KOSPI 278470) and Cosmax (KOSPI 192820) are KOSPI-listed and not direct beneficiaries of the KOSDAQ reform. Individual-name judgments are analytical inferences and may be wrong. Data cut: May 10, 2026 KST.

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.

Korean Quality-Compounder Screen — Only 4 Names Pass 9 Filters: VM, Pamicell, Silicon2, Samyang Foods (And Why Pamicell Survives Both Screens)

Thu, 07 May 2026 08:30:00 +0900

🔗 Related: Earnings-Explosion Screen — 67 Names · Korea Daily Market Hub

The previous post showed an “earnings explosion” screen returning 67 names with the lesson “most have already moved.” This piece runs the inverse experiment: stack filters to the limit and see what actually clears. Nine simultaneous conditions — ROE ≥25%, revenue growth ≥+15%, OP growth ≥+20%, OPM ≥10%, P/E ≤25×, target headroom ≥+15%, daily turnover ≥₩3.0bn, foreign+institution net-buy, RSI <75 — and only four Korean equities pass.

Executive Summary

Only four names clear the nine simultaneous filters: VM, Pamicell, Silicon2, Samyang Foods. Out of the entire Korean listed universe. The narrowness of the result is the point.
The formula targets the intersection of “good business” and “what the market is currently buying.” ROE ≥25% catches efficient-capital users; P/E ≤25× catches names not yet expensively priced; foreign+institution net-buy catches names with active flow support. When all three filter dimensions clear simultaneously, quality, price, and flow are confirmed in one cohort.
Pamicell survives both screens. It cleared the previous earnings-explosion screen of 67 names and this 9-filter screen of 4 names. ROE 38%, OPM 36.3%, OP growth +72.3%, foreign+institution 20D +₩53.7bn. Two independent screening lenses, same conclusion — the strongest single signal in this cohort.
Silicon2 has the cleanest quality-vs-price combination. ROE 38.7%, P/E 13.3×, target headroom +36.8%. Among the four, the price-relative-to-quality combination is the cleanest. But its current leadership intensity is below Pamicell’s.
VM ranks #1 on raw quant but with elevated late-stage volatility risk. OP growth +242.9%, P/E 19.8×, target headroom +29.8%. On numbers alone the strongest. But +451% off 52-week low is “already-moved” territory — a tactical/aggressive holding rather than a core position.
The narrow-miss list is the more interesting research output. HD Hyundai Marine Engine (P/E 25.7×, missed by 0.7×), Classys (only flow filter failed), PharmaResearch (only flow filter failed), Isu Petasys (P/E too high), Samsung Electronics (RSI too high). One condition flips and these names enter — useful “watch list” candidates.

1. Why ROE 25% — The Foundation of the Screen

1.1 The One-Line Definition

ROE 25% means: for every ₩100 of equity, the company generates ₩25 of net income annually. In plain terms, capital is being deployed very efficiently.

1.2 Why This Is a Strong First Filter

ROE ≥25% carries three implications:

Implication	Read
High capital efficiency	Modest equity base, large absolute earnings
Reinvestment value	Re-deployed earnings compound the equity base
Multiple justification	Higher P/E or P/B has analytic basis

The critical question is whether the ROE is sustainable. A single year of ROE 25% is meaningless. To be a “good ROE,” it has to hold for 2–3+ years, be accompanied by revenue growth, and not depend on leverage.

1.3 Good ROE 25% vs Bad ROE 25%

Good ROE 25%	Bad ROE 25%
Comes with revenue growth	Revenue is flat; only NI spikes
OPM is high	Debt-driven ROE inflation
Sustainable across 2–3+ years	One-year peak from a cyclical top
Reinvestable market is large	One-off (asset sale, FX gain, divestiture)
P/E / P/B is justifiable vs growth	P/B ≥10× already, no headroom

That distinction matters. ROE ≥25% by itself is a first-filter for “good-business candidates,” not a buy signal. A real buy decision requires growth durability, valuation, flow, and chart position — which is exactly why the screen stacks nine filters.

2. The Nine Filters — Why Stack Them

2.1 The Formula

① ROE ≥ 25%
② 2026F revenue growth ≥ +15%
③ 2026F OP growth ≥ +20%
④ 2026F OPM ≥ 10%
⑤ 2026F P/E ≤ 25×
⑥ Target-price headroom ≥ +15%
⑦ 20-day average turnover ≥ ₩3.0bn
⑧ Trailing 20-day foreign + institution net-buy
⑨ RSI < 75 (or 20-day MA pullback confirmed)

2.2 What Each Filter Catches and Filters Out

Filter	Catches	Filters out
① ROE ≥25%	High capital-efficiency businesses	Profit-poor businesses
② Revenue ≥+15%	Top-line-growing businesses	ROE-from-cost-cutting / one-offs
③ OP ≥+20%	Earnings leverage	Top-line growth without margin lift
④ OPM ≥10%	Pricing power	Low-margin top-line expansion
⑤ P/E ≤25×	Not yet expensively priced	Already-discounted-future-expectations
⑥ Target headroom ≥+15%	Reward room remains	At or near target price
⑦ Turnover ≥₩3.0bn	Tradable	Ultra-small-cap illiquidity
⑧ Foreign+institution net-buy	What the market is currently buying	Where smart money is exiting
⑨ RSI <75	Not yet overheated	Short-term overbought

2.3 The Core Insight Behind Stacking

When all nine clear simultaneously, three independent dimensions confirm at once:

Quality: ROE + OPM + revenue/OP growth
Price: P/E + target headroom
Market validation: foreign+institution flow + RSI + turnover

A “good business” (quality) that is “not yet expensively priced” (price) and “currently being bought” (market validation) — the intersection.

The fact that nine simultaneous filters are hard to clear is itself the value. Where the earnings-explosion screen returned 67, this screen returns 4. Strict filters produce meaningful results.

3. The Four That Pass

Rank	Name	ROE	P/E	OP growth	OPM	Target hd	20D return	Foreign+Inst 20D	Profile
1	VM (KOSDAQ: 081000)	34.2%	19.8×	+242.9%	29.4%	+29.8%	+42.2%	+₩25.1bn	Aggressive momentum
2	Pamicell (KOSPI: 005690)	38.0%	21.1×	+72.3%	36.3%	+21.9%	+36.7%	+₩53.7bn	Focus candidate
3	Silicon2 (KOSDAQ: 257720)	38.7%	13.3×	+30.3%	18.0%	+36.8%	+15.1%	+₩31.7bn	Quality candidate
4	Samyang Foods (KOSPI: 003230)	35.3%	18.8×	+31.7%	23.4%	+38.2%	+7.1%	+₩16.3bn	Steady compounder

All four share ROE 34–39%, OPM 18–36%, foreign+institution net-buy. Common: capital efficiency + earnings growth + active flow. Differences are character.

4. Per-Name Read

4.1 Pamicell — The Cross-Screen Survivor

Why Pamicell is the practical #1: it cleared the previous earnings-explosion screen of 67 and this 9-filter screen of 4. Two independent screening lenses, same conclusion.

ROE 38.0% → top-tier capital efficiency
OPM 36.3% → highest margin among the four
OP growth +72.3% → earnings leverage confirmed
P/E 21.1× → not yet expensively priced
Foreign+Inst 20D +₩53.7bn → flow strongest in cohort

Pamicell was already analyzed in Series Part 3 with a GARP Score of 150 (top of the materials-stock universe). The fact that it also clears this independent ROE-25%-multi-factor screen means the appeal holds across very different filter logic — quality + price + flow as a distinct lens from the original GARP composite.

That said, 20-day return +36.7% raises near-term overheating risk. The right posture is “wait for pullback or scale in after 1Q earnings confirm” rather than “full conviction now.”

Tracking signals:

₩19,000–20,000 holding as support (on pullback)
1Q26 OP ≥₩11.5bn with OPM ≥30%

4.2 Silicon2 — Cleanest Quality-vs-Price

Silicon2 has the cleanest combination of capital efficiency relative to multiple in the cohort.

ROE 38.7% → highest ROE of the four
P/E 13.3× → cheapest of the four
Target headroom +36.8% → second-highest in cohort
20D +15.1% → most stable of the four (not overheated)

ROE 38.7% with P/E 13.3× is rare in the Korean market. When ROE is this high but P/E sits at 13×, the market is either (a) skeptical that the ROE is sustainable, or (b) hasn’t fully discovered the name yet.

Silicon2 is a global K-beauty distribution platform. Strong overseas-revenue mix and good margins. But distribution-business dynamics — freight, FX, margin variability — matter. If the market’s P/E 13× reflects “distribution-business discount,” that’s defensible; if it reflects “growth deceleration concern,” it warrants caution.

Recent leadership intensity is weaker than Pamicell’s. 20D +15.1% is steady but not the strongest mover. Closer to “high-quality name that can move again” than “currently leading.”

Tracking signals:

₩41,500–43,000 holding as support (on pullback)
2026F OP growth ≥+20% maintained
Freight / margin pressure not materializing

4.3 VM — Best Quant Score, Highest Variance

On quant alone, VM is the strongest of the four.

OP growth +242.9% → dominant #1 in cohort
OPM 29.4% → second highest
P/E 19.8× → reasonable
Target headroom +29.8% → adequate
Foreign+Inst 20D +₩25.1bn → solid

The issue is price location. 52-week-low to current: +451%. A 4.5×-in-a-year run. This isn’t “undervalued discovery” — it’s “already-moved territory.”

OP growth +242.9% may carry “turnaround base-effect” character. If 2025 OP was depressed, even modest absolute lift produces a large growth rate. Whether the growth is “structural” or “base effect” is the analytical question.

VM is more of an aggressive position than a core hold. Upside is meaningful when right; if wrong, the +451%-off-low position implies large downside.

Tracking signals:

₩52,000–55,000 pullback support
Order-momentum maintained
2026F OP estimate revisions

4.4 Samyang Foods — Solid Compounder, Not the Current Leader

Samyang Foods clears all nine on numbers.

ROE 35.3% → high
OPM 23.4% → strong
P/E 18.8× → reasonable
Target headroom +38.2% → highest in cohort

A genuinely strong business with a clean narrative — Buldak ramen global expansion, rising overseas-revenue share, margin expansion.

But 20D +7.1% is the weakest of the four. In the current market regime where momentum-leadership matters, Samyang isn’t on the front line. A portfolio-stability slot, not a short-term-momentum slot.

Tracking signals:

₩1.25M–1.30M pullback for entry consideration
Overseas growth not decelerating
OPM peak-out check

5. The Near-Miss List — Where the Real Optionality Sits

Name	Disqualifier	Failed condition	Read
HD Hyundai Marine Engine	P/E 25.7×	⑤ P/E ≤25×	0.7× miss. Effectively at-pass
Classys	Foreign+Inst net-sell	⑧ Net-buy	Single-flag: flow recovery → in
PharmaResearch	Foreign+Inst net-sell	⑧ Net-buy	Pullback-style quality. Flow confirmation needed
Isu Petasys	P/E 41.4×, target +10.5%	⑤⑥ P/E + target	Leader, but priced
Samsung Electronics	RSI >75	⑨ RSI	Hold OK; new entry per filter blocked

The near-miss list is more interesting than it looks because a single condition relaxation puts each one in the active cohort.

HD Hyundai Marine Engine is 0.7× away. In an aggressive-market regime where the P/E filter is loosened from 25× to 30×, it enters. Classys and PharmaResearch enter the moment flow turns positive. Samsung Electronics enters when RSI cools to <75.

Tracking these “waiting candidates” is part of the screen’s value.

6. The Cross-Screen Intersection — Where 67 Meets 4

The previous post returned 67 names from the earnings-explosion screen. This post returns 4 names from the ROE-25%-9-filter screen. Cross the two — and Pamicell appears in both.

[Earnings-explosion screen — 67 names]
Conditions: 2025/2026 profit + OP/NI each ≥+80%
Result: 67 → most already moved
Pamicell: ✓ pass (GARP Score 150)

[ROE-25% screen — 4 names]
Conditions: ROE ≥25% + growth + P/E ≤25× + flow + RSI
Result: 4 → quality+price+flow intersection
Pamicell: ✓ pass (ROE 38%, OPM 36.3%, flow +₩53.7bn)

Cross-screen result:
→ Pamicell survives both lenses with the same conclusion
→ The "earnings explosion" growth lens AND the "ROE+price+flow" quality lens both confirm

This is why cross-screen analysis is more meaningful than any single screen. A single filter creates confirmation bias; two independent filters provide robustness.

Whether Silicon2, VM, and Samyang Foods also cleared the earnings-explosion screen is a separate cross-check. The fact that Pamicell appears in both is, by itself, the strongest analytical signal in this cohort.

7. Two Honest Caveats

7.1 All Four Names Have Recently Run

VM: +451% from 52-week low; 20D +42.2%
Pamicell: +105% from 52-week low; 20D +36.7%
Silicon2: 20D +15.1% (relatively stable)
Samyang Foods: 20D +7.1% (relatively stable)

VM and Pamicell are up 35–42% in 20 sessions. The screen flagged all four at the screening moment, but at already-elevated prices, “full conviction now” is inefficient. Pullback wait or post-earnings entry is the cleaner setup.

7.2 The Screening Formula Itself Can Be Wrong

These nine conditions are not an absolute formula. Several traps:

ROE may be temporary: a strong 2025 ROE that drops in 2026 is meaningless
Consensus estimates can revise: 2026F OP-growth is a sell-side estimate, and estimates can be wrong
Flow can reverse: today’s foreign+institution net-buy can become tomorrow’s net-sell
RSI <75 is a snapshot: the next day’s print may be ≥75

The screen is “look at this name more deeply,” not “buy this name.” That four cleared means “worth a closer look,” not “guaranteed buy.”

8. Tracking Signals — Four Names + Watch List

8.1 Pamicell (Focus Candidate)

₩19,000–20,000 holding as support
1Q26 OP ≥₩11.5bn with OPM ≥30%
Same tracking variables as Series Part 3

8.2 Silicon2 (Quality Candidate)

₩41,500–43,000 holding as support
2026F OP growth ≥+20% maintained
Freight / margin pressure not materializing

8.3 VM (Aggressive)

₩52,000–55,000 pullback support
Order momentum maintained
2026F OP estimate revisions

8.4 Samyang Foods (Steady-Compound Watch)

₩1.25M–1.30M pullback for entry consideration
Overseas growth not decelerating
OPM peak-out check

8.5 Watch List

HD Hyundai Marine Engine: when P/E moves below 25×
Classys / PharmaResearch: when foreign+institution flow turns to net-buy
Samsung Electronics: when RSI cools to <75

9. The Single Closing Line

ROE ≥25%, revenue growth ≥+15%, OP growth ≥+20%, OPM ≥10%, P/E ≤25×, target headroom ≥+15%, daily turnover ≥₩3.0bn, foreign+institution net-buy, RSI <75. Stack nine simultaneous filters and only four Korean names clear: VM, Pamicell, Silicon2, Samyang Foods.

If the prior post’s 67-name earnings-explosion screen taught “most have already moved,” this 4-name screen teaches “strict filters surface what actually compounds.” Sixty-seven is too many — the market has discovered most. Four is few enough to matter.

The four share efficient capital deployment, growing earnings, and active flow. Their differences are in character: Pamicell is the focus candidate (cleared both screens, ROE 38%, OPM 36.3%, flow +₩53.7bn); Silicon2 is the quality candidate (ROE 38.7% with P/E 13.3× — best quality-vs-price); VM is the aggressive position (OP growth +243%, dominant on pure quant, but +451% off low); Samyang Foods is the steady compounder (clean numbers, but not the current momentum leader).

Pamicell surviving both screens is the strongest analytical signal in this cohort. Two independent filtering logics — earnings-explosion growth and ROE+price+flow quality — produce the same answer.

That said, all four have recently run hard. The screen result is not a “full-conviction-now” signal. Pullback wait or post-earnings entry is cleaner. The narrow-miss watch list (HD Hyundai Marine Engine, Classys, PharmaResearch, Samsung Electronics) is also worth tracking — relax one filter and any of them enters.

The screen is the start, not the destination. Whether the four names print 67 or 4, the next step is asking why each one cleared and whether that clearance is sustainable.

FAQ — Korean Quality-Compounder Screen May 2026

Q: How many Korean stocks pass a strict 9-filter ROE-25% screen? A: As of May 7, 2026: only 4 names — VM (KOSDAQ 081000), Pamicell (KOSPI 005690), Silicon2 (KOSDAQ 257720), Samyang Foods (KOSPI 003230). The criteria stack ROE ≥25%, revenue growth ≥+15%, OP growth ≥+20%, OPM ≥10%, P/E ≤25×, target headroom ≥+15%, daily turnover ≥₩3.0bn, foreign+institution net-buy, RSI <75.

Q: Why does ROE 25% matter as a screening threshold? A: ROE 25% means the company generates ₩25 of net income per ₩100 of equity per year — a strong indicator of capital-deployment efficiency. Stacked with growth (revenue / OP / margin), valuation (P/E / target), and flow (foreign+institution / RSI) filters, it isolates “good business + reasonable price + active market support” simultaneously.

Q: Why does Pamicell stand out in this cohort? A: Pamicell cleared both the previous earnings-explosion screen (67 names) and this ROE-25% screen (4 names). Two independent filtering logics produced the same answer, which is the strongest analytical signal in either screen. ROE 38%, OPM 36.3%, OP growth +72.3%, foreign+institution 20D +₩53.7bn.

Q: Is Silicon2 (257720) publicly traded? A: Yes — Silicon2 is listed on KOSDAQ under ticker 257720. It is a global K-beauty distribution platform with high overseas-revenue mix and ROE 38.7%. Its P/E 13.3× combined with ROE near 39% is unusually clean quality-vs-price math by Korean-market standards.

Q: What is VM (081000)? A: VM is a KOSDAQ-listed Korean equipment / industrial-products company that cleared the screen with the strongest pure quant score (OP growth +242.9%). However, it has run +451% off its 52-week low, putting it firmly in “already-moved” territory — closer to an aggressive position than a core hold.

Q: Why is Samyang Foods (003230) listed but flagged as steady-compounder? A: Samyang Foods (KOSPI 003230) clears all nine filters with strong fundamentals (ROE 35.3%, OPM 23.4%, P/E 18.8×). However, its 20-day return at +7.1% is the lowest in the cohort, meaning it is not currently the market’s momentum leader — making it a steady-compound portfolio slot rather than a short-term-momentum slot.

Q: Should I just buy all four screen passes? A: No. The screen is “worth a closer look,” not “guaranteed buy.” Four key caveats: ROE may be temporary, consensus estimates can revise, flow can reverse, and RSI <75 is a snapshot. Two of the four (VM and Pamicell) are up 35–42% in 20 sessions — pullback wait or post-earnings entry is cleaner than chasing.

Q: Why is the near-miss list (“HD Hyundai Marine Engine, Classys, PharmaResearch”) interesting? A: Each missed by exactly one condition. HD Hyundai Marine Engine: P/E 25.7× (0.7× miss). Classys / PharmaResearch: foreign+institution net-sell (only flow filter failed). One condition flips and they enter the active cohort. This makes them productive watch-list candidates.

This post is research and commentary only, not investment advice. The 9-filter screen formula, per-name ROE / P/E / OPM / growth figures, and flow data are sourced from local DB, sell-side consensus, and public materials and reflect analyst inference at the screening moment (2026-05-06). Subsequent price / flow / earnings changes can change results. Short-selling, lending, and program-flow data are not confirmed for the local DB. The screen is an analytical starting point, not a buy signal. Analysis can be wrong. Data as of May 6–7, 2026 KST.