Ideas
Small/mid caps lead, not large caps
The index itself is less important than the active small/mid-cap and KOSDAQ tape. He would trade selective leading names in small/mid caps rather than chase large-cap index moves, while keeping position count small.
KOSPI rally has already started
The KOSPI rally has already started: BOJ, FOMC, AI speed-control worries, and US options expiration are already known, and the market held up. Foreign selling appears to be turning, and the historical pre-Chuseok weakness is a contrarian setup, so he expects a move toward 7,500.
Kolmar and Cosmecca lead cosmetics ODM
Within cosmetics, he sees Kolmar and Cosmecca as ODM leaders that held up better and rose earlier than APR and Silicon2. He remains positive on these relative leaders, though the broader cosmetics sector is crowded and may need time to rotate back.
Optical communication sector is crowded
Optical communication had a sector-wide spike including laggard names, which historically marks short-term crowding. It needs fresh money to keep going, so he would sell optical communication and rotate to secondary batteries or nuclear.
Secondary batteries support AI power efficiency
Secondary batteries are needed to improve power efficiency while nuclear plants take years to build, and they are now aligned with semiconductors in the AI data-center supply chain rather than competing for the same flows. He flags them as a rotation candidate while they rest.
AI data center power is must-own
AI data centers are now the priority theme over semiconductors alone. The coming bottleneck is power, so the whole AI data-center power bundle—equipment, transformers, batteries, nuclear, construction, and wires—should be represented in a portfolio.
US gas power fills immediate power gap
In the US, gas power generation is the fastest way to meet data-center electricity demand, so gas power utilities and generators should benefit as the immediate power solution.
Construction benefits from data center buildout
Construction companies benefit whether data centers or nuclear plants are built, and they have recently lagged. He suggests buying Korean construction names on weakness as part of the AI data-center power buildout.
Korean biotech lacks momentum now
Biotech has lost momentum and leading investors are leaving; rallies are being sold to buy other sectors. A sustainable biotech turn would require semiconductor leadership to break, which he does not expect now.
Samsung and SK hynix are core
SK hynix should stage a technical rebound to around 2.1 million won, with the 2.0-2.14 million won zone as a first target. The timing may coincide with the end of buybacks, and November large-shareholder tax selling is a later risk.
Samsung and SK hynix are core
SK hynix and Samsung Electronics are unconditional core holdings for an AI portfolio. Even though he holds less memory than in the first half, these two remain the fixed base for Korean semiconductor exposure.
Semiconductor equipment and back-end favored
Within Korean semiconductor equipment/materials, equipment is preferred over materials, and back-end packaging/OSAT is better than front-end. AI demand and capacity expansion confirm that equipment, materials, and parts will sell, and the rally is broadening beyond memory.
Nuclear power benefits from power bottleneck
Nuclear power is a must-have for the AI data-center power bottleneck expected around 2027. Westinghouse is now aligned through equity investment rather than blocking Korean projects, and any sell-on-news from a US investment announcement should be bought.
Doosan Enerbility can win both models
Doosan Enerbility can receive orders under both the Westinghouse model and the Team Korea model, making it a key nuclear beneficiary. The Westinghouse equity investment reduces the historical intellectual-property conflict risk for Korean nuclear projects.
Take profits in electric wires
Electric wires rallied strongly today, so he would take profits and rotate into lagging nuclear and construction names. The long-term power buildout still needs wires, but near-term crowding favors a tactical sell.
Sanil Electric leads transformer market entry
Sanil Electric is entering the large transformer market despite strong incumbents, and its share price is already stronger and above moving averages. If the transformer theme works, he sees more upside here than in the established names.
Buy KOSPI dips; target 7,500-8,000
Next week is a buy-on-dips week, especially if the market weakens Monday-Wednesday around Chuseok and US-China summit uncertainty. The first rebound target is 7,500-8,000, but he prefers a slow rise because sharp moves trigger leveraged ETF crowding and backlash.
This 3PRO TV (삼프로TV) video, published September 18, 2026,
features Hwang Yoo-hyun, Myung Min-jun
discussing KOSDAQ, Korean small/mid caps, EWY, 161890.KS, Cosmecca Korea, Korean optical communication sector, Korean secondary battery sector, AI-SECTOR, US gas power utilities, Korean construction sector, Korean biotech sector, 000660.KS, 005930.KS, Korean semiconductor equipment, Back-end packaging/OSAT, Korean nuclear power, 034020.KS, Korean electric wire sector, 062040.KS.
17 trade ideas extracted by AI with direction and confidence scoring.
Speakers:
Hwang Yoo-hyun,
Myung Min-jun
· Tickers:
KOSDAQ,
Korean small/mid caps,
EWY,
161890.KS,
Cosmecca Korea,
Korean optical communication sector,
Korean secondary battery sector,
AI-SECTOR,
US gas power utilities,
Korean construction sector,
Korean biotech sector,
000660.KS,
005930.KS,
Korean semiconductor equipment,
Back-end packaging/OSAT,
Korean nuclear power,
034020.KS,
Korean electric wire sector,
062040.KS
