What countries apart from the US is a quant model selecting — and why
We run a quantitative stock selection model that ranks ~800 tickers by risk-adjusted performance across multiple time windows and outputs a concentrated portfolio of 15. Right now, three of those 15 are non-US positions: ASML (Netherlands), EWY (South Korea), and TSM (Taiwan). Here's what the model sees.
Taiwan Semiconductor (TSM) — Taiwan
The model has covered this one before but it keeps ranking. The quant reasoning isn't "AI is big" — it's about structural position. TSM manufactures at advanced nodes for everyone: NVIDIA, Apple, AMD, Qualcomm. You don't pick a winner in the AI arms race; you own the factory that arms both sides.
What makes it score well risk-adjusted rather than just high-return: TSM has low idiosyncratic volatility relative to its returns. The market treats it like high-beta semiconductor but its revenue behaves more like a utility with pricing power. That compression between perceived risk and actual risk-adjusted return is where the model finds alpha.
ASML (ASML) — Netherlands
This one gets less attention than it deserves outside quant circles.
ASML makes the EUV lithography machines that physically print transistors onto silicon. 100% of EUV systems shipped globally come from one company, in Veldhoven, Netherlands. It took 30 years and ~$10B in R&D to build, depends on sole-source suppliers (Carl Zeiss for optics, Cymer for lasers), and has 16,000+ active patents. TSMC, Samsung, and Intel cannot manufacture sub-7nm chips without ASML hardware. There is no alternative supplier.
The numbers: Q2 2026 revenue €9.3B, gross margin 54%, Q3 guided at €11-12B with 55-57% margins. Year-end backlog: €38.8B — nearly one full year of revenue already contracted before the year started.
EWY (iShares MSCI South Korea ETF) — South Korea
The model selected an ETF, not a single stock. Here's why that's interesting.
EWY's top two holdings are Samsung (21%) and SK Hynix (20%) — together 41% of the fund. In practice, you're primarily buying Korean memory and logic semiconductors wrapped in ETF structure. But the ETF selection over single stocks is deliberate from a risk-adjusted perspective: Korean individual stocks have wide bid-ask spreads for US investors, single-name execution risk, and concentrated chaebol exposure. EWY gives correlated AI upside with partially diversified downside.
The HBM thesis: SK Hynix holds ~62% of the global High Bandwidth Memory market. HBM is the performance-critical memory stack that sits directly on AI accelerators — without it, you cannot build a competitive GPU. SK Hynix's Q1 2026 operating margin: 72%. They were first to mass-produce HBM3E, hold the majority of NVIDIA's supply agreements, and are already shipping HBM4.
The governance catalyst: Korea has been running a Corporate Value-Up Program since 2024, following Japan's playbook. March 2026 saw formal Commercial Code amendments mandating fiduciary duty to minority shareholders, treasury share cancellation requirements, and tighter audit independence. The "Korea discount" — historically 30-50% P/E gap to global peers driven by chaebol governance concerns — is beginning to compress. Multiple expansion on top of an earnings supercycle is what EWY's +74% YTD in 2026 reflects.
The model entered EWY in February 2026, right as the governance reform implementation was accelerating and HBM demand confirmation was coming through NVIDIA's B200 supply agreements.
What these three have in common
None of these are speculative bets on future growth. They're monopoly or near-monopoly positions in the AI infrastructure supply chain — the layer that everyone in the chain depends on. ASML makes the machines that make the chips. TSM makes the chips. SK Hynix (via EWY) makes the memory that goes on the chips.
The model doesn't read earnings calls. It sees Sharpe ratios, revenue stability, and momentum signals. These three have all of that — plus fundamental moats that explain why the signal has been persistent.
Full 15-stock portfolio and methodology at Quantin. Happy to answer questions on the selection process in the comments.
Walk-forward validated since 2018. Not financial advice.