Japan tech & semiconductors

Claim

This sweep is preserved mainly because it corrected an error in the repo's own prior research, and the correction generalises.

The FX artifact

The 7/29 global screen flagged Japan as discounted on the strength of EWJ −8.4%. But the yen fell 8.41% over the same window, and the TOPIX set a record close of 4,101.96 on July 6. The entire "discount" was the currency.

EWJ is unhedged, so it measures the yen as much as the market. For a dollar investor wanting Japanese equity exposure rather than a short-yen position, DXJ (yen-hedged) is the honest instrument. Any international screen using unhedged country ETFs as a value signal is measuring FX.

The governance thesis is name-specific, not country-specific

Japan's corporate-governance/buyback story is real but does not travel as a country trade: buyback + dividend yield ranged 0.03% to 4.48% within a single field. A 144-fold spread inside one sector means the country-level framing carries no information.

No AI discount

Advantest at 60x P/E and 28x P/B. Whatever was cheap in Japan, it was not the AI complex — same conclusion as the US ([[pattern-ai-levered-fields-trade-above-own-band]]), reached independently.

⚠️ Method note — why this sweep happened at all

The Japan detour was a misstep: the field was screened before US coverage was exhausted, on the strength of the FX artifact above. The user challenged it directly. Recorded because the sequencing error is more reusable than the field result: exhaust the domestic universe before adding currency risk to the search.

Ticker traps hit here

SCRYY → SCOR SE (not Screen Holdings) · CPCAY → Cathay Pacific (not Capcom). See [[pitfall-adr-ticker-resolution]].

What would falsify this

A yen move large enough that hedged and unhedged Japan diverge in the opposite direction, or governance reform producing sector-wide rather than name-specific capital returns.

History

• 2026-07-29 — swept; the prior global screen's Japan flag corrected as an FX artifact.