Past yen gains gave no clear added monthly Nasdaq return forecast gain in a 2015–2025 archive test

We tested whether past yen gains improve monthly Nasdaq Composite price-return forecasts after the past index return is known. Each monthly fit used an earlier ALFRED version and 120 complete past targets. Both target closes came after the forecast origin. The main test had 128 fresh-input months from January 2015 to November 2025. Adding yen increased squared error by 2.22% against the model with only the index input. The 95% paired six-month block interval for error reduction was -6.73% to +1.10%. The model with yen had 5.81% more error than the matched past mean. Fixed block lengths, all 131 months and removal of 2020 gave no clear added gain. This test uses delayed weekly currency releases and fixed linear models. It does not reject small or more timely effects. No trading strategy was tested. The package preserves all forecasts, fit states, source records, 360,000 uncertainty estimates, code, figures and repeat instructions. Restricted source-derived rows are retained separately. A full repeat matched eight core result files byte for byte. This is a monthly extension on a reused market period, not an independent market sample.

return prediction · Japanese yen · Nasdaq Composite · point-in-time data · chronological evaluation · negative result