The Treasury yield spread gave no clear gain in 131 monthly Nasdaq return forecasts with dated inputs in 2015–2025

We tested whether the recent 10-year minus 3-month Treasury yield spread improves next-month Nasdaq Composite log price return forecasts. We used dated ALFRED archives, fixed monthly fits, and 131 nonoverlapping outcomes from January 2015 to November 2025. A ridge model with 120 earlier monthly targets had forecast skill of -0.04% versus the past mean. The 95% paired six-month block interval was -1.91% to +2.20%. Adding the twelve-month spread change gave skill of -1.63%, with an interval of -4.31% to +1.69%. The primary spread forecast was always positive, so its 70.23% sign accuracy matched always-up. Removing four stale price-input months did not establish a gain. An independent reconstruction checked all 15,720 training rows. Four future-source removal checks passed. A full repeat matched eight core output files byte for byte. The package preserves code, forecasts, fit states, source requests and hashes, all planned cases, uncertainty draws, figures and negative findings. This is one historical price-index test, with no trading strategy or net-return claim. Small gains remain possible. The exact Treasury archive is retained under its source rights; raw provider rows are excluded from the public package.

return prediction · Treasury yield spread · point-in-time data · Nasdaq Composite · chronological evaluation · negative result