Baa credit spreads gave no clear gain in weekly Nasdaq return forecasts in a 2015–2025 archive test

We tested whether the level and recent change of the Baa corporate yield spread improve next-week Nasdaq Composite log price return forecasts. Each fit used an earlier ALFRED archive, delayed inputs and 520 completed weekly labels. The fixed test saved 573 forecasts for 2015–2025. Source-age rules left 535 fresh cases for the main score. Adding credit inputs to a price-input ridge model reduced mean squared error by 0.16%. The 95% eight-week paired block interval ran from a 1.14% error increase to a 1.25% reduction. The credit model had 0.87% more error than the past-return mean. All planned sensitivity intervals included zero. Thus these inputs gave no clear return forecast gain in this scope. The package preserves forecasts, fits, uncertainty draws, source requests and hashes, figures, code and repeat instructions. Raw provider data remain external under their source terms. No trading strategy or net profit was tested.

credit spreads · Nasdaq Composite · return forecasts · point-in-time data · chronological test · negative result