Large price moves gave no clear gain in delayed Nasdaq return forecasts on 1,908 fresh sessions in 2018–2025

We tested whether large daily price moves help to predict a later Nasdaq Composite mean return. Inputs used dated ALFRED records through the previous calendar day. The target ran from the entry session close to the next session close. The main test used 1,908 sessions with a fresh previous close in 2018–2025. A model that grouped moves by size relative to prior volatility reduced mean squared prediction error by 0.016% against the past mean. The paired 95% interval was -0.230% to +0.248%. Fixed 2% groups and a linear control also gave no clear gain. After 69 large negative scaled moves, the delayed mean log return was -43.52 basis points, with interval -116.97 to +17.10. The test gives no clear evidence of a delayed rebound or a reliable mean-return advantage. The package preserves 2,513 forecasts, 120 monthly fit states, all 15,000 paired block draws, source requests and hashes, code, figures, checks, and repeat instructions. Limits include one index, sparse events, a specified input delay, and archive gaps. Source prices and individual observed labels remain external under source terms. No portfolio profit is claimed.

stock return prediction · short-term reversal · point-in-time data · Nasdaq Composite · chronological test · negative result · volatility scaling