A fixed Student t shape gave no clear weekly Nasdaq distribution score gain over normal at equal variance in 2016–2025

We tested whether a Student t distribution with five degrees of freedom improved full weekly Nasdaq Composite return forecasts over a normal distribution with the same changing variance. The forecasts used dated ALFRED inputs and a fixed weekly schedule. The main test had 497 non-overlapping weeks with fresh inputs in 2016–2025. Student t reduced mean continuous ranked probability score by 0.280 return basis points, or 0.19%. The 95% paired eight-week block interval for t minus normal was -1.126 to +0.509 basis points. Density and 5% tail scores also gave no clear primary-model gain. Omitting 2020 left almost no CRPS gain. The nominal 90% t interval covered 87.73% of outcomes, compared with 89.94% for normal. At equal variance, t5 has a narrower 90% interval despite its heavier far tails. The package preserves all seven fixed model paths, 4,011 forecasts, aggregate scores, source hashes, code, figures, and repeat instructions. Direct outcomes and resampling draws are retained separately under source terms. This is one historical index sample, not a new market replication or a trading-profit test.

return distribution · Student t · CRPS · point-in-time data · weekly forecast · Nasdaq Composite · negative result