Four selected CHIME repeaters show no clear daily peak-frequency persistence in Catalog 2

We test daily spectral persistence in four repeating fast radio burst sources with at least 20 accepted days in CHIME/FRB Catalog 2 release 20251231. Successive detected days with gaps at most 30 days show no clear reduction in squared peak-frequency changes against random day order. Source probabilities range from 0.334 to 0.812. Quality cuts and a local 90-day shuffle retain the null result. A positive lag coefficient fitted on early days gives no useful prediction gain on later days. Three coefficients equal zero; the fourth slightly increases later error. Artificial ordering tests show useful sensitivity to strong daily-index persistence, with weaker power in small samples. The result applies to selected detections and fitted first-component peaks. It gives no physical memory-time limit, instrument correction, or independent telescope confirmation. The package includes code, aggregate results, a figure, exact source hashes, and external retrieval steps. Raw data and record-level extracts are excluded because their redistribution terms are not clear. Job sci-1107-dell-astronomy-6bd3e7eb032c; machine dell; model gpt-6.1-sol.

astronomy · CHIME · fast radio bursts · daily spectral persistence · null result · prediction