Stationary Gaussian emission delays fit ASASSN-14ko peak times and approach the period-kick law after trend removal

We test whether correlated flare times can identify random orbital-period kicks in ASASSN-14ko. A stationary Gaussian delay with a double-root autoregressive response approaches the exact discrete kick law on any finite cycle grid after the unknown epoch and period are removed. The result allows gaps and additional polynomial clock terms. The delay standard deviation grows as the memory to the power 3/2 at fixed kick scale. We compare both laws using 30 prior peak times with two peak definitions and quadratic or cubic clock trends. The best tested stationary delays have standard deviations of 3.19-4.00 days and raw likelihoods at least as high as the kick model, with one extra memory parameter. A 3-day SD cap also retains that fit. At 40-cycle memory, fixed delay and kick laws have total-variation upper bounds below 0.033, but require much larger delay amplitudes. We provide likelihood profiles, amplitude constraints and distribution bounds. This is a reanalysis and a scoped application of an established unit-root limit. It adds no observations and does not show that a positive physical emission delay follows this model. Independent limits on delay amplitude or memory, or a pericentre marker, are needed to assign timing noise to orbital kicks.