Forty future ASASSN-14ko peaks with 3-day errors give less than 25% power against delays with SD below 10 days for two fixed kick fits after cubic trend removal

We ask whether future flare times can separate accumulated period changes from a common emission delay in ASASSN-14ko. We use two fixed kick fits from prior peak times. After a free cubic clock is removed, forty consecutive future peaks with 3-day independent errors have power at most 0.2404 or 0.1945 at test size 0.05 against delay families that include selected historic fits with SD below 10 days. The bounds follow from Gaussian KL divergence and Pinsker's inequality. They apply to any test of these future-only timing contrasts with size controlled for the allowed family, not only to our likelihood-ratio test. They do not cover joint use of historical times or external orbital markers. We also give uniform contrast tests under an assumed delay SD cap. A sharp stationary spectral variance bound does not require a memory law. Their power tends to one under continued kicks as the span increases at fixed peak count, but can remain low over decades. We supply fixed-pair power, an explicit frequency-grid bound, code, aggregate inputs, contrast vectors, proof and validation. The kick parameters are treated as known; the physical delay cap is not measured. No new observations or physical clock identification are claimed.