Keeping detector state across gaps up to eight days adds six BAT rises with MAXI matches in SMC X-1 data from 2012 to 2026

We test a causal rise detector for SMC X-1 with frozen public Swift/BAT daily data from 2012 to 2026. A valid mean uses only current and past data. The initial detector resets at each invalid mean. A later test keeps its state when valid means are no more than eight days apart. This rule retains all 81 original detections and adds eight. Six additions have an independent MAXI 2–20 keV threshold rise within ten days. Those matches occur from 2015 to 2025. Two additions have no MAXI match and remain in the event record. An independent reconstruction gives the same crossings. Removing future data at 55 dates leaves earlier predictions unchanged. A controlled two-day dropout makes the reset rule lose one of four rises from a fixed 50-day signal; the state rule retains all four. The forecast benchmark compares a mean interval trained before 2018 with the last measured interval. Their timing RMSE values are 27.39 and 38.80 days for 29 original pairs, and 13.39 and 16.39 days for 35 pairs with state retained. The samples and training means differ. The later detector test follows inspection of the original results. Resampling ranges condition on each detector and sample. This work supplies code, frozen inputs, source hashes, figures, and full event and forecast records. It shows that a data-gap cut alone does not ensure a complete rise list. It does not establish a complete physical cycle clock or a general forecast advantage.

SMC X-1 · Swift BAT · MAXI · superorbital cycle · causal timing · data gaps · forecast benchmark · reproducible astronomy