A block test with quoted BAT errors does not reject a common cycle phase for 2S 0114+650 in 2012–2025

We test the phase of the known cycle near 30.8 days in the neutron star binary 2S 0114+650. We use 4,635 accepted daily Swift/BAT rates from 2012 through 2025. The null model has one phase and seven nonnegative amplitudes. Both models fit period, annual means, changing binary profiles, and changing second cycle harmonics. Separate phases improve weighted squared error by 45.72. A restricted residual bootstrap with 128-day sign blocks gives p = 0.311 from 1,999 draws. The common-phase period is 30.72174 days. This test does not reject a common phase at the 5% level. Its calibration assumes approximate sign symmetry of each block's joint errors and independence between blocks. It does not prove a stable physical clock. An added-variance control gives p = 0.048 from 999 draws. This result is near the 5% boundary and limits inference across error rules. All planned controls are reported. A separate MAXI 4–10 keV test gives p = 0.796. Imposed fundamental phase steps from 2020 measure conditional sensitivity for the actual sampling and observed residual noise. We derive calendar intervals from MJD, including 33 BAT rows with following-year labels. Exact source tables, data hashes, selections, code, fitted coefficients, simulations, numerical checks, and figures are preserved.

2S 0114+650 · Swift BAT · MAXI · superorbital cycle · phase coherence · block wild bootstrap · reproducible astronomy