Matched MAXI and BAT data show lower daily flux correlation in 2023–2026 than in 2012–2023 for 4U 1820-30

We compare public MAXI 2–20 keV and Swift BAT 15–50 keV daily data for 4U 1820-30. We match calendar days, apply quality cuts, and give each accepted day equal weight. The Pearson correlation is 0.579 on 3,095 earlier days, from MJD 55931 to 60157. It is 0.177 on 529 later days, from MJD 60158 to 61333. The difference is −0.402. For 56-day pair blocks, its conditional 95% range is −0.654 to −0.198. The range stays negative at 28, 84 and 167.43 days. Subtraction of listed error variances retains the contrast under independent zero-mean noise assumptions. However, a comparison with only the last seven earlier periods is less clear. We also fit two Fourier harmonics and a trend at a fixed period of 167.43 days. The later minus earlier relative first-harmonic phase change is −0.069 cycle. Its 56-day residual-sign range is −0.268 to +0.071 cycle. All four phase ranges include zero. The ranges are conditional on block models, with 6,400 draws per length and sample. They do not establish coverage for all source changes. Larger later BAT errors, gaps, native flux calibration and a nonlinear flux relation limit physical inference. The result does not establish intrinsic loss of coupling or place a change at the sample boundary. The contribution is a fixed matched-day table and a reproducible correlation and phase comparison. The package contains source data, hashes, code, all draws, figures, checks and repeat instructions.

4U 1820-30 · MAXI · Swift BAT · X-ray binaries · daily flux correlation · relative phase · block resampling · public astronomical data