ZTF data from 2022 to 2025 do not continue the sinusoids of 86 ZTF periodic-quasar candidates, and their forecasts are no better than red-noise control quasars

Chen et al. (2022, arXiv:2206.11497) reported 86 periodic-quasar candidates (periods 500-1000 d) from ZTF DR11 data (March 2018 to March 2022). Such candidates are possible signs of supermassive black hole binaries. We did an out-of-sample test with ZTF DR24 data from the SNAD mirror. For each candidate, we fitted the period and a sinusoid on the DR11 time span only. Then we scored the forecast on the later data, 2022-03-08 to 2025-10 (median 1.45 cycles). We applied the same procedure to 2,575 SDSS DR16Q control quasars matched in magnitude and redshift, and to the same controls with injected sinusoids. The median forecast correlation was 0.05 for the candidates and 0.07 for the 57 controls that pass the same in-sample periodicity cuts (KS p = 0.96). Phase hits within 45 deg were 29% for the candidates and 26% for the selected controls (chance 25%). With injected sinusoids of the candidates' fitted amplitude, the median correlation was 0.62 and phase hits were 61%. A mixture fit gives a best-fit coherent fraction of zero, with a 95% upper limit of about 6% (bootstrap 84th percentile 9%). If the true amplitudes are half the fitted values, the limit is 13-28%. None of the three candidates that Chen et al. kept after adding older surveys continued its sinusoid (correlations 0.13, -0.58 and 0.08; amplitude ratios 0.30, 0.39 and 0.11). A 180-day gap, the g band and a per-candidate damped-random-walk null give the same result. Limitations: the test window covers only about 1.5 cycles, quasi-periodic or intermittent signals are not excluded, the selected-control null is small (57), and only the 86 SPL candidates were available.

periodic quasars · supermassive black hole binaries · ZTF · out-of-sample forecast · red noise · damped random walk · Lomb-Scargle · quasar variability · SDSS DR16Q · null result