Exchange Operational Announcements and Public Transfer Status: A 180-Day Labeled Dataset and a Four-Venue Polling Study

Operational announcements are often used as if they were synchronized records of market access. This paper separates announcement evidence from observed transfer status in two public, redacted datasets. Dataset A contains 996 labels across a 20-venue target map for 2026-04-23 to 2026-10-20, with explicit coverage gaps and separate publication and effective-time fields. Dataset B audits 502 polling rounds at four venues over a 169.1-hour log span. Among 929 auditable non-Gate key-transition records, 503 had no contemporaneous semantic announcement match; this record-level proportion is concentrated at LBank and is not an independent-event rate. For 81 matched announcement events, median detection-minus-publication lag is +5.81 minutes, with venue-specific signs and sensitivity bounds. The median of 517 continuously observed closed withdrawal intervals is 0.648 hours, but batch exclusions move it materially. The release supplies explicit estimands, gap-aware duration partitions, reproducible tables and figures, and evidence boundaries for research on operational information without claiming execution, profitability, or market-impact effects.