Historical clock synchronization and timestamp semantics for CLOB + RTDS

#1
by RiccioToday - opened

Hi Polydata5,

Thank you for publishing this dataset. We are assessing whether historical CLOB and RTDS events can be aligned with an independently collected Binance stream for causal offline replay. Could you clarify the following points, ideally for the 1 October 2026 BTC files and whether the answers differ across dates?

  1. Were the CLOB and RTDS streams recorded on the same physical host and wall clock? What do recv_ts, recv_mono, conn_id, frame_i and hour_start mean? In the inspected 00:00–01:10 UTC sample, recv_mono is zero: is this an unavailable field or an export placeholder?

  2. Where is recv_ts assigned: socket arrival, WebSocket callback, after JSON decoding, or after a queue/write step? Could you share the collector source or the exact code/version used for these historical files? Were timestamps ever shifted, normalized or reconstructed during export?

  3. What synchronized the collector clock (NTP/chrony/PTP/provider clock)? Are historical offset/error bounds, clock-step/drift logs, host/run identifiers or local-send/local-receive timestamped server-time probes available? Timestamp precision alone would not establish clock accuracy; one-way receive-minus-publisher differences do not separate clock offset from network/queue delay.

  4. Does an event="connect" marker occur before the subscription request is sent? For the first CLOB book and RTDS subscription response, is the outer timestamp generated after that particular request, or can it describe cached data? Please distinguish the RTDS outer publisher timestamp from the inner oracle observation timestamp. This determines whether those messages can safely bound clock offset.

  5. Is native Binance trade data (including both E and T and local receive timestamps) available from the same collector host, even if not currently published?

  6. For closed historical dates before 23 October 2026, are initial market metadata snapshots available with their acquisition timestamps: condition/token/outcome mapping, opening reference price and its source, tick size and changes, minimum order size, fee parameters and resolution rules? Later metadata alone cannot establish what was available at entry time.

If some evidence was not recorded, an explicit statement of what is unavailable would also help. Which historical dates offer the most complete CLOB + Chainlink spot + TWAP60 coverage with the above provenance? No paid service or purchase is requested.

Thanks!

Sign up or log in to comment