Replay exists so that a configuration can be tested against the same conditions twice. In ATAS the same chart objects are used, and only the data source changes, which means a panel behaves identically in live and recorded mode when the feed configuration and the session template agree.
01 Feed configuration and tick depth
The feed determines what can be reconstructed. Depth levels, tick granularity and the historical window stored locally all belong to the feed configuration rather than to a chart. A chart cannot show more history than the feed holds, and a ladder cannot show more book levels than the feed publishes.
| Property | Constrains |
|---|---|
| Tick granularity | Footprint rows, delta accuracy, tape filters |
| Book depth | Ladder columns, cumulative size, pull detection |
| Stored history | Replay range, profile windows, scan coverage |
| Session template | Bar boundaries, composite splits, daily resets |
When two readers disagree about the same historical bar, the first comparison is tick depth: a coarser feed loses prints, and a footprint row built from fewer prints is a different row.
02 Playback controls
Playback speed determines how much of the recorded stream is delivered per unit of wall time. Speed multipliers above the tick rate of the instrument will coalesce updates; the bars that result are still correct in total volume, but their internal sequence is no longer visible. For configuration work that is usually acceptable; for studying sequence it is not.
- Step forward one bar at a time to inspect row structure.
- Step forward one print at a time to inspect sequence inside a row.
- Run at an intermediate speed to check that a filter behaves the same way at pace.
03 Session alignment and boundary effects
Replay rebuilds session boundaries from the session template, not from the recording. A recording made under one template and replayed under another will produce different daily resets, different profile splits and different composite windows. This is the single most common reason a replayed session does not match the notes taken during it.
- Record the session template name used during the original observation.
- Set the same template before replaying.
- Compare one known bar — volume and delta — before drawing conclusions about the rest.
Pick a bar with a distinctive print size near a session edge. If its footprint row structure differs between live and replay, the template or the tick depth changed; nothing else in the panel needs to be examined until those two match.
04 Storage, retention and volume
Recorded ticks are the largest files an order-flow desk keeps. The storage question is therefore a retention policy question: how many sessions must remain instantly available, and how many can be archived. Depth snapshots are considerably larger than trades, so a book-recording policy multiplies storage requirements quickly.
05 Reproducing a documented observation
A replayed observation is reproducible when five items are recorded: instrument, date range, session template, bar rule and feed depth. With those, a second reader can arrive at the same picture; without them, the observation is an anecdote. The template that produced the panel should be listed too, since inputs are instance-scoped and cannot be recovered from a screenshot.
Storage and throughput limits are quantified on the performance and hardware notes page, and the panel geometry that replay sessions are usually reviewed in belongs to workspace layouts and templates.