Rebuild missing options flow

Request a repair for a known partial session

Net Drift can offer Rebuild missing flow when coverage is partial. This is an explicit bounded recovery request for eligible raw options tape, not a general refresh button or a way to estimate missing flow from candles. Select the intended underlying, trading date and session before requesting it. The raw recovery scope is not narrowed by your displayed premium, size or aggressor filters.

Run one controlled recovery

Keep the selection stable long enough to see the accepted job's status.

1. Open the affected Net Drift view and inspect its reported gaps.
2. Confirm underlying, effective trading date and RTH/GTH/Combined selection.
3. Choose Rebuild missing flow once when offered.
4. Watch the normal session status while the page remains visible.
5. After completion or a meaningful status change, refresh the chart.
6. Compare reported gaps and observed bounds before deciding another request is needed.

Bounded work and partial results

Recovery stops at the latest completed minute and respects historical/overnight access, account request limits and job capacity. The browser watches status for a limited period; backend work can continue after that local wait ends. Completed windows can persist for resume, while failed or unavailable provider windows remain partial. RTH and GTH are not substitutes for each other, and Combined excludes unopened segments and known intermissions rather than filling them with false prints.

Worked example: overnight complete, daytime incomplete

Suppose a GTH repair finishes and the overnight drift becomes continuous, while RTH still reports several gaps. The completed GTH window does not certify RTH. Select the affected regular-session context and inspect its exact gaps before a separate eligible request. Filtered drift totals may change after repair because previously missing actual prints were admitted. That is correction of the historical aggregate, not a burst of new live trading at the old timestamps.

Troubleshoot retries and apparent inactivity

Do not repeatedly press recovery to force progress; accepted work is bounded and duplicate targets need not create new jobs. A rejected quota/capacity request is different from a running job. If the local waiting message ends, inspect current status before retrying. Price bars cannot serve as a fallback for missing option executions. Include selection, accepted/request time, gap intervals and final status in a ticket rather than claiming repair succeeded because the button returned normally.

Requested range and observed coverage are separate

A requested start/end window describes what you asked for. Observed bounds, gaps and completeness describe which eligible retained inputs actually exist. Retained futures executions, BBO and captured depth cover the current and previous futures session; standard option, chain and reference events have seven-day retention. Individual tools can impose narrower query, event, row, byte or preparation limits. Selecting a wider window does not cause every missing exchange observation to be downloaded automatically.

A running feed or a successful current snapshot proves neither an earlier session prefix nor continuous historical depth. Corrections are resolved before time filters, so a corrected timestamp can move an event into or out of a selected range. Incomplete capture must stay partial even if the visible line looks continuous at the chosen scale. Inspect source and coverage metadata before relying on totals. When asking for help, provide requested bounds, observed bounds, the exact dated instrument and the gap or limit message rather than only the page's selected date.

Judge recovery by accepted coverage, not by a successful button

An accepted repair request starts bounded work; it does not instantly certify every target window. Watch whether observed bounds and individual gap intervals change. Completed windows can remain useful even if another window fails, so a partial result is not necessarily all-or-nothing. Duplicate identities must not add volume twice during resume. Compare before/after coverage and selected totals together so a larger cumulative drift can be explained by repaired actual prints rather than mistaken for a new live burst.

The browser watches ordinary session status after acceptance and stops its local wait after a limited period. Those status requests are observations of the existing job, not new recovery requests. If the wait ends, inspect status before pressing retry; work may still be running. An overnight repair can involve date boundaries different from the display trading date, but the user selection should remain the effective underlying/date/session. Do not attempt to fill missing provider windows with interpolated candles or copy GTH success into RTH completeness. Each accepted source window needs its own evidence.