Troubleshoot missing, stale or disconnected market data

Identify the exact observation you expected

Begin with the affected page and tool, exact symbol or dated futures contract, selected session, trading date and minute. A chart panned to history can look frozen while its current feed is healthy. A different futures delivery contract or SPX option-root selection can also explain a mismatch. Record the expected observation and when you expected it, including time zone. This gives the investigation a concrete target rather than comparing two screens that happen to use similar market names but request different instruments or time windows.

Check selection and filters before refreshing

Review expiry, strike, premium, size, side and trade-type filters in the affected tool. An active market can legitimately produce no rows for a restrictive selection. Clear the filters that are not essential to the test and keep the instrument, session and date fixed. For a chart, inspect interval and visible range; for a tape, inspect the filter and pagination context; for a surface, inspect expiry and source availability. Do not change every control simultaneously, because then a recovered display cannot identify which original selection caused the apparent gap.

Read freshness, coverage and source states

Inspect available timestamps, source, delayed flags, coverage, pending-history and access messages. Missing, stale, delayed, partial and not configured are different states. An old quote can remain old even though the browser successfully loaded the application shell. Partial capture means totals describe admitted observations, not an inferred complete session. History building means preparation is in progress for a bounded selection. A provider or plan restriction needs its own remedy; a successful HTTP response or a populated neighboring widget does not establish fresh inputs for this exact tool.

Use a controlled refresh sequence

Use the page's Refresh data action once when available and wait for its actual result. If the view is paused or away from the live edge, use the relevant Live, reset or follow control deliberately. Keep a screenshot of the original status when it will matter for support. A reload can change the evidence you were investigating, and rapidly refreshing several views can obscure whether history preparation completed or the current selection changed. Return to the same coordinates after refresh before deciding that the original issue is resolved.

1. Confirm instrument, contract, session, date, minute and visible range.
2. Clear unintended filters and read the source and coverage messages.
3. Refresh once, wait for its result and compare the same selection again.

Separate reconnection from historical recovery

A reconnect can restore a synchronized current book without filling earlier missing executions or depth. The resumed picture is not proof that the gap before it was repaired. Futures depth, executions, BBO and other data layers can have different captured coverage. A historical rebuild or recovery action must report its own progress and result for the requested selection. Do not infer complete earlier history from a healthy current connection, and do not replace a documented gap with a remembered value or synthetic interpolation when reporting the problem.

Recognize display failures separately from source failures

Some visual tools have a fallback, such as a 2D surface when 3D graphics cannot be used. That is a rendering choice rather than automatic evidence that the market data is absent. Conversely, a beautifully rendered retained frame can still be stale. Note whether labels and surrounding controls load, whether the plot layer is missing, and whether the source reports current observations. If exporting a chart image fails because the graphics context is unavailable, preserve that specific error instead of describing it only as a disconnected provider.

Create a reproducible data ticket

If the problem remains, include the exact selection, earliest and latest relevant times, filter state, visible source and coverage, safe error text and a screenshot with axes and status. Explain whether it affects one tool, one instrument or several independent views, and what your controlled refresh changed. For API comparisons, add request IDs and safe parameters. Keep credentials and unrelated private windows out of captures. A provider subscription or configured endpoint is not verification of this missing interval; the report should distinguish what was observed from what you expected the service to supply.