Best Trading Mentors

Feedback turnaround journal

Keep returned durations, pending ages, withdrawn tasks and missing-deadline coverage separate.

Added 14 October 2026

The submitted population

A task row contains an opaque identity, submitted timestamp, optional due timestamp, state and optional returned timestamp. The as-of time bounds every actual observation. Submissions cannot occur after it, and returns must be between submission and cutoff. No accepted-work population is fetched from a provider. The tool cannot discover omitted tasks or authenticate timestamps. Its five base tasks are fictional. A real service review needs the agreed starting event and actual source record; a supplied submission time may differ from the contractual acceptance time. That distinction remains outside the arithmetic.

Returned durations and pending ages

Returned duration runs from submission to return. Pending age runs from submission to as-of. Pending work stays outside returned-only mean and median without disappearing from the trace. Withdrawn rows have no inferred returned duration or pending age. In the base packet, one-, two- and twenty-four-hour returned durations give a two-hour median and nine-hour mean. The open task has a separate fifty-one-hour age. A short median therefore does not prove every submission was handled quickly. The model makes no typical-provider or customer-satisfaction conclusion from those fictional values.

Declared deadline comparisons

An inclusive operator treats equality as on time; a strict operator does not. Pending work becomes overdue after an inclusive due time or at a strict due time. The operator is a reader-supplied convention, not an authenticated agreement. Missing due times give no deadline comparison while returned duration remains available. Those identities are listed as unclassified returned work. A due time may be after the cutoff if the source supplies it, but it cannot precede submission. The journal does not infer weekends, business hours, grace periods or a contract clock from these timestamps.

Closed and open populations stay visible

The output separately reports returned, pending, withdrawn, returned-with-due, on-time, late and overdue-pending counts. These are task counts, not independent learners or trades. A returned task without a due time cannot inflate on-time coverage. A withdrawn task cannot silently become completed work. The pending-feedback-arrives variant changes the returned population using a later authored return while preserving the old snapshot as a separate case. This prevents a summary from imputing a convenient completion time or deleting unfinished work simply to improve a response statistic.

Distribution arithmetic

Returned durations are sorted for the median. Even populations use the average of their two middle durations. The mean uses the complete returned population and six decimal places for display. UTC whole-second inputs produce second-based durations; tied submissions need no invented subsecond ordering. Zero-duration returned feedback is permitted when its supplied timestamps coincide. Empty returned populations produce null statistics, not zero waits. The tool does not evaluate the substance of feedback, urgency of the task or quality of the explanation. Those questions require the actual artifact and criterion.

A bounded timing result

Impossible dates, duplicate task identities, unknown states, conflicting returned fields and unsupported precision are rejected. Valid inputs still may be selected, misclassified or incomplete. Unobserved tasks remain null, and learning improvement, strategy validation and investment suitability remain null in all variants. A late response is a result under the entered due convention, not a legal finding or a fabricated complaint about a provider. A timely response is not an endorsement. Keep the applicable service terms and source coverage with the exported record before drawing a stronger conclusion.

Learning and service records