A gold candle is missing. Can the target audit still reach a verdict?

Start with the evidence window

A nominal target audit usually asks whether a level was touched during a defined period, sometimes before another level. That question requires more than finding one candle with a convenient high or low. It requires knowing what was observed throughout the relevant window and what was not. A missing interval can change an ordered outcome even when a later target touch is visible.

This is a proposed research workflow, not an audit result for a particular CyGuru signal. No new OHLC export has been acquired or evaluated for this article. The examples below are deliberately hypothetical and establish neither a target reached nor a broker fill. Their purpose is to show how a research system should preserve uncertainty instead of converting a data gap into a favourable result.

A completed bar is not a complete history

OANDA’s candle definition identifies a start time, separate bid, ask or midpoint data when requested, a price-count volume field and a complete flag. The flag means the candle’s ending time is no longer in the future. It does not, by itself, certify that a downloaded file contains every interval needed for a historical study.

That distinction is easy to miss in a data pipeline. Every retained row can be marked complete while an entire row is absent. I would therefore maintain two checks: the state of each candle and coverage across the requested window. Neither should be treated as a substitute for the other. A field name is only as strong as the provider’s actual definition.

Source notes: [1] OANDA v20 — Instrument and Candlestick Definitions

A later touch does not settle what happened earlier

Imagine an invented BUY setup with a frozen entry, stop and TP1. A visible candle later exceeds TP1, but an earlier interval between activation and that candle is missing. The later observation can support the narrow statement that the chosen price series subsequently reached the target. It cannot exclude a prior stop touch during the unobserved interval.

A careful result might therefore retain target-touch evidence while leaving TP1-before-SL unresolved. Those are different claims. Filling the gap with the last observed price would manufacture a path and could remove the very event the audit is intended to detect. A smooth-looking chart is not a reason to replace missing evidence with assumed inactivity.

Define expected coverage without inventing a session

The coverage gate should begin with the provider, instrument, interval convention and known trading schedule. An expected maintenance break is different from an unexplained missing interval while observations should be available. Do not simply demand one bar for every wall-clock minute across weekends and holidays; establish what the selected feed is supposed to contain.

My proposed record stores request boundaries, retrieval time, returned timestamps, duplicates and gaps. Keep the original response before any cleaning. When additional data are obtained, append the repair evidence and record how the verdict changes. If another provider fills a gap, disclose that substitution and assess whether its price definition is compatible. A different feed may be useful corroboration without becoming the same original evidence.

Preserve the side and the original levels

A midpoint high and an ask high answer different price questions. Before testing an event, freeze the original Entry, SL and TP values and specify which series supports each nominal comparison. Do not choose the side after seeing which one produces the desired target status. Retain the indicator’s reported events separately from any independently recomputed OHLC classification.

The final audit should distinguish confirmed nominal observations, ambiguous ordering, missing coverage and unresolved setups. It should also state that none of these labels includes broker fills, commissions, spread, slippage or P&L. The CyGuru terminal can display a reported milestone while the independent audit still needs more data. Showing both states clearly is stronger evidence practice than forcing every record into a single success-or-failure label.