From Target Touch to Evidence: Build a Gold Signal Lifecycle Ledger

Begin with a frozen proposal

Give every signal a stable identifier before evaluating it. Preserve the original publication time, receipt time, instrument and feed, timeframe, direction, entry rule, invalidation, target levels and expiry rule. If a signal depends on a candle close, record which candle and when that close became observable.

Store subsequent changes as new events rather than replacing the original message. Moving a target after price has approached it creates a different rule set. A ledger should let a reader distinguish the original proposal, a documented amendment and the final outcome without relying on the author’s memory.

Use an explicit sequence of states

A practical sequence is received, eligible, triggered, entered if execution is documented, partially closed, closed, expired or cancelled. A separate evidence field should identify whether a state comes from a provider label, a price-data rule or an authenticated execution record. These are different levels of support.

This is a proposed research design, not a claim that CY66 currently implements every field. A supplied card marked successful should initially be recorded as a provider-reported status. Independent verification requires the original rule, the relevant price history and the agreed evaluation procedure. An account return additionally requires actual position and cost records.

A target touch is not automatically a fill

CME’s order-type explanation distinguishes price conditions from execution: limit orders constrain price but may remain unfilled, while stop orders require their specified trigger. For a gold research ledger, also identify which price series is being checked. Bid, ask, midpoint and last trade do not answer identical questions.

For a hypothetical short position, a chart low below a target does not by itself establish the price at which the short could be bought back. Record the applicable execution side and any spread assumption. If only candles are available, a bar that reaches both stop and target may leave their order unresolved. Mark that case ambiguous or apply a conservative rule chosen before testing.

Source notes: [1] CME Group — Futures Order Types

Partial exits need quantities

A three-target plan does not mean the full position earned three separate target returns. The ledger needs the fraction allocated to each exit, the remaining quantity, any stop change and the treatment of fees. Without quantities, report price milestones rather than a fabricated aggregate profit.

For teaching only, imagine half an initial position exits at one risk unit and half at two. Before costs, the weighted result is 0.5 times one plus 0.5 times two, or 1.5 risk units on the original position. It is not three. This invented arithmetic illustrates bookkeeping; it does not describe a real CY66 signal or verified trading performance.

Keep the denominator visible

A selected successful example can explain a setup’s mechanics. It cannot establish a win rate. To estimate a rate, define the complete eligible population, the observation period, the outcome rule and the treatment of cancellations, missing data and unresolved cases before calculating anything.

Publish counts as well as percentages, and show losses and ambiguous cases alongside favourable examples. If several signals overlap the same gold move, disclose that dependence. Counting closely related alerts as independent experiments can make the evidence appear broader than it is. A result archive should make these relationships easy to inspect.

Make later review possible

Retain source files, timestamps, rule versions and the calculation code or worksheet. Link each outcome to the relevant evidence rather than only to a cropped victory image. Keep a correction history when data or implementation errors are found. The ability to revise transparently is part of a credible research process.

My preferred article format separates observed price milestones, provider-reported status, independently checked rule outcomes and actual execution evidence. If the last two are missing, say so. A ledger built this way can become a useful research asset over time; an isolated successful example remains an illustration with clearly bounded claims.