Before paying to copy an MQL5 or Myfxbook signal, turn its presentation into questions that a file can answer. A smooth curve, a recent streak and a win percentage summarise different things. None shows, by itself, how much risk produced that history. Start by requesting exported trades, the dates they cover and a curve that includes open positions. That lets you separate observations from the provider's account of events and from questions that still lack data. The purpose is to understand the evidence behind the signal, including its gaps.
A win rate does not describe the size of losses
Declared · Imagine a signal advertising a 60% win rate and little drawdown. This is a declared example, not a measurement of MQL5, Myfxbook or any particular account. The percentage counts trades that ended positive; it does not compare the size of their outcomes. Many small exits can coexist with large losses. You need the full distribution, trading costs and the time that positions which eventually lost remained open.
The definition of a trade also matters. A group of entries into the same market move can increase the count without adding independent observations. Consider those entries as episodes and ask whether they share exits, exposure and direction. A screenshot of the aggregate percentage cannot answer that. Keep the original export so position changes and sequences remain visible when reviewing the history with someone else.
The coin: a small, explicit reference model
Declared · Assume 100 independent trades, each with a 50% chance of a win, like a fair coin. Reaching a win rate of at least 58% has an approximate probability of 6.68%, rounded to 7%, using a normal approximation with continuity correction. The correction moves the boundary by half a win because the count is discrete. This is the probability of observing that threshold or more within the model; it is not the probability that a particular signal is explained by chance.
The calculation fixes the sample size and threshold before looking at outcomes. If you pick a signal after browsing a catalogue, you also select among many histories. The highlights may include extremes that appear through selection. A coin also ignores the amount lost when a trade fails: this example concerns the count of wins, not the financial result of copying a strategy.
The number of trades depends on the question
There is no universal minimum trade count that makes a signal's evidence sufficient. Define the difference from the coin you want to detect, the uncertainty you can tolerate and how previous attempts enter the analysis. Then consider dependence, changes in position size and the history's duration. Many trades concentrated in the same market move contain less information than their raw count suggests. The model's assumptions matter as much as its sample.
A history covering different periods helps reveal behaviour that a favourable window hides. Separate the period used to choose the signal from a later period that did not influence that choice. Keep dates of system changes too. If the provider changes the rules after each setback, the combined record mixes different decisions instead of describing a stable experiment. Extra rows alone do not resolve that problem.
Little drawdown: request a curve that includes open positions
A balance curve records closed outcomes; an equity curve includes the value of open positions when available. A system can close small positive trades while keeping a losing position open for a long time. A shallow balance drawdown therefore does not describe all exposure. Check which curve Rigor received and whether its dates cover the trades. The labels attached to those curves are part of understanding the result.
With closed trades alone, floating drawdown within them remains Not measured. A figure printed in the platform summary can remain Declared if the file cannot reconstruct it. Deposits and withdrawals need their own treatment: incoming capital changes the balance but is not a trade. Also ask which positions were still open at the end of the export and whether the report describes that gap.
What Rigor reviews in a signal's file
With suitable history, Rigor looks for patterns such as increasing size after losses, adding entries to average down a losing position and concentrating results in a few trades. It reviews large losses relative to the typical loss, simultaneous exposure, cash movements and positions left open at the end. These are findings about supplied data; their absence does not establish that risk is absent outside the file or outside the period it covers.
When usable trades and costs are available, the report shows sensitivity to higher trading costs. It also separates file-derived metrics, Measured, supplied information, Declared, and questions that could not be measured, Not measured. That separation helps formulate the next question for the provider. Rigor does not enter the provider's account, copy orders or decide whether you should subscribe to the signal. Its scope is the uploaded evidence.
Prepare a review someone else can repeat
Request the complete export, identify whether its documentation describes a live or demo account and record the requested interval. Keep provider explanations, itemised fees and curve limitations separately. The source account's history does not describe your own executions: latency, sizes and costs can differ. The free report lets you start with questions the file can answer and keep unanswered questions visible, with a clear distinction between an observation and an assumption.
FAQ
Is a ranking screenshot enough?
It cannot reconstruct trades or open losses. Request the export and keep the file's dates and limitations alongside it.
Does the report decide whether I should copy?
No. It describes evidence in the uploaded history and questions that remain open. That decision is outside the audit's scope.