Before presenting a model portfolio or signal track record, establish which questions the file behind the chart can support. A record may describe a simulation, a published theoretical portfolio or an account containing trades. Those sources are not interchangeable, even when their charts look similar. A statistical review examines the supplied returns and distinguishes calculations from declarations. It does not reconstruct the investment process or turn a simulation into observed transactions. Preparation starts by defining what each period represents and which information is missing from the interpretation of the complete history.
Define the record under review
State whether the series represents a model portfolio, published signals or an account export. Describe the currency, calendar and frequency, together with dates when the process changed. If the history joins simulated and observed stages, preserve that distinction in the documentation. A change in the investment universe, rebalance rule or treatment of distributions can alter the interpretation even when the chart remains continuous.
Keep the original file and an explanation of any transformations you made before supplying it. If returns were calculated from portfolio values, explain the treatment of contributions and withdrawals. A rise caused by an external contribution is not a strategy return. Reviewing the file cannot independently discover movements that were never included in the supplied record or recover a valuation policy that was not documented.
Supply dated returns for each period
For a portfolio without a trade list, prepare dated period returns or portfolio values. The upload form accepts CSV or Excel. Use clear column headings, preserve the original frequency and distinguish period returns from cumulative figures. Explain the units, including whether returns are expressed as decimals or percentages. Do not fill gaps just to make the curve look continuous: document why they are missing.
For trades, the universal reader needs columns that make the records interpretable: entry and exit times, quantity and prices for closed trades, or time, quantity, price and direction for fills. The export guide explains column mapping. That file answers different questions from an aggregated return series. Do not assume a list of closed trades includes open positions, distributions or every expense incurred by the portfolio throughout its history.
Separate gross, net and benchmark series
Identify which return version you supply: gross or net. Keep both separately when available and list the fees, expenses and slippage assumptions applied. A net label does not measure missing costs. The declaration that a fund's returns are net of fees has a specific scope in the upload form; it does not replace the execution cost breakdown for a signal. Nor does supplying these descriptions imply an automatic comparison between gross and net versions.
Supply the benchmark as a dated CSV series under Advanced options. Document its currency, frequency and distribution treatment, and explain why it is an appropriate reference. The comparison uses the compatible dates available, not a headline figure covering a different period. Choosing a benchmark after inspecting the result is also a research decision that belongs in the accompanying account of the process.
What the report can answer
Where sufficient data are available, the report examines statistical uncertainty, concentration in the best periods and differences between the recent segment and the rest. Selection among variants matters: supply the number of research attempts and its source to contextualise the deflated Sharpe. If that count comes from your explanation, it remains a declaration even when the calculation using it can be reproduced from the recorded inputs.
The labels Measured, Declared and NOT Measured explain that boundary. Measured identifies a calculation on the supplied data; Declared identifies information supplied by the author; NOT Measured indicates insufficient evidence for a check. Benchmark comparison depends on the benchmark series, while cost tests depend on the detail available. An absent section should not be interpreted as a favourable conclusion about the question that section would otherwise address.
What cannot be answered without the file
A screenshot or cumulative return does not establish the ordering of returns, gaps in the history or dependence between periods. Without the series, Rigor cannot conclude whether that record withstands the statistical checks. Without trades and costs, it cannot measure every execution effect either. These limits should accompany the result whenever it is shared with someone who did not supply the data.
Even with the file, the audit does not establish that a signal was published before each market move, that an account belongs to its presenter or that every discarded attempt was disclosed. It does not reconstruct the code, recommend allocations or promise future results. A statistical review of received material has a different scope from an operational, legal or ownership review. Those questions require their own evidence and procedures.
Prepare the conversation and inspect an example
Use the contact link below to describe the source, frequency and available files before defining the review. Credentials and access codes are not needed for that conversation. The examples page shows the structure of a report made with synthetic data: it helps you understand the format, not infer a conclusion about your portfolio. When you present your own report, keep the analysed period, declarations and unmeasured sections alongside it so the reader can see the limits of the supporting evidence.
FAQ
Can we start without supplying the signal's code?
Yes. The statistical audit starts from a return or trade file. It does not reconstruct the code or the signal's logic.
Can the example predict the class of our record?
No. The example uses synthetic data to show the format. Your report's class and limits depend on the material you supply.