How to Verify Trading Results

A percentage in a profile is a question, not an answer. Before repeating it, identify the subject, period, account or model, measurement method and original source. Those five labels prevent a personal return from being presented as a subscriber result, or a contest division result from becoming a claim about an entire career.

This archive guide is a desk procedure, not a ranking of traders. It was updated to remove unsourced industry-wide assertions from the earlier version. The SEC investor bulletin on performance claims warns readers to examine how returns are calculated and whether losing periods have been left out. That is the starting question for every record below.

1. Write the claim in one testable sentence

Copy the number exactly, then add a named actor and a start and end date. State whether the result belongs to a personal account, a competition entry, a managed fund, a model portfolio or a stream of published signals. Write the metric beside it: account return, return on capital, pips, win rate or risk-adjusted score. A figure with no denominator or period is not ready for comparison.

Record whether the claim is gross or net of commissions, financing, subscription fees and slippage. Ask how deposits and withdrawals were handled. If capital changed during the period, a start-to-end balance ratio alone is not a reliable return calculation. Keep the raw statement and the method that turns it into the headline number.

2. Find the source that owns the record

For an account claim, request a dated statement or audit report with enough scope to identify the account, period, calculation basis and independent reviewer. Check who prepared it and what they actually examined. A letter that confirms account existence is not an audit of every trade. A cropped balance screenshot has no visible population of deposits, withdrawals, costs or losing periods.

For a competition result, use the organiser's own results and rules. Match the entrant, year, division, placing and return. Note any special scoring or eligibility rules. A strong contest performance is noteworthy evidence of what happened in that contest; it is not a substitute for a multi-year account history or the experience of a subscriber following signals elsewhere.

For a person selling or managing financial services, check the legal entity and permissions separately from the performance record. In the US derivatives market, NFA BASIC is an official background-search starting point. A clean identity search does not prove profitability, and a return statement does not establish authorisation.

3. Reconstruct the complete sequence

A signal archive needs more than highlighted winners. Request the original message, publication timestamp, instrument, direction, entry rule, invalidation or stop, exit rule, every update and final scoring rule for all calls in a defined period. Count cancelled, missed and losing calls. If the seller uses several models, keep a separate denominator for each and state when the model began.

A cryptographic commitment can help with chronology. Recalculate the digest from the exact original payload and verify its receipt against an independent timestamp. OpenTimestamps explains the distinction between creating a timestamp proof and upgrading it to a confirmed Bitcoin attestation. A valid proof supports the existence of those bytes by the attested time; it does not prove that the message reached subscribers, that a trade filled, or that the archive includes every losing call.

Check the version history. If a stop, target or grade changes after publication, the initial and revised instructions need separate timestamps. An outcome cannot be scored under whichever version looks best after the fact. Where original files are unavailable, mark the chronology unverified instead of guessing from an image.

4. Test the arithmetic against the trade path

Start with a small sample you can reproduce. For each call, record the publication time, the subscriber's receipt time, the quoted entry, the price available at the subscriber's venue, the stop and the exit. A reference entry may be useful for evaluating a model, but it is not the same as a realised fill. Spread, order size, delay and fees can change the result.

Here is an illustrative check, not a historical return: a long signal with entry 100, stop 96 and target 108 offers 4 units of reference risk and 8 of potential gain, or 2R before costs. If the available fill is 100.8, the same stop is 4.8 away and the target 7.2 away, or 1.5R before costs. Report both rows, identify the assumed prices and never convert the example into a provider performance statistic.

A model backtest, an audited personal account and a live subscriber ledger are three different objects. They can be compared, but one cannot silently inherit the other's assurance. The most useful result may be a narrow one: the organiser confirmed a placing, while subscriber execution remains untested.

5. Publish the uncertainty as carefully as the number

When a source is missing, say which one. Useful labels include organiser-confirmed, account statement supplied, publisher-reported, timestamped publication and subscriber outcome not independently tested. Avoid the unqualified word "verified" when only one layer has been checked.

Keep a link or archived copy of each source, the date checked and any correction. If a provider later changes its story, readers should be able to see which claim moved. The editorial job is not to convert incomplete evidence into certainty; it is to make the boundary of the evidence easy to inspect.

Primary sources: SEC performance-claims bulletin, NFA BASIC and OpenTimestamps documentation. The arithmetic example above is hypothetical and carries no claim about a trader or publisher.