A verdict on your trading system that you can show.
Independent statistical audit of a trade history: open-source engine, published method, thresholds fixed in advance, report reproducible byte for byte, attestation anyone can verify. The engine states facts; it advises nothing and predicts nothing.
Start with the free pre-check See a reportWhy this verdict is worth something
The thresholds are written before your file
Five eliminatory barriers, eleven detectors, their parameters: everything is published. A NO-GO is argued on facts, not on an opinion.
Same file, same report, byte for byte
Every report prints the SHA-256 fingerprint of your data, the method version and the seed. The engine is open source: anyone holding the file can redo the computation.
A numbered attestation, a public registry
Issued only on GO, recorded in the registry with the file's fingerprint. Your clients verify it without taking your word.
The five questions put to your results
Is chance ruled out?
Deflated Sharpe: does the performance survive the number of variants you tried before keeping this one?
And with bad luck?
10,000 block replays of the same trades: in the 5 % least favourable scenarios, do you still win?
Do the losses come clustered?
Your drawdown compared with thousands of reorderings of the same trades.
Would a system with no edge do as well?
Systems with no edge at all, same sizes of gains and losses: how many beat you?
Does it hold outside the periods it was tuned on?
Walk-forward on your parameter sets: what is left of the performance out of sample.
Plus eleven detectors of the way the trading is done: martingale, grid, missing stops, profit concentration, too-smooth curve, asymmetry, single-period dependence, unrealistic holding times and, with quotes, look-ahead, impossible prices, over-sensitive stop.
How it works
You send the raw export
MetaTrader 4 or 5 report (account or Strategy Tester), Deriv export, or a CSV with one row per trade. Never your code: the audit covers your trades. How to export it.
Free pre-check
Within fifteen minutes: readable file, fields present, sufficient sample. If the data does not hold, you know before any money changes hands.
Audit re-run within 72 hours
We run the engine again on your file: verdict on page one, an “In plain words” section that reads without us, barriers, findings, simulations, benchmarks. Attestation recorded in the registry on GO. A 45-minute debrief call.
Services
| Service | Price | Attestation | |
|---|---|---|---|
| Open-source engine The whole engine — command line, desktop window, library — and every report, to run yourself. AGPL-3.0 licence. | free | — | See the code |
| Pre-check Is your file usable? Answer within fifteen minutes, before any payment. | free | — | Send a file |
| Transferable attestation We re-run the audit on your file; on GO, a numbered attestation in the public registry that your buyers verify themselves. Due whatever the verdict. | 390 € | transferable attestation | Order |
| Portfolio of systems Several systems audited together, with the statistical correction for multiple selection. | on quote | attestation on GO | Request a quote |
| Due diligence Everything, plus an investment-committee note; scope and price on quote. | on quote | transferable attestation | Request a quote |
| Commercial licence Embed the engine in a closed-source product or service, without the AGPL obligations. | on quote | — | Request a quote |
The engine is the same for everyone and its code is public: the verdict depends neither on the service nor on who runs it. What you pay for is an audit re-run by a third party and recorded in a public registry.
The rules
A NO-GO stays a NO-GO
The report is due and paid whatever the verdict. No conclusion is changed because it displeases — that is what makes the attestation credible when it is favourable.
Unusable data: refunded
The free pre-check makes the case rare. Should it happen after payment anyway, full refund.
What the audit does not do
It predicts no performance, is not investment advice, does not fix your system and recommends no settings. It analyses only the data provided.
Frequently asked questions
Why pay if the engine is free?
Running the engine yourself gives you the verdict. An attestation is worth something else: a third party re-ran the computation on your file and recorded it in a public registry that your buyers check without taking your word.
Does my source code stay confidential?
We do not need it. The audit covers your trades, not your code: you never hand over your strategy.
Why insist so much on the number of variants tested?
Because it is the question nobody asks and it eliminates most systems. A result obtained after three hundred attempts is not worth the same result obtained first time. The report shows from how many attempts your result stops being significant.
And if my system is good?
You receive the numbered attestation, recorded in the registry, and the drawdown benchmarks: the loss level beyond which your system no longer behaves as in its history. Often the most useful part of the report, even for a sound system.
What happens to my data?
The source file is deleted 30 days after delivery; the archive kept contains no name, no account number, no individual trade. Data policy.