AUDIT · MONTHLY REPORT & EVIDENCE
Lock the format first, fill in the numbers after
A fraud does it the other way round: flattering numbers first, format adjusted to suit. We fixed all seven items, their methodology, and what we say when we cannot deliver one — and only then started filling them in. This page is that format.
Evidence comes in three layers
The earlier the layer, the less it asks you to trust us. We fill layer one first because it is the only one that depends on no numbers at all — and the only one you can verify yourself today.
- Layer 1
Structurally verifiable · no trust required
Available nowThis layer asks you to believe nothing we say. Every line below is visible in your own Binance account, without going through us.
Your money stays in your own account
How you check it yourselfWe take no custody and cannot move funds out — it stays in your own Binance account throughout. (We do place orders in that account on the signal, so the balance changes; but the money never leaves your account.)
The key we hold cannot move money out
How you check it yourselfOpen Binance API management and check that key's permissions. Confirm all three are off: withdrawals, internal transfer, and universal transfer — checking "withdrawals" alone is not enough, since either of the other two is sufficient to move funds out.
You can pull the plug at any time
How you check it yourselfDelete the key in Binance. We lose every capability on the spot, with no need to notify us first. ⚠️ Know the consequence though: any open positions stay open, and we can no longer close them for you — from that moment they are yours to manage.
- Layer 2
Live trading · our own real money
Awaiting sign-offOur own account runs on real money. When this layer opens it will give a weekly equity series, state which period it covers, and state which stretch is excluded.
When this layer opens is not ours alone to decide; the data and its methodology need confirmation from the strategy side. Until then we publish no figures — and give no hint as to what those figures look like, including no hint that they are good.
- Layer 3
Backtest · disclosed in institutional format
Awaiting sign-offA backtest is not a selling point. It is material that has to be disclosed in a fixed format: time window, capital specification, cost assumptions, and known limitations — none of them optional.
A return curve with no cost assumptions, no drawdown and no time window is one you should simply not look at. Ours included.
Where we tied our own hands
Every line below is a restriction on us, not a promise to you — a restriction is harder to walk back, and easier for you to hold us to.
- Fixed formatThe same seven items every month. No cherry-picking the month, the methodology, or the definition of a metric.
- A missing cell must be explainedIf we cannot produce an item, we say why, rather than quietly shipping one cell fewer.
- Known limitations may not be blankThe disclosure table has a cell dedicated to what is weak about the material. A disclosure listing only strengths is not a disclosure.
- Excluded stretches must be markedIf any stretch of data is excluded, it is marked on the chart with the reason, not spliced over to make the curve longer.
- We do not draw the conclusion for youWe give you the comparison, not the verdict. What is yours to judge, we do not write up as an assertion.
- No claims of effectivenessWe do not claim our strategy works. We give you a checkable record and the full methodology; the judgement stays yours. This clause is reviewed once a year — what gets reviewed is whether to change the clause, not whether to start making the claim.
What that curve will look like
Below is a sketch of the disclosure format — not data, and not a time scale. The three widths do not represent durations, and the vertical axis carries no numbers.
Sketch only · widths are not durations · no data on either axis
One stretch is excluded in full
We wrote the rule — an excluded stretch gets marked with its reason — into the format itself, which leaves us no room to be vague about it later. The stretch below is what that rule produced:
Over the 11 trading days from 2026-08-07 to 08-17, one of the system's self-protection mechanisms was in a triggered state: positions could be reduced but not added to, and actual exposure ran at roughly thirty percent of design. That stretch does not represent the strategy's designed behaviour, so we exclude it in full and count it in no statistic.
The seven slots in each monthly report
The template is fixed. From the first issue onward, every month is published against the same fields — no cherry-picking the month, no cherry-picking the methodology. The four marked "awaiting" wait on the live ledger; the three marked "first issue" depend on no upstream data and will be written in full from issue one.
| Field | Status at first issue |
|---|---|
| Equity curve | In first issue |
| Win rate | Awaiting sign-off |
| Worst single trade | Awaiting sign-off |
| Max drawdown | In first issue |
| Market context | In first issue |
| Team retrospective | In first issue |
| Next-month plan | In first issue |
The three marked "first issue" depend on no upstream data and are written in full from issue one — and they are precisely the three most often left out elsewhere. A retrospective has to say what we got wrong, and a plan has to be specific enough for next month to hold it against us.
Disclosure fields for the backtest material
The first three lines below were confirmed as publishable item by item by the strategy side and are reproduced verbatim; the fourth, "known limitations", is written by us. A missing cell is visible to you on the spot.
| Field | Value |
|---|---|
| Time window | 2024-12-13 → 2026-06-13, 548 trading days |
| Capital specification | 3,000U (canonical backtest baseline, same spec as the lowest customer tier) — relative L1 truncation error 4.64% |
| Cost basis | Net of costs: 7bp/side fees + 5bp/side slippage, with funding accrued per instrument, either actual or estimated |
| Known limitations | This is a backtest, not live trading. The capital spec is fixed at 3,000U; a different size gives a different truncation error. Funding is estimated for part of the book rather than measured throughout. The live segment also has an excluded stretch, described above. |
The "known limitations" cell is not allowed to be empty. A backtest disclosure that lists only the good parts is not a disclosure.
Four questions to ask anyone — including us
The four below are generic; use them on any provider. Under each is our answer as of today, so you can check us against them directly.
- 01
Ask about costs: are fees and slippage deducted, and at what rate.
Our answerThey are. Net of costs: 7bp/side fees plus 5bp/side slippage, with funding accrued per instrument. It is in the disclosure table below.
- 02
Ask about drawdown: reporting returns without a maximum drawdown is reporting only the winning days.
Our answerDrawdown is what our risk gate is set against, not an explanation added afterwards. The risk methodology is public in the docs.
- 03
Ask about intervals: was any stretch excluded, and why. Excluding one is not the problem; being unable to say why is.
Our answerOne stretch is excluded in full. Its dates and the reason are stated above, not glossed over.
- 04
Ask what is checkable: which single claim can you verify in your own exchange account rather than take on their word.
Our answerAll three claims in layer 1, right now, entirely inside your own Binance account, without going through us.
How we stop ourselves from editing the numbers afterwards
Each issue will be published together with a hash, which shows that the issue was not altered after it went out. What it does not show should be said plainly: it places no constraint on running several versions beforehand and publishing the best-looking one — that would require publishing a commitment hash before the data exists (commit-reveal). We intend to do that, but have not built it yet, and the hash is not anchored to any external timestamp either. Until it is actually live, we will not say it is protecting you.
Not live yet · currently a plan, not code
Past issues
In the meantime, layer one is verifiable today
You do not need to wait for our numbers — key permissions and custody of funds are two things you can check yourself in the Binance console in ten minutes.
Read the onboarding & permissions guide