The checkboxes that close the loop from paper to workbook.

The four confirmations

Each box is a different claim about the same trade.

The Compliance Transfer Audit asks for four separate confirmations, and they are not redundant. Entered into Journal claims the trade exists in CP3. Verified against TradeZella claims the paper record and the broker record agree. Included in weekly review claims it was actually looked at rather than merely typed. Date verified claims someone knows when the reconciliation happened. A trade can satisfy the first and fail the second — entered accurately from a paper sheet that was itself wrong — which is precisely why the boxes are separate.

The reconciliation direction

When paper and broker disagree, the broker wins — and the paper gets corrected.

The instruction is specific: when TradeZella data is available, reconcile the paper sheet against the digital record, and correct the paper sheet if TradeZella reveals a factual difference. The direction of correction matters. The paper sheet is the honesty record — it holds the operator's contemporaneous account of adherence, emotion, and intent, which nothing else captures. The broker record is the fact record for prices, fees, and durations. Reconciliation lets each be authoritative over what it actually knows, rather than treating either as the single source of truth.

What the boxes actually detect

The missing trade — the failure no downstream tool can see.

Every diagnostic in MARS computes on the trades it was given. None of them can detect a trade that was never entered, because from the workbook's perspective that trade did not occur. A week with sixteen trades and fifteen journal rows produces branch counts that are wrong, hit probabilities computed on a truncated sample, and a quota audit that reads compliant when it isn't — and every one of those outputs looks entirely normal. The transfer audit is the only place in the system where the count of trades taken can be compared against the count of trades recorded, and the comparison only happens if the boxes are used.

• Systematic omission is worse than random: the trade least likely to get entered is the one the operator would rather not revisit.
• That bias runs one direction — non-compliant and losing trades are the ones that go missing.
• Which means an unaudited journal drifts toward flattering, without a single dishonest entry.

The filing rule

Paper is kept until the weekly review verifies it, not until it is typed.

The QA rules specify that physical trade sheets are filed by week and kept until the CP3 weekly review is completed and verified. Note where the release point sits — after verification, not after entry. The gap between those two moments is where transcription errors live, and discarding the source as soon as it has been copied removes the only artifact capable of catching them. Once the review has run and agreed, the sheets become audit backup and the loop is genuinely closed.

Why this survives being boring

The audit is the cheapest control in the framework and the first one abandoned.

Ticking four boxes generates nothing, teaches nothing, and takes seconds. It is precisely the kind of task that erodes first under time pressure, and its erosion produces no immediate symptom — the journal keeps filling, the workbooks keep computing, everything continues to look fine. The symptom arrives months later as an unexplainable divergence between what the system reports and what the account balance says, at which point the missing trades cannot be recovered because the paper was discarded and the memory is gone. The control is trivial. Its absence is not recoverable.

Why the count is the real output

The audit's value is a number nobody thinks to compute.

Beyond catching individual omissions, the transfer audit produces something the rest of the system cannot: a count of trades taken that can be set against the count of trades recorded. Every workbook downstream knows how many rows it holds. None of them knows how many trades actually happened. That single comparison — sixteen sheets filed, sixteen rows entered, sixteen boxes ticked — is the only integrity check in the framework capable of detecting the absence of data rather than an error within it, which is a categorically harder thing to see and a categorically worse thing to miss.