The return leg is the half of the handoff that gets skipped.

The loop

Four transfers, and the cycle only closes when all four have happened.

The two-workbook rail moves material in four directions rather than two. The console sends the throttle its tactical evidence and exposure state. The throttle returns the final tier, authorised pool, authorised per-trade average and the effective deployment figures. The throttle also writes the full decision snapshot into the log. And the next cycle's executed trades return to the ledger, sized according to what was authorised, so that future evidence describes trades the system actually sanctioned. Each leg has a destination and a reason, and the loop is what makes the following cycle's evidence truthful rather than merely recorded. Three of the four legs feel consequential. The one that feels administrative is the mirror, and it is the one that fails.

What the mirror carries

Two values with named cells, and both are inputs to the console's own arithmetic.

The authorised cycle pool and the authorised per-trade average return to specific destinations in the calculator, and their roles there are not decorative. The pool is the gross figure the console's smart open-exposure logic subtracts active open risk from in order to compute remaining capacity. The per-trade average is the ceiling that smart fresh sizing is not permitted to exceed. Without them, the console's exposure block has no authorised envelope to reason against — it can still measure what is open, and it cannot say what remains, because remaining is defined relative to an authorisation that lives in the other workbook. The effective figures and the directive travel alongside, but those are for the log and next-cycle reference. The two mirrored values are what the console consumes.

Why it fails

The step has no immediate consequence, which is why it disappears.

Every other leg of the loop produces something the operator wants in the moment. Sending evidence gets a decision. Taking the decision permits trading. Logging closes the cycle and is the last act before moving on. The mirror produces nothing the operator needs right now — the tier is already known, the trades can be placed from the throttle's own output, and the console will open perfectly well tomorrow. Its consequence arrives a full cycle later, when the exposure block computes remaining capacity from whatever pool figure it still holds. Failures whose cost is deferred past the point of attention are the ones that survive, which is why the manual names this one as an edge case rather than trusting it to routine, and why the checklist gives it its own numbered step.

The failure signature

A stale pool figure produces confident capacity arithmetic against last week's authorisation.

The specific damage depends on which way the authorisation moved. If the tier was reduced and the console still holds the older, larger pool, remaining capacity is overstated and the console suggests fresh deployment the throttle never authorised — the operator is sized against a permission that has been withdrawn. If the tier rose and the console holds the smaller figure, capacity is understated and the account under-deploys quietly. Neither case produces an error, because the arithmetic is correct throughout; only its input is wrong. And the mismatch is not detectable from within the console, which has no way to know that the number in its pool cell is older than the decision it is being used to implement. Detection requires comparing the two workbooks, which is exactly the act that was skipped.

Making it stick

The mirror belongs with the log, at the close of the cycle.

Because the step has no natural prompt, it needs to be attached to one that does. The checklist places it immediately after the throttle decision and alongside executing with effective sizing, and pairing it with logging works well in practice: both are closing acts, both are administrative in feel, and both have consequences that arrive later. A useful confirmation is to read the console's pool cell against the throttle's authorised pool before beginning the next cycle's setup — two numbers, one glance, and a mismatch means the previous cycle did not close. That check is cheap enough to run every time and is the only thing standing between a skipped mirror and a cycle sized against a permission that expired.

• The pool and per-trade average are inputs to the console's own capacity block.
• A stale pool overstates capacity after a demotion and understates it after a promotion.
• Neither workbook can detect the mismatch alone — the check is a comparison.

The key idea

A one-way transfer is a report; a two-way transfer is a contract.

The distinction matters because it determines what happens when a leg is missed. If material only ever flowed from console to throttle, an omission would mean the throttle deciding on stale evidence — visible immediately, because the decision would be about a cycle the operator remembers differently. Because the flow returns, an omission instead corrupts a calculation one cycle downstream in a workbook that appears to be functioning. Contracts with a return leg are more capable and more fragile in exactly this way, and the price of the capability is a step that has to be performed on discipline rather than on need.