The chain ends with a number that is not an instruction but a verification.

The two kinds of output

Sizing outputs are instructions; the total is a prediction to be tested.

Effective fresh trades, effective per-trade risk and the fresh deployment pool are instructions — they tell the operator what to do next. Total active risk after new trades is arithmetically simple, being active open risk plus the fresh deployment pool, and it answers a different question: given that the instruction is followed, what is the account's total exposure going to be? That is a statement about a future state rather than an action, and its value comes entirely from the comparison it invites. It should remain inside the authorised pool unless the full-tier override is in use, which is a testable claim with two outcomes. The line exists so the chain's result can be falsified rather than assumed, which is a different property from the chain being correct.

Why a chain needs a terminal check

Each step is individually correct and the composition can still be wrong.

The smart-capacity chain is eight steps of arithmetic, and each one is simple enough to be obviously right. That is exactly the condition under which composition errors survive. A wrong cell reference three steps up produces a valid number that flows through five further correct calculations and emerges as a well-formed deployment instruction; nothing in the sequence has an opportunity to notice, because every step is faithfully computing from what it was handed. A terminal check breaks that property by re-deriving the outcome from a different direction — open risk plus fresh deployment, against the authorised pool — so a chain that drifted anywhere along its length lands somewhere the comparison can catch. It is the same reasoning that makes the derived columns in the journal function as an integrity check on the typed ones.

The override interaction

The one mode permitted to fail the check is the one required to warn.

Two of the three throughput modes cannot fail this comparison, because both are constructed to fit inside remaining capacity — one by reducing count, one by reducing size. Full Tier can fail it, since it authorises four trades at full tier risk regardless of what is already open. That is the mode's declared behaviour rather than a defect, and the integrity requirement attached to it is that the workbook raises a warning when total active risk after new trades would exceed the authorised pool. The check and the warning are therefore the same mechanism seen from two sides: the comparison establishes the fact, and the warning ensures the fact is stated rather than merely computed. An override that breaches the pool silently is indistinguishable from a chain that miscalculated, and the warning is what separates the deliberate case from the broken one.

What it protects against

Exposure stacking is invisible at the level of any single decision.

The failure this line guards is cumulative rather than momentary. An operator carrying two positions authorises four fresh trades at the tier's per-trade average, and each of those five decisions is defensible in isolation — the tier is correct, the per-trade average is correct, the carryover positions were correctly sized when they opened. The account is nonetheless running six positions of risk against a four-position budget. Nothing about the individual sizing was wrong; the arithmetic of the total was simply never performed. Computing it explicitly, as a named line at the end of the chain, converts an emergent property that nobody was responsible for noticing into a rendered number sitting beside the ceiling it needs to be compared against.

Reading it

Two numbers, one glance, and the comparison is the whole procedure.

The operational content is deliberately small. Before executing a cycle, read total active risk after new trades and read the authorised cycle pool, and confirm the first is not above the second. If it is, one of two things is true: the full-tier override is deliberately in use and the excess is the known price of that decision, or something in the chain is wrong and needs tracing before any trade is placed. There is no third explanation, which is what makes the check worth performing every cycle rather than periodically. It also belongs in the decision log, where it is recorded as the final exposure reality check — so the comparison survives the cycle and a pattern of near-ceiling totals becomes visible across weeks rather than being re-noticed and forgotten each time.

• Open risk plus fresh deployment, compared against the authorised pool — every cycle.
• Above the ceiling means either a deliberate override or a broken chain. Nothing else.
• Logging the total is what turns a per-cycle check into a detectable pattern.

The key idea

A calculation that cannot be wrong is a calculation nobody is checking.

It is tempting to treat a verified formula chain as settled — the arithmetic was correct when it was built, and arithmetic does not decay. What decays is the wiring around it, and what changes is the situation it is applied to. Ending the chain with a figure whose only purpose is to be compared against a constraint means the system routinely asks whether its own result is admissible, using a route that does not depend on the steps that produced it. That is cheap, it is mechanical, and it converts the pool from a parameter the chain respects into a constraint the chain is measured against.