A single bad grade changes your posture. Only a pattern should change your system.

The two responses

Posture is weekly and automatic; structure is monthly and deliberate.

The postural response requires no analysis and no permission. The grade lands, the doctrine applies, and the operator's selectivity and expansion appetite adjust accordingly for the coming week. It is cheap, reversible and correct even when the red grade turns out to have been noise. The structural response is expensive and hard to reverse: changing a weight alters every historical grade, retiring a branch removes a source of expectancy that may take months to prove was needed, and rewriting a rule resets the evidence clock on everything downstream of it. Matching the cost of the response to the strength of the evidence is the whole discipline, and one week is never strong evidence.

Why one week says so little

The grade is computed from probabilities estimated on a handful of trades.

A trading week contains a small number of positions, and the branch probabilities derived from it inherit all the imprecision that implies. Two or three outcomes landing differently would have moved several of those probabilities materially, and the blended figure with them — quite possibly across a threshold. This is not a defect in the workbook; it is what a weekly cadence costs. It does mean that the difference between a yellow week and a red one is frequently within the range that chance alone could have produced, and that treating the boundary crossing as a discovery about the system is reading a precision the sample does not contain.

What a pattern looks like

The monthly rollup counts statuses, and the count is the level at which the question becomes answerable.

The period view exists to convert single readings into a distribution. It reports how many weeks fell into each status, what the average and weighted expectancy were across the period, whether branch participation quotas were met and whether pacing is ahead or behind. Repeated reds inside one period is a governance signal in a way that one red is not. A status mix that has shifted from mostly green to mostly yellow across two periods is a signal even if no single week was alarming. Those are the observations that justify a structural response, and none of them can be made from a row.

• One red inside a green month is variance and needs no structural response.
• Repeated reds, or a shifted status mix across periods, is the actual trigger.
• Quota failure across a period is a separate governance finding with its own response.

The asymmetry that makes this hard

The pressure to act is highest during the week whose evidence is weakest.

The structural response is most tempting immediately after the bad week, when the discomfort is fresh and doing something feels responsible. It is least justified at exactly that moment, because the sample is one week old and the diagnosis has not been performed. Waiting is not passivity — the postural response is already in force, the diagnostic surfaces are available, and the period is accumulating the evidence that would make a structural decision defensible. What waiting costs is the feeling of having responded, and that feeling is not worth a weight change that will make three months of history unreadable.

The separation that protects both

Because posture responds immediately, structure is free to respond slowly.

The two loops are not merely tolerated alongside each other; they depend on each other. Structural patience is only safe because something else is already reacting — an operator whose only available response to a bad week was a system change would face real pressure to make one. Since the doctrine has already reduced selectivity and blocked expansion, the account is protected while the evidence accumulates, and the structural question can be held open for as long as answering it properly requires. That is the same arrangement the gate and the review cadence run everywhere else in the system: a fast brake so the slow judgement can be slow.

The key idea

Systems are rarely destroyed by a bad month. They are destroyed by the response to one.

A strategy with genuine expectancy will produce red weeks, and it will produce them in clusters occasionally, because that is what variance does. If each cluster triggers a revision, the strategy is never actually run — what gets run is a sequence of strategies, each abandoned before its evidence matured, and the resulting record shows an operator who has traded for years without ever accumulating a testable sample. Matching the response to the evidence is what allows a plan to exist for long enough to be evaluated at all.