A red grade names the symptom and is structurally incapable of naming the cause.

The four candidate causes

One red grade, four possible stories, and they call for four different responses.

Expectancy may genuinely be decaying, in which case the strategy or the market has changed and something structural needs attention. The sample may be thin or the week distorted by news, in which case the correct response is to record it and continue. Trades may have been classified into the wrong branches, which corrupts the probabilities and means the grade is describing an accounting error rather than performance. Or friction may have consumed the difference, which is a cost problem and will not respond to changes in selection or management at all. Nothing in the grade distinguishes them, and choosing between them by instinct means acting on the story the operator finds most available rather than the one the evidence supports.

Why the monitor is deliberately shallow

A weekly instrument that could diagnose would be too slow to be weekly.

The shallowness is a design choice rather than a limitation nobody got round to fixing. The scorecard is meant to be completed every week without fail, which puts a hard ceiling on how much work it can require. An instrument capable of distinguishing edge decay from classification error would need trade-level data, several statistical families and a research posture, and it would not get filled in on a Saturday morning after a difficult week. Splitting the fast shallow monitor from the slow deep diagnostics is what makes the fast one survivable, and survivability is the property that matters most for a weekly cadence.

The routing question

Ask what kind of answer would settle it, and the surface follows immediately.

In practice the routing is easier than the four-way split suggests, because each candidate cause has a distinctive kind of evidence attached. Questions about whether the edge exists at all are statistical and belong to the expectancy lab. Questions about whether opportunity was generated and kept are path questions and belong to the execution lab. Questions about whether a pattern is a condition or noise are horizon questions and belong to the structural engine. Questions about what may be deployed next are capital-state questions and belong to the gate. The scorecard's contribution is to say that one of these questions now needs asking, which is genuinely useful and is where its contribution ends.

• Statistical questions to the EV lab; path questions to the execution lab.
• Horizon questions to the structural engine; deployment questions to the gate.
• The grade triggers the routing and never performs it.

The notes column

The one place the workbook carries causal information, and it is written by hand.

There is a partial exception worth knowing about. The weekly row includes a notes field, and it is the only part of the workbook that can hold a reason. Recording that a week contained three high-impact news days, or that a branch was likely misclassified, or that spreads were unusual, gives the monthly review something the numbers cannot supply — context, written while it was still remembered. Those notes are not evidence in the statistical sense and should never override a number. Their function is narrower and still valuable: they stop a monthly review from constructing an explanation for a strange week out of whatever the operator can still recall four weeks later.

The impatience problem

The instinct to diagnose from the grade is strongest exactly when it is least reliable.

The temptation to skip the routing is at its peak after a bad week, when the operator wants an explanation immediately and the deeper surfaces require effort. What gets produced instead is a diagnosis assembled from memory and mood — usually that the edge is decaying, because that is the most alarming available story, and it is the story that leads to changing the system in response to a single noisy sample. The status alone can never support that conclusion. If the answer matters enough to act on, it matters enough to get from the surface that can actually produce it.

The key idea

An instrument that knows what it does not measure is worth more than one that guesses.

The scorecard could have been built to offer a suggested cause alongside the grade, and it would have been used constantly and been wrong often, because the information required is genuinely not present in its inputs. Declining to guess is what keeps the grade trustworthy: when it says red, that statement means precisely one thing, it means it reliably, and the operator knows to go and find out why somewhere else. A layer that stays inside its evidence is the only kind that can be believed without qualification.