Dual Weakness: the one state the manual calls a danger condition.

What it means

Neither instrument has anything reassuring to report.

RAER asks whether deployed risk is converting into useful return. RAPF asks whether the profit being produced holds up as quality against the risk it consumed. Under Dual Weakness both answers are negative, which removes the interpretive move available in every other adverse state. Under-monetisation at least establishes that the execution is sound and the problem is located at the exits; profit-without-efficiency at least establishes that results are arriving, even if the risk consumed to produce them is not justified. Dual Weakness offers no such foothold. There is no half of the machine performing well enough to build a repair around, and that absence — rather than the severity of either individual reading — is what makes it categorically different.

The playbook

Stop adding, start reading — and do not attempt a fix from inside it.

The correct response has three parts and none of them is a trade. First, the state is a review priority that outranks everything else the diagnostic layer has produced this cycle; whatever else was scheduled for the monthly review gives way. Second, the enquiry is deliberately wide rather than targeted, because a state with no healthy half offers no starting hypothesis — the branch-level SDE pipelines, the execution efficiency lab, the checklist adherence record, and the regime reading all get opened, precisely because the engine cannot say which of them holds the cause. Third, and most importantly, the operator resists the instinct to redesign. A system reading Dual Weakness is a system whose operator is under pressure, and structural changes made under pressure are the ones the record shows most reliably regretted.

• The state directs attention and does not name a cause — the engine is composite-only and cannot locate the branch.
• It is a diagnostic layer output and still holds no deployment authority; if capital needs restricting, that is the gate's finding, not this one.
• One week of Dual Weakness is a reading. Persistence is what converts it into a characterisation.

Severity and conviction

How strongly the state is being asserted is part of the state.

Severity in the engine derives from the conviction values on both sides, usually taking the stronger. A Dual Weakness carrying low conviction on both metrics is a different object from one carrying high conviction on both: the first says the reading is adverse and the underlying drift is not firmly established, the second says two independent pipelines are confident. The response scales accordingly, and reading the state label without its severity discards the distinction. This is the same discipline the SDE applies with its stability layer and the regime engine applies with its QA flag — the system consistently asks how much a reading should be believed before asking what to do about it, and the contradiction engine is no exception.

The false comfort

Dual Weakness can arrive while the account is profitable.

Both metrics are risk-adjusted, and both are read cumulatively rather than weekly. That combination means the state describes structural condition rather than recent results, and structural condition can deteriorate through a period whose surface numbers are acceptable — a run of small wins consuming disproportionate risk, or profits arriving from a narrowing subset of the flow. An operator checking equity and finding it intact will reasonably conclude the engine is being alarmist. The whole purpose of the layer is to be readable before equity confirms it, which necessarily means being unwelcome at the moment it is most useful. If the state only appeared once the curve turned, it would be a report rather than a diagnostic.

The key idea

The dangerous state is the one with nothing left to build on.

Severity is not what makes Dual Weakness the danger condition — a severe under-monetisation can cost more in a given month. What makes it different is structural: every repair needs a functioning part to work outward from, and this is the state defined by not having one. That is why the instruction is to widen the enquiry rather than narrow it, and why the manual reaches for stronger language here than anywhere else in the engine.