Every layer can lower the tier. Only some are allowed to raise it.

The two columns

Reading a layer's power means reading both directions separately.

A layer's entry in the hierarchy answers two independent questions, and collapsing them into a single notion of influence loses the design. The gate defines the maximum survivable exposure row and the maximum risk tier — it can demote and it cannot promote, which sounds like a limitation and is the opposite. A gate that could promote would mean shallow drawdown actively authorising more risk rather than merely permitting the existing ladder, and the account would then be sized largest at precisely the moment its own good fortune was the evidence. The structural expectancy read can do both, because it is a claim about whether aggression is deserved. The risk-efficiency and profit-factor reads can do both as modifiers. Acceleration adjusts timing in both directions. Each entry is a deliberate answer to two questions, and the answers differ across layers.

Why demotion is universal

Any layer noticing something wrong is sufficient reason to deploy less.

Every layer in the stack can lower the tier, and the reason is that the cost of a false demotion is bounded while the cost of a missed one is not. If the risk-efficiency read suppresses aggression during a stretch that turns out to have been fine, the account deployed less than it could have and forfeited some upside — a real cost, measurable, recoverable. If a layer notices genuine deterioration and has no power to act on it, the account deploys at full size into a condition that at least one instrument had already identified. Making demotion universally available means the system's caution is the union of everything it knows rather than the intersection, and any single instrument's warning is sufficient. Nothing has to be corroborated in order to reduce risk, which is why the survivability rule resolves conflicts downward.

Why promotion is rationed

Raising risk requires structural evidence, and speed is not evidence.

Promotion runs on the opposite logic, because the cost asymmetry inverts. The layers permitted to promote are the weekly structural reads — expectancy, efficiency, profit quality, acceleration — and what they share is that each summarises a period long enough for the reading to mean something. Daily expectancy is described as a fast tactical signal that can support but should not prove structure alone, so its promotion power is limited rather than absent: it may select a higher tier inside a cap that structural evidence has already justified, and it may not create the justification. The monthly regime has no direct promotion power because a macro overlay measured monthly should not be used as a cycle selector at all. The rule underneath is that a layer may raise risk only on the strength of evidence gathered over a period commensurate with the commitment being made.

The arithmetic

The asymmetry is enforced by minimums rather than by judgement.

The construction makes the rule mechanical rather than aspirational. Modifiers accumulate into a raw core recommendation bounded to the ladder's range, and that recommendation is then resolved by taking the minimum of it, the gate's cap, and any operator-imposed ceiling. The structural overlay stage takes a further minimum against the composite confirmation cap and the diagnostic cap. Because every resolution stage is a minimum, any layer holding a low value binds the outcome regardless of how enthusiastic every other layer is, and no combination of favourable readings can lift the result above the most restrictive one. Promotion power therefore only ever means the ability to raise the raw recommendation before the minimums are applied — which is exactly as much power as the design intends a modifier to have.

The failure it prevents

A stack where everything can promote drifts upward, permanently.

Consider a version of the panel in which every layer could raise the tier as freely as it could lower it. Because these readings are noisy and partially independent, at any moment some subset of them is favourable, and an operator seeking a reason to size up would rarely have to look far. Over a long enough run the tier would ratchet toward the top of the ladder and stay there, not through any single bad decision but because the system had been given many independent routes upward and the same number downward, while the operator's attention is asymmetric between them. Rationing promotion to the slow structural layers means going up requires a specific kind of evidence that is not always available, which is the only reliable way to make higher tiers rare without relying on restraint.

• Any single layer may demote — caution is the union of everything the system knows.
• Only slow structural layers may promote; the fast ones select inside a settled cap.
• Every resolution stage is a minimum, so the most restrictive reading always binds.

The key idea

Two-way influence is the default assumption and almost always the wrong one.

Most scoring systems treat their inputs as symmetric by construction — a metric contributes positively when favourable and negatively when not, and nobody writes down which direction each input is trusted in. That symmetry feels neutral and encodes a strong claim: that being right about danger and being right about opportunity are equally reliable for every instrument. They are not. A drawdown reading is highly reliable about danger and says nothing about opportunity. Splitting the two directions and answering them separately for each layer is what allows the stack to be trusting where trust is cheap and sceptical where it is expensive, instead of setting one confidence level for everything.