It is a hierarchy, which means the order carries meaning.

Ranked, not listed

The word hierarchy was chosen over panel deliberately.

A panel implies peers; a hierarchy implies precedence. The distinction becomes operational the moment two metrics disagree, which they do routinely — and without an ordering the operator resolves the conflict by preference, which reliably means by whichever number is currently more flattering. The ranking exists to remove that discretion, in exactly the same way and for exactly the same reason the gate ladder removes sizing discretion under drawdown. Fixing precedence in advance also removes a specific argument the operator would otherwise have with themselves every time the readings split, and removing recurring arguments is most of what the governance layer does.

The order

Expectancy first, drawdown second, then the auditors.

Expectancy leads because a system without edge cannot be rescued by any reading beneath it — efficient conversion of nothing is still nothing. Drawdown sits second because it determines whether the edge can be harvested at all: a positive-expectancy method that produces unsurvivable holes is unusable regardless of its EV. RAPF, RAER, and acceleration follow as the audit layer, qualifying and cross-examining the top two rather than competing with them. That ordering resolves most conflicts before they need to be argued.

Conflicts, resolved

Three common disagreements and what the ranking says.

The ordering earns its keep on ordinary cases rather than exotic ones:

• Expectancy strong, efficiency weak — the edge is real and being deployed badly. Repair deployment; do not touch the method.
• Efficiency strong, expectancy weak — excellent conversion of an inadequate edge. Efficiency work is wasted here; the method is the problem.
• Drawdown fine, acceleration falling — nothing is currently unsafe and something is decaying. Investigate rather than intervene.

What the ranking is not

Authority in a conflict is not importance in a diagnosis.

A caution worth stating: acceleration sits at the bottom of the ordering and is frequently the most informative reading on the panel, because it is the earliest. Low precedence means it does not override expectancy when the two disagree — it does not mean it should be consulted last or valued least. The hierarchy governs conflict resolution only. Confusing precedence with importance leads operators to under-weight exactly the metric that gives them the most warning time, which inverts the panel's main advantage.

Why not a composite

A single blended score would destroy the information.

The obvious simplification is to weight the five into one number, and it is rejected for a specific reason. The diagnostic value lives entirely in the disagreements, and a composite averages disagreements away by construction — a system with strong expectancy and collapsing efficiency and a system with mediocre everything can produce the identical composite while requiring opposite interventions. The score would be easier to read and would have discarded the only thing worth reading it for.

The hierarchy under stress

Precedence matters most when the readings are unwelcome.

Ordering is unnecessary when all five agree and indispensable when they do not — and they most often disagree during difficult stretches, which is also when the operator is least inclined to accept the ranking's answer. A trader in drawdown with strong expectancy and weak efficiency is being told by the hierarchy to fix deployment and leave the method alone, which is the harder path and the correct one. That the ranking was fixed in advance, while nothing was at stake, is what allows it to be followed at the moment it is inconvenient. The pattern to watch for is an operator who accepts the ranking readily in calm weeks and discovers a reason it does not apply during difficult ones — that discovery is the ranking working correctly and being resisted, not a flaw in the ranking.