The governance overlay governs capital and is blind to selection.

What the overlay is

A capital-authority layer that never inspects the method beneath it.

The overlay consists of drawdown gates, brake states, risk tiers, cycle-pool authorisation, per-trade limits, open-exposure adjustment, System Lock and override logging. Every one of those operates on the size and timing of deployment, and none of them examines what is being deployed into. This is deliberate rather than incomplete: it is what makes the overlay adoptable at all, because a governance layer that required approval of the strategy would not be an overlay. The price of that portability is a specific and permanent blindness, and buyers should know which side of it their actual problem sits on.

The consequence for a leaking edge

An overlay on a decaying method makes the decline orderly.

If the underlying strategy is deteriorating, the overlay behaves exactly as designed and produces a result that is easy to misread. Drawdown accumulates, the gate steps down, tiers compress, deployment shrinks, and eventually the lock engages. Every one of those responses is correct, and none constitutes a diagnosis. The operator observes a system reducing their risk and may reasonably conclude the governance is working — which it is, in the sense that the capital is being protected, and is not, in the sense that the cause remains unexamined and unaddressed. The overlay converts a fast loss into a slow one and provides no theory of the loss.

The consequence for a sound edge

Where the overlay is genuinely the whole answer.

The opposite case is the one it was designed for and the one it solves outright. A trader with a real edge whose results are damaged by oversizing after wins, exposure stacking across concurrent positions, or aggressive attempts to recover a drawdown does not have a strategy problem. They have a deployment problem, and the deployment problem is fully inside the overlay's territory. Here the overlay is not a partial adoption in any meaningful sense: it addresses the entire binding constraint, and adding analytics would produce interest rather than improvement. This is also the cleanest case in the whole adoption question, because the constraint and the instrument line up exactly and the result is visible within a cycle or two.

• The overlay governs deployment and never inspects the method.
• On a decaying edge it produces an orderly decline and no diagnosis.
• On a leaking deployment it is the complete answer, not a partial one.

Diagnosing which case applies

The distinguishing question is answerable before adopting anything.

The two cases separate on a question the operator can put to their own history: over the last extended stretch, were the losses driven by trades that went wrong, or by the sizing of trades that were roughly as expected? A method producing outcomes broadly in line with its historical distribution while the account declines is a deployment problem. A method whose outcomes have themselves shifted is not, and no amount of capital governance addresses it. That is a rough test rather than a rigorous one, and it is still considerably better than assuming the overlay covers both.

The lag on the one signal it does have

Aggregate expectancy is real evidence and it arrives late.

The overlay is not entirely without a view of the method, because expectancy computed across the trades actually taken is a genuine reading on whether the set still pays. The difficulty is timing. That reading needs enough closed trades to separate a shift from ordinary variance, which places it months behind the change it is describing, and the gates will have been compressing deployment throughout the interval on evidence that never named a cause. An overlay-only operator therefore learns the same fact as a full adopter, later, and without the branch attribution that would say which part moved.

The key idea

Portability and blindness are the same property.

The overlay works above any strategy that produces loggable trades precisely because it declines to have an opinion about strategies, and that indifference is what makes it both immediately adoptable and permanently unable to answer the selection question. It is not a limitation to be engineered away in a later edition — an overlay that judged setups would have stopped being an overlay, and would have required the operator to surrender the one part of the process they adopted it specifically to keep. What follows is a buying instruction rather than a caveat: adopt it for the problem it owns, and do not wait for it to raise a concern it has no instrument to detect.