Capital state is routed on the rare trading question that has an exact answer.

The scarcity

Objectivity is the constraint, and there is very little of it available.

Set out the questions a trading system has to answer and sort them by whether two competent people looking at the same evidence would agree. Is this a valid setup — no. Is the trend intact — no. Should this branch be used here — no. Is the market in expansion — arguably, with a defined method, but the method has parameters. How far is current equity below its highest realised value — yes, unambiguously, forever, with no method to argue about. That question is almost alone in its category, and the system's response is to build its non-negotiable layer on it and leave the contestable questions to discretion, where they belong. The gate is not the most important thing MARS measures. It is the thing MARS can measure without argument.

Why authority follows objectivity

A layer that can be argued with will eventually be argued with.

The reason the objective quantity gets the highest authority rather than the most predictive one is a claim about durability under pressure. Any layer whose input requires interpretation offers a route to a different answer, and the route will be taken at some point — not through dishonesty but through the ordinary process of an operator reconsidering a judgement they were always entitled to make. A layer whose input is a subtraction offers no route. There is nothing to reinterpret, no window to lengthen, no threshold to reconsider, and therefore no version of the conversation in which the ceiling turns out to be higher than it was five minutes ago. Predictive power is worth less here than unarguability, because the layer's job is to hold precisely when someone wants it not to.

What it costs

The gate knows nothing about why the drawdown happened.

The price of routing on an objective number is that the number is thin. Drawdown from peak does not know whether the losses came from a difficult market, a run of poor execution, a branch that has stopped working, or ordinary variance around an intact edge. Two accounts at minus fifteen percent with completely different underlying conditions get identical treatment, and the treatment is calibrated to the worse interpretation because the gate cannot distinguish them. That is a genuine loss of information and it is accepted deliberately: the diagnostic layers exist to answer why, they carry the resolution the gate discards, and none of them touches deployment. The system separates the question that can be answered exactly from the questions worth answering well.

The one place it bends

Even the objective number has a documented exception, and it proves the rule.

There is exactly one route by which the gate can read other than the arithmetic says, and it is worth setting beside the objectivity argument rather than hiding from it. Capital movements — a deposit, a withdrawal, a broker adjustment — change equity without any trading having occurred, which makes the comparison against the peak temporarily unrepresentative. In those cases a documented divergence is permitted. Notice what that exception is not: it is not a judgement about market conditions, about how the drawdown was earned, or about whether the account deserves more room. It is a correction to an input that stopped describing trading. The gate remains a subtraction; the exception exists because the subtraction occasionally has the wrong operands, and it requires writing down which ones and why.

The alternative that fails

A capital state assembled from several inputs would be a judgement wearing a number's clothes.

It is tempting to build a richer capital-state reading — combine drawdown with expectancy, efficiency and volatility into a composite health score, and route on that. The composite would be more informative and it would forfeit the property that makes the layer work. Each additional input carries an estimation window, a threshold, or a weighting, so the composite has parameters, and parameters are choices. The number produced would look as objective as a drawdown percentage and would be a considered estimate, which is exactly the situation in which an operator under pressure discovers that a different reasonable configuration produces a friendlier reading. The gate stays thin so that it stays unarguable.

• Drawdown from peak is nearly alone among trading questions in having one answer.
• Authority follows unarguability rather than predictive power, because the layer must hold under pressure.
• The gate knows nothing about cause — that is the diagnostic layer's job, and it touches no ceiling.

The key idea

Build the binding layer on the measurable thing, not on the important thing.

The instinct in designing a risk system is to govern on whatever matters most, which usually turns out to be something requiring estimation. The inversion here is deliberate and worth generalising: govern on whatever can be established beyond dispute, keep it narrow, and let everything that requires judgement inform rather than authorise. The result is a system whose hardest constraint is also its simplest calculation — which sounds like an underuse of the analytics and is the reason the analytics can be as rich as they are without ever destabilising the deployment decision.