Profit is not an input to the health score, and that is the point.

The omission

A health score that included profit would just be a lagging equity curve.

If profit were an input, a profitable month would score healthy by construction, and the score would tell the operator something they already knew from their account balance. Every genuinely useful thing the structural layer does depends on being able to disagree with the equity curve — to say the machine is deteriorating while the month was green, or holding up well while the month was red. A metric that takes profit as an input forfeits that ability at the moment of definition. The exclusion is not an oversight about what matters; it is what allows the number to be a second opinion rather than an echo.

The three that remain

Compliance, compression, gate improvement — all behaviour, none of it outcome.

What survives the exclusion has a common property: each of the three describes something the operator or the governance layer *did*, not something the market delivered. Compliance asks whether the plan was followed — a question with an answer that does not depend on whether following it paid this month. Compression measures how hard the capital envelope was squeezed, which is a fact about the drawdown path rather than the result. Gate improvement asks whether the capital state ended better than it started, which is the closest of the three to an outcome and is still a governance reading rather than a profit one. Together they describe how the machine conducted itself through whatever the month happened to be.

• Behaviour is controllable; outcome is partly not. A score built on behaviour is a score you can act on.
• All three inputs are available regardless of whether the month was profitable, so the score never goes blank in the months it is most needed.
• None of the three can be improved by getting lucky, which is what makes an improving score meaningful.

The uncomfortable readings

The score exists to produce the two combinations nobody wants.

Most months, health and profit agree, and on those months the score adds little. Its entire value is concentrated in the two disagreements. A profitable month with a poor health score describes a system that made money while breaching its own plan, running deep compression, or ending in a worse capital state than it began — money earned in a way that will not repeat safely. A losing month with a strong health score describes a system that followed its rules, contained the damage, and improved its gate position while the market simply did not cooperate. The first should worry an operator far more than the second, and no profit-inclusive metric will ever tell them so.

What it is not

Health is not a verdict on whether the edge works.

The boundary matters as much as the definition. A high structural health score does not say the strategy is profitable, has an edge, or should be scaled — those are expectancy questions answered by the scorecard and the EV layer, and the health score has no input that could speak to them. It is entirely possible to run a disciplined, compliant, well-governed system around a strategy with no edge, and the score will read healthy the whole way down. Structural health measures how well the machine is being operated, not whether the machine is worth operating, and conflating the two is the fastest way to draw a false conclusion from a genuinely well-built number.

The key idea

Excluding the obvious input is what makes the metric say something new.

The temptation with any composite score is to include everything that matters, which produces a number that correlates with everything and distinguishes nothing. Leaving profit out is a decision to answer a narrower question well: given whatever the month produced, was this machine operated in a way that deserves to keep running. That question has no other instrument in the system.