What may move while a trade is open — and what may not.

The permitted moves

A short list, all pointing the same way.

The page records the legal mid-trade events precisely: stop moves are recorded, a reduced stop means reduced active risk, and break-even means zero active risk. Partials fill at their declared levels. The trail activates at its contracted unlock. What unites every item is direction — each one takes risk off the table or leaves it unchanged, and none adds. There is no permitted action that widens a stop, increases position size, or extends a target beyond the contract. The asymmetry is not an oversight in the list; it is the list's organizing principle.

The prohibited move

Widening a stop is the one action that rewrites the trade's identity.

Every risk number attached to a trade — its size, its slot in the cycle, its contribution to open exposure, its R denominator — derives from the stop distance declared at clearance. Widen the stop and all of them silently become wrong: the position is now risking more than the throttle authorized, the R multiple the trade will eventually report is measured against a denominator that no longer matches what was actually risked, and the cycle's exposure accounting understates what is at stake. The trade continues to look compliant in every record while having ceased to be the trade that was cleared.

Why floating R stays out of it

Unrealized profit has no vote in a management decision.

The demotion of floating R to context is what makes the improvisation boundary enforceable in practice. A management framework that permitted unrealized profit to inform decisions would license exactly the renegotiations the boundary forbids — a trade showing 2.8R invites a wider trail, a trade showing −0.6R invites a wider stop, and both invitations feel like responsiveness to new information. Floating R is not new information about the trade's edge. It is one path's current position, and it will move again before the trade resolves.

• Active risk governs deployment; floating R is demoted to context. The two are treated with opposite seriousness.
• The trade's contribution to the cycle is what it can still cost, never what it currently shows.
• A trade in profit and a trade in loss face the same rulebook — which is the entire point of writing it beforehand.

The unmonitored assumption

The boundary is easier to hold because the trades are pre-configured.

MARS assumes trades are configured through a trade-assistant EA and run without active monitoring, and that assumption does real work here. A management contract enforced by a human watching a chart depends on the human declining to intervene; a contract configured into the platform before entry simply executes. This is why conditional and momentum-dependent management structures were evaluated and rejected — not because they lack merit, but because anything that isn't a fixed price or R-level trigger reintroduces the moment of discretion the boundary exists to remove.

When the contract is wrong

A bad contract is a review finding, not a mid-trade correction.

Sometimes the management rules genuinely are wrong for the conditions — the trail is too tight for the regime, the break-even level is being reached and immediately stopping out, the partial is banking too early. These are real findings and they matter. They are also findings about the contract, and the place to act on them is the review, where the MAE/MFE Lab's capture and giveback metrics can establish whether the pattern is real across a sample. Fixing the contract mid-trade fixes one trade and destroys the evidence that would have fixed all of them.

The cycle consequence

Every reduction hands capacity back to the system immediately.

The monotonic descent of active risk is not merely a safety property — it is how the cycle refills. Because a trade at break-even carries zero active risk, it stops consuming pool capacity the moment its stop reaches that level, and the freed capacity becomes available to the next deployment through the throttle's smart open-exposure calculation. This gives the boundary a constructive reading rather than a purely restrictive one: the permitted moves are precisely the ones that return resources to the system, which is why the rulebook can afford to be generous about them and absolute about everything else.