Hidden recovery trading: the failure Overflow was designed to make visible.

Why this branch

Recovery trading needs a legitimate label, and Overflow has the loosest one.

An operator who wants to trade their way out of a losing stretch faces a practical obstacle: every branch in MARS has an identity that a marginal trade will fail. Normal wants a clean static setup. The trend branches want confirmed expansion and an evidence stack. Overflow wants additional high-quality opportunity — a definition elastic enough that a determined operator can fit almost anything inside it. That elasticity is not a drafting error; some latitude is what makes supplemental flow possible at all. But it means Overflow is where the pressure lands, and the branch is governed accordingly.

The loop

The pattern is self-reinforcing, which is why it needs catching early.

Hidden recovery trading is not a single bad decision — it is a loop that tightens. A loss creates the impulse. Overflow absorbs it, because Overflow will take the trade. The additional trades add friction and variance at exactly the moment the account is least able to absorb either, which produces more losses, which strengthens the impulse. Left alone for a few weeks, the loop moves the account toward gate demotion, and gate demotion withdraws the aggressive variants that the operator was implicitly counting on to recover the deficit. The trap closes from both directions.

The signature

Timing, not count — the abuse fits comfortably inside a compliant number.

The most important operational point about this pattern is that it does not require a quota breach to be doing damage. A week with three Overflow trades is unremarkable if they are distributed across sessions that qualified. The same three, compressed into the two sessions immediately following the week's largest loss, are the pattern. Overflow entries clustering after losing days is the signature, and it is legible in sequence long before it is legible in totals. This is why the audit reads the journal's timing data alongside the count rather than treating compliance as a number that either passes or fails.

• Cluster after a loss → behavioral finding, regardless of whether the count was compliant.
• Extended holds inside a static-exit branch → duration-mismatch review; Overflow should close short.
• Large negative R inside Overflow → named in the lab's red-flag list; the branch's structure should not permit it.

The prosecutor's eye

What the weekly audit actually examines.

The review is deliberately adversarial rather than confirmatory. Trade count against quota establishes whether participation drifted. Fee and swap drag per trade establishes whether the branch's cost profile still resembles short static holds. R distribution establishes whether any single Overflow trade produced an outcome the branch's structure should have made impossible. And timing establishes the pattern above. The Weekly Summary tracks Overflow count and contribution separately from the blend for exactly this reason — folded into an aggregate, a bad Overflow week disappears inside a decent one from the other three branches, and the behavioral signal is lost precisely when it matters.

The honest frame

This is a control against a normal human response, not against a character flaw.

It is worth stating plainly that the impulse this control targets is close to universal and does not indicate a defective operator. Wanting to recover a loss quickly is the expected response to a loss; a system that assumed otherwise would be designing for a person who does not exist. What MARS does is refuse to rely on the impulse being resisted. The branch's quota is counted, its timing is logged, its costs are tracked, and its contribution is reported separately — so that the pattern becomes an observation in a weekly review rather than a discovery made three months later in a drawdown post-mortem. The control works because it does not require willpower to function.