Full Tier by exception: the override doctrine.

The three modes

Smart by default. Compress under pressure. Full Tier by exception.

Smart Capacity is the default because it respects the pool: fresh deployment is resized around open exposure, so the account never carries more risk than the authorized tier intended. Compress Risk keeps trade count but shrinks per-trade risk to remaining capacity — the mode for high-conviction sequences inside a crowded book. Full Tier Risk forces full authorized risk per trade regardless of what is already open, potentially stacking exposure above the pool. It exists because rare conditions justify it. It is dangerous because most conditions that feel like they justify it do not. The asymmetry is the whole design: two modes negotiate within the pool, and one negotiates with the pool itself.

The legitimacy test

Three questions before the switch is touched.

An override is legitimate only when all three answers are yes — and the honest answer to the third is the one that fails most often: Any answer that requires a qualifier is a no.

• Is the gate state GREEN with drawdown well inside the guardrail? Override authority does not exist under a compressed gate.
• Is the open book independent of the fresh deployment — different pairs, different regime exposure — so stacking is nominal, not real?
• Would you document this override, with reasoning, before the trade — and be comfortable reading that reasoning after a full loss?

What the modes cost

Each mode buys something and sells something else.

No throughput mode is free. Smart Capacity buys exposure integrity and sells trade count — on a crowded book it will tell you to take three positions when you wanted four. Compress Risk buys trade count and sells per-trade magnitude, so a correct call pays less than it should. Full Tier buys full magnitude on every position and sells the pool ceiling itself, which is the only one of the three that can put the account above its authorized exposure. Reading them as a menu of preferences is the error. They are a menu of trade-offs, and only one of them trades away the thing the governance stack exists to protect.

The documentation rule

An override that was not written down did not happen for a reason.

The Cycle Decision Log exists so that override reasoning survives the outcome. Written before the trade, the reasoning is a hypothesis; recalled after a win, it is a rationalization, and after a loss it is usually silence. The log forces the sequence that keeps the doctrine honest — state the condition, state why it is exceptional, then act. The review value is downstream: a quarter of logged overrides is a readable record of whether the operator's exceptional conditions were genuinely exceptional, or whether ‘rare’ quietly became ‘most Fridays’.

• Log before the trade — the entry is a hypothesis, not a report.
• Name the specific condition, not the feeling that accompanied it.
• Review the quarter in aggregate: frequency is the real verdict.

The frequency test

Count the overrides, not the outcomes.

The single most useful review question about this doctrine is arithmetic rather than judgemental: how many times did the switch get touched this quarter? A genuine exception is rare by definition, so a log showing two overrides across sixty cycles is a doctrine working exactly as designed. A log showing eleven is not a run of unusual markets — it is a mode that has quietly become the default while retaining the language of exception. Outcomes cannot answer this, because a profitable override proves nothing about whether the condition was exceptional; a well-sized bad bet and a badly-sized good one look identical in the P&L column. Frequency is the only measure that survives the outcome, which is precisely why it is the one the review uses.

The cost of casual use

Every casual override reprices the whole system.

The exposure-governance stack works because the operator trusts it under pressure — and the operator trusts it because it is never bypassed casually. Each undocumented override teaches a quieter lesson than the trade's outcome: that the rails are advisory. Once rails become advisory, gate states become suggestions, drawdown guardrails become negotiations, and the system reverts to exactly the discretionary chaos it was built to replace. The doctrine is therefore behavioral, not mechanical: rare, deliberate, documented — or not at all.