The variant axes: time and exposure, and the aggression they trade.

The two axes

Every variant is a position on time and exposure.

The five variants resolve into two independent dials. The time axis controls when the structure starts protecting: Time-Aggressive delays the first monetization — holding past the standard partial point to 1.3R, or delaying a trail unlock to 1.8R — buying more right-tail at the cost of a longer unprotected window. Time-Conservative advances protection — monetizing rungs at 0.75R and 1.25R before the standard points. The exposure axis controls how much of the position stays uncapped: Exposure-Aggressive banks less at the first event (33% instead of 50%) so more size rides the runner; Exposure-Conservative banks more, adding a 1.4R rung so less remains at risk. Standard sits at the origin of both axes. The full aggression ordering follows directly: Time-Aggressive, then Exposure-Aggressive, then Standard, then Exposure-Conservative, then Time-Conservative.

The locked table

Per branch, per variant: exact structures, not tendencies.

The doctrine's strength is that every branch-variant cell is a locked, explicit management structure — checkpoint prices, partial fractions, trail activations — not a vibe. Normal Time-Aggressive holds to 1.3R for the 50% bank, with an optional 2.5R extension target. Trend Partial Exposure-Conservative banks 50% at 1R and another 25% at 1.4R before trailing at 1.6R. Trend No-Partial Time-Aggressive delays its trail unlock to 1.8R — the single most aggressive cell in the system, flagged exceptional-only. Trend No-Partial Time-Conservative takes a single 33% at 1.25R and trails at 1.4R. Because every cell is specified, variant selection is a choice among known structures whose checkpoint hit rates the attribution layer can measure — which is the entire difference between a management system and a collection of exceptions.

Selection is evidence-gated

Aggressive variants are earned; defensive ones are cautioned.

The selection matrix pairs every cell with use-when and avoid-when doctrine, and the pattern is consistent: aggression must be earned through stacked evidence, defense must not become a habit. Time-Aggressive anywhere requires green branch EV, validated checkpoint hit rates (1.3R, 1.8R), clean ATR expansion, authority-timeframe confirmation, and a permissive gate — and Trend No-Partial Time-Aggressive adds 'exceptional only' on top. The defensive variants carry the opposite warning: overusing Time-Conservative in clean trends under-monetizes the right tail, and an over-banked trend branch becomes an over-defensive scalp. The failure modes are named in the matrix itself — using aggression because the setup feels strong is feeling replacing evidence; the variant note field requiring a written reason for exceptions is the enforcement.

• Every aggressive selection cites evidence: EV state, checkpoint validation, ATR, gate. 'Feels strong' appears only in the failure-modes table.
• Defensive selection is free but audited — capture efficiency and giveback metrics reveal chronic under-monetization.
• The variant note is mandatory for exceptions: selection reasoning enters the record with the trade.

The attribution loop

Variants are measured against Standard, always.

Because Standard is the origin of both axes, it's also the permanent control group: the attribution layer computes each variant's average R against the same branch's Standard over the same period — the variant-versus-Standard ratio — plus checkpoint hit rates, net-of-fees R, and capture/giveback quality. That closes the loop the axis system opens: a variant is a hypothesis that deviating from baseline in a specific direction pays under specific conditions, and the attribution tables are where the hypothesis meets its record. Variants that persistently underperform their Standard lose their claim to selection; variants that outperform earn wider (still evidence-gated) use. The five options stay five options — but their usage evolves with proof.

The key idea

Two axes turn discretion into a measurable coordinate.

Trade management is where most discretionary systems dissolve into per-trade improvisation that no analytics can grade. The variant doctrine solves it with structure: five locked positions on two named axes, per branch, each with entry conditions, failure modes, and a permanent comparison against baseline. The operator still chooses — that's the discretionary-quant bargain — but every choice is a coordinate in a measured space, and the space itself never moves. Aggression, in this system, is not a feeling. It's an address.