Skip to content
Montex Alpha Rail System
$497
← Back to System Architecture

Operator brief · 50

Why every decision publishes its own reasoning.

The key idea

The two failures

Blind trust and emotional override are symptoms of the same missing information.

An engine that emits only a tier invites two opposite pathologies. The first is blind trust: the operator deploys at whatever the box says without knowing why, which means they can't catch a bad input, can't notice when a stale reading is driving the answer, and learn nothing from years of decisions. The second is emotional override: the output feels too low after a good week or too high after a scary one, and with no visible reasoning to argue against, the operator substitutes their mood. Both are the same deficiency — the decision is opaque — and both dissolve when the engine shows its work.

The decomposition

Each layer's role, contribution, and warning, published per decision.

The engine's transparency table breaks the decision into its sources and names what each one did. Gate/DD as the hard capital authority defining the row and ceiling. Daily all-blended EV as the fast tactical throttle affecting cycle-level adjustments and never structural truth. The weekly structural read as the permission layer determining promotion eligibility. Category 6 as deep composite confirmation of whether promotion is genuinely deserved. Calculated diagnostics as the fragility and pressure overlay introducing caution, suppression, or veto. And the final engine as the synthesis — effectively the minimum of all caps, with veto logic applied. Every row carries its own operator guidance and log reminder, so the audit trail is generated as a byproduct of the decision rather than reconstructed afterward.

FigureThe engine's three blocks — intake, resolution, structural overlay
Intake & selectionwhat environment are we in?· Current gate & drawdown· Daily EV state· Weekly permission tier· Category 6 state· Cycle context & overridesCalculation & resolutionwhat would pure logic choose?· Structural modifiers applied· Raw core tier — pre-overlay· Manual override cap· Core permission tierStructural overlaydoes governance permit it?· Category 6 cap & permission· Diagnostic overlay cap· Hard veto check· Final structural cap

The decision engine's own layout mirrors the doctrine: context in, unconstrained logic resolved, governance overlays applied, then the final directive. Block hues identify the workbook's own colored regions.

The raw-versus-final distinction

Publishing the unconstrained recommendation is what makes governance visible.

One field in the engine does disproportionate work: the raw core tier — what the machine would have chosen from pure logic, before overlays, vetoes, and caps. Alongside the final resolved tier, it makes the governance layer's contribution literally measurable: a raw T6 resolved to a final T4 says the diagnostics removed two rungs of deployment this cycle, and names which overlay did it. Without that field, suppression is invisible; the operator sees a modest tier and can't distinguish 'the evidence was mediocre' from 'the evidence was strong but structure vetoed expansion.' Those are completely different weeks, and only one of them should feel disappointing.

What it protects

Auditability, learning, and the legitimacy of the override log.

Published reasoning pays out in three places. Audit: months later, any cycle's deployment can be reconstructed and defended, which is what makes attribution and drift detection possible at all. Learning: an operator who reads the chain each cycle develops a working model of the system's logic rather than a superstition about its moods. And override legitimacy: when a manual override is invoked, the engine's stated reasoning becomes the thing the override is arguing against — which forces the logged reason to be specific and makes casual overrides visibly indefensible. An override against a published chain is a documented disagreement. An override against a black box is just a feeling with better paperwork.

The key idea

A governed system has to be able to explain itself, or it won't stay governed.

Rules that can't be interrogated get abandoned the first time they're inconvenient, because there's nothing to weigh against the urge. The explainability engine ensures every deployment decision arrives with its own defense — the layers, the caps, the vetoes, the raw-versus-final gap — so the operator can see the governance working rather than merely obeying it. That visibility is what converts a rule set into something an operator can actually live inside for years.

Connected inside MARS

Authority Hierarchy · Why It Matters →
System Architecture →
Throttle Control Panel →
Authority Hierarchy →

Every brief documents the same shipped system.

The complete MARS package — eleven workbooks, three TradingView indicators, the full manual library — $497.

Get MARS Pro — $497Explore the Platform