Skip to content
Montex Alpha Rail System
$497
← Back to The Comparison Stack

Operator brief · 176

Every row in the stack can flatter alone. None can flatter together.

The key idea

The design choice

A single-number comparison is a single point of failure.

Reduce the comparison to one figure and the review inherits every blind spot that figure has. Equity ahead of model is the obvious candidate and the worst one, because it is satisfied by outcomes the system exists to prevent: risk taken beyond authorisation, overrides used as habit, a branch mix drifted toward variance. The stack's answer is redundancy of a specific kind — not five measurements of the same thing, which would merely be reassuring, but five measurements that fail for different reasons. Expectancy can be strong while drawdown is abnormal. Drawdown can be clean while conversion is wasteful. Each row is a different question, and their disagreement is the signal.

FigureWhat each row can look good while hiding
Expectancy vs model82strong EV bought with abnormal drawdownDrawdown vs model58clean risk, wasteful conversion underneathProfit quality64quality intact, deployment velocity stalledRisk conversion61efficient per unit, too little deployedAcceleration88fast growth, override-driven and unearnedBehaviour34hardest row to fake, easiest to skipconcealment exposure

Schematic exposure score: how much a strong reading on this row alone can conceal, judged by how many other rows would have to be checked to rule out the flattering explanation.

The ranking rule

Clean behaviour with a flat result outranks a hot result with drifting discipline.

The stack carries an ordering that surprises people, and it is the module's own: a strong expectancy row alongside degrading tier discipline is a worse report than a flat expectancy row with clean behaviour. The reasoning is about what each state predicts. Flat results with intact discipline describe a system operating correctly through an unremarkable period, which is the majority of periods and reliably reverts. Strong results with drifting discipline describe a system whose governance is loosening while its outcomes happen to be cooperating — and outcomes are the thing least likely to keep cooperating. The second report is worse because it is a warning wearing a good month.

The single sitting

Reading the rows on different days is the same as reading one row.

The cadence rule specifies that the stack is read as a set at a scheduled review, and the reason is about attention rather than logistics. Rows examined separately are examined in a context set by whichever was read first, and the first row read is disproportionately likely to be the flattering one. By the time the behavioural row is reached — days later, after the encouraging expectancy figure has already been mentally filed — its job has changed from informing a verdict to overturning one, which is a much heavier lift. Simultaneity is what keeps the rows peers. Sequence quietly makes the first row the thesis and the rest the objections.

The no-cherry-picking property

A fixed row set removes the choice of which comparison to make.

The most consequential discretion in any review is not how the numbers are interpreted but which numbers get looked at, and that discretion is exercised before anyone notices they are exercising it. A fixed stack closes it. The rows are the same every period regardless of what happened, which means a month cannot be reviewed on its strongest axis and a weak axis cannot go unmentioned by simply not being raised. This is the same structural idea as the closed verdict vocabulary in the comparison workflow, applied one level earlier: constrain the inputs to the judgement, not merely the language the judgement is recorded in.

The behavioural row's status

Not a cross-check that overturns a verdict — a row that helps produce one.

Gate dwell, tier usage, throttle behaviour, and open exposure appear in this stack as a first-class row rather than as an audit applied to a result. The distinction is more than semantic. Treated as an audit, behaviour is consulted when something looks suspicious, which means it is consulted least often in the months where discipline is quietly loosening and results are fine. Treated as a row, it is read every period with the same weight as expectancy, and its normal readings accumulate into a baseline that makes an abnormal one legible. The row that is easiest to skip is the one describing the behaviour that produced everything in the rows above it.

The key idea

The stack's answer is the pattern across rows, never the best row in it.

Five metrics and a behavioural row produce something no single comparison can: a shape. Ahead on expectancy, adverse on drawdown, clean on behaviour is a different report from ahead on expectancy, clean on drawdown, drifting on behaviour, and the two demand different responses despite sharing their headline. Reading the stack as a set is what makes those shapes visible, and reading it any other way collapses six independent readings back into the single number the design was built to avoid.

Connected inside MARS

Live-Trade Benchmark Comparison →
Classifying Deviation →
Throttle Control Panel →

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