Operator brief · 08

Regime classification: the system's weather report on itself.

The key idea

What is classified

Internal system state — not the market's mood.

The engine reads six structural inputs — gate pressure, structural health, gate volatility, transition count, persistence, and compression — scores each from −2 to +2, and sums them into a composite that maps onto a regime ladder: Expansion, Balanced, Transitional, Compression, Instability, Degradation, Drawdown Control, Critical State, and their variants. The subject is always the machine, not the market. It answers: how hostile are conditions to this system right now? That framing is deliberate: a system can be in perfectly good health during hostile markets, and in poor health during friendly ones.

FigureSix inputs, one composite, one regime

The subject is always the system's own internal state — never the market's direction.

The sub-label layer

Each regime carries exactly four allowed descriptors.

A regime alone says how bad; the sub-label says why. Instability driven by High Transition (excessive state-flipping) calls for cutting churn; the same regime driven by Low Persistence (failed follow-through) calls for shortening the trust horizon. The four-per-regime ontology is enforced by QA — a sub-label outside its regime's menu flags the classification as provisional rather than letting a mislabel steer posture. The response to a regime is almost never determined by the regime alone — it is determined by which of its four causes is driving it.

Why monthly

The cadence is chosen to outlast noise.

A regime read on a daily cadence would flip constantly and mean nothing; read annually it would arrive far too late to change anything. Monthly is the interval at which the six structural inputs accumulate enough evidence to be stable while still leaving room to act on the answer. It also matches the review rhythm the rest of the system already runs on, so the classification lands beside monthly performance, parameter review, and capital dynamics rather than as an isolated signal demanding its own attention. A regime that changes every month is itself a finding — usually Instability, and usually accurate.

Using the classification honestly

It changes how results are read, not whether they count.

The most valuable thing a regime label does is retroactive: it supplies the context in which a month's numbers were produced. A strong month earned during Degradation is genuinely stronger evidence than the same month during Expansion, and a weak month during Compression is weaker evidence of decay than the same month during favourable conditions. Without that context an operator is left comparing raw results across incomparable environments, which is how good systems get abandoned in bad months and mediocre ones get promoted in good ones.

The failure the engine guards against

Systems are usually abandoned in the wrong month.

Without a regime layer, a run of weak results reads as one undifferentiated signal: the system is not working. With one, that same run splits into two very different findings. Weak results produced during Expansion — favourable internal conditions, low gate pressure, stable structure — are genuine evidence of a problem, because the environment offered no excuse. Weak results produced during Degradation are substantially less informative, because the conditions were hostile and the system's job in hostile conditions is to survive rather than to perform. Reading both the same way leads directly to the most expensive mistake available: retiring a functional system during the exact stretch it was designed to endure, then rebuilding it from scratch in time for the next one.

Advisory by design

Regimes shape posture. Gates keep the hard authority.

The engine is deliberately interpretive: it produces a call-to-action stance — aggressive, neutral, defensive, protective — without overriding gate limits. Hard authority stays with the drawdown ladder because regimes are a model and models can be wrong. What the regime layer adds is the discipline of context: knowing whether this month's results were earned in Expansion tailwind or survived through Degradation headwind changes what those results mean.

Connected inside MARS

Regime Classification Engine →

Structural Diagnostic Engine →

Gate Pressure & Compression →

Every brief documents the same shipped system.

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

Get MARS Pro — $497 Explore the Platform