Uneven rungs: why the regime ladder is finest near the middle.

The bands

Twenty-five points of composite score, cut ten ways.

Six metrics, each scored −2 to +2, sum to a composite ranging from −12 to +12. That range maps onto ten main regimes grouped into four bands of severity. At the top, Expansion occupies +10 to +12 and Expansion (Developing) occupies +7 to +9. The middle holds Balanced at +4 to +6, Transitional at +2 to +3, and Compression at 0 to +1. Below zero, Fragile Recovery sits at −1 to −2 and Instability at −3 to −5. The negative group runs Degradation at −6 to −8, Drawdown Control at −9 to −10, and Critical State at −11 to −12. Read as widths rather than labels, the pattern is immediately visible: three-point bands at the extremes, two-point bands through the middle.

Why the middle is finer

Resolution belongs where the state is genuinely in question.

A month scoring +11 and a month scoring +12 are, for every operational purpose, the same month: broad structural strength across the six metrics, nothing pressing to decide. Distinguishing them would produce a label difference with no behavioural consequence. A month scoring +1 and a month scoring +3 are not the same month at all — one is Compression, a neutral-to-defensive squeeze building beneath the surface, and the other is Transitional, a system rotating or handing off between internal states. Those carry different postures, different selectivity, and different review focus, and the two-point gap between them is where an operator most needs the engine to be precise. The ladder spends its resolution where the classification is doing work.

The zero crossing

Compression and Fragile Recovery straddle zero, and both are two points wide.

The tightest section of the ladder is the four-point stretch from +1 down to −2, which holds two entire regimes. This is not accidental. A composite hovering around zero is the state in which a month is most genuinely undetermined — the six metrics are not agreeing on a direction, and the same score can be reached by a system that is quietly tightening or one that is slowly recovering. The engine cuts finely here because the operational difference between Compression and Fragile Recovery is large: one warrants defensive posture as constraint builds, the other warrants patience while an incomplete recovery finishes. Coarser bands would collapse the two into a single indeterminate label, and an indeterminate label is what the engine exists to replace.

• Positive and negative are not the boundary that matters — Compression at 0 and Fragile Recovery at −1 are adjacent states, not opposites.
• Balanced is the widest of the neutral group at three points, because 'stable enough not to classify as stress' is a genuinely broad condition.
• The narrow bands are the ones an operator should expect to move between month to month.

Reading the position, not just the label

Where inside a band a month sits is real information the label discards.

Because the bands are unequal, a regime name alone under-describes the month, and the composite score should always be read alongside it. Degradation at −6 sits one point from Instability and describes a system that has just crossed into structural weakening; Degradation at −8 sits one point from Drawdown Control and describes a system approaching capital-defence territory. Both are labelled identically. The manual's stated reading order — QA flag first, then final label, then composite score, then primary driver — exists partly for this reason: the label compresses, the score restores what compression removed, and the driver names which of the six metrics is doing the pushing. An operator who reads the label and stops has taken the compression without taking the correction.

The key idea

A ladder's rung widths are a statement about what deserves attention.

Uniform bands would have been simpler to build and would have said that every part of the range matters equally. This ladder says something more specific: the extremes are already unambiguous and do not need to be subdivided, while the region around zero is where a month's identity is genuinely contested and where an extra point of precision changes what the operator does next.