Three grade bands, not two — and the middle one is the only place judgment is authorised.

The three bands

Clear, contested, and refused, with the colours doing the work.

The score line is rendered in three states. Five or six is the high band — the components substantially agree and the continuation thesis has support. Three or four is the middle: real evidence, incomplete agreement. Two or below is refusal, and there is nothing to discuss in it. The bands are not statistical quantiles or anything so grand; they are a stated division of a six-point scale into ranges with different standing, and their value comes from being fixed before the week rather than from being finely calibrated.

Why not two bands

A binary scale relocates the ambiguity rather than removing it.

Setting a single cut point does not make the borderline cases disappear; it moves them next to the line, where they become arguments about whether a four should really count. The pressure lands on the threshold itself, and thresholds under pressure move. Three bands take the argument off the threshold and give it a home. A four is not a failed five. It is a four, in the band where the operator is expected to decide, and the decision is a legitimate part of the process rather than an exception to it.

What the middle band actually permits

It authorises a decision, not a smaller version of the trade.

The most common misreading is to treat the middle band as licence to take the same trade at reduced size. That is not what it means, and the reason is that size is not the indicator's to allocate — deployment belongs to the governance layer, which is reading gate state and open exposure rather than a chart score. The middle band permits exactly one thing: the operator deciding to trade or not trade, on the strength of the specific rows, in a state the instrument has flagged as underdetermined. The size that follows is whatever the rails say it is, unchanged by the score.

Why the refusal band has no middle

Two or below is not a weak setup; it is a different market.

The low band is where four or more components are failing, which means the conditions the strategy is defined against are largely absent. There is no reduced version of a continuation trade in that environment, because what is missing is not conviction but the phenomenon. Keeping this band hard is what makes the middle band affordable — an operator who knows the floor will not move can afford to exercise judgment above it without that judgment gradually eroding the floor. The two bands work as a pair, and softening either one costs the other its meaning.

• Two or below is an absence of the setup, not a version of it.
• Discretion above the floor stays cheap only while the floor stays fixed.
• The band a trade was taken in belongs in the record, alongside the score.

Recording the band

A judgment call that is written down can eventually be graded.

The band matters most after the fact. Trades taken from the cleared band and trades taken from the operator's band are two different populations, and comparing their outcomes over a sufficient sample answers a question no amount of reasoning will settle — whether this operator's discretion in underdetermined conditions adds anything. That is a genuinely useful thing to know in either direction. It requires only that the score at entry was captured, which costs a field, and it is the difference between discretion that improves and discretion that merely persists.

The key idea

Give judgment a designated place, or it will take an undesignated one.

Discretion inside a governed system is not eliminated; it is located. A scale with a marked middle says where operator judgment belongs and, by implication, where it does not — which is the whole content of the design. The alternative is not a more disciplined system but a less honest one, where the same judgment gets exercised at the threshold, unlabelled, and never appears in the record as something that could have been wrong.