Reporting needs a record. Adjudicating needs a rule that existed first.

The distinction

One is a description; the other is a judgement against a standard.

A report states that a trade was taken on a given instrument, risked a given amount, and closed at a given result. Every element is retrievable from the record itself, and two competent people reading the same record produce the same report. A judgement states that the trade was non-compliant, or inside the expected envelope, or inefficient. None of those is retrievable from the record, because each requires a second input: the checklist that defined compliance, the distribution that defined expectation, the efficiency standard that defined conversion. Adjudication is the record plus a prior standard, and the standard has to have existed beforehand to mean anything.

Where each standard lives

Five verdicts, five standards, all of them written in advance.

Branch identity is judged against the branch definitions, which is why a trade's classification is fixed at entry rather than inferred from how it behaved. Compliance is judged against the pre-trade control checklist — approved asset, valid session, stop and target inside plan, risk authorised, capacity available. Expectation is judged against the Monte Carlo envelope produced by the branch mix. Efficiency is judged against the capture and giveback standards in the MAE/MFE Lab. Permission is judged against the gate row and the authorised pool. In each case the standard predates the trade, and could not be assembled afterwards without circularity.

Why the order cannot be reversed

A standard written after the result is a description wearing a verdict's clothes.

The temptation in every measurement system is to set the standard once the outcome is known, and it rarely feels like cheating — it feels like calibration. A drawdown band widened after an uncomfortable month, a branch reclassified once it is clear which bucket flatters it, an efficiency threshold relaxed because the previous one seemed harsh. Each move preserves the appearance of adjudication while removing its content, because a standard that adapts to results can never contradict them. This is the reason the benchmark refreshes on approved change rather than on an unrepresentative week, and why promotion out of the sandbox is a deliberate act. Both rules exist to make the retroactive adjustment inconvenient enough to require a decision, since it will never feel like one on its own.

The cost of the harder mode

Adjudication produces confident verdicts, including confidently wrong ones.

A report can be incomplete but is rarely misleading, because it claims little. A verdict claims a great deal, and inherits every weakness of the standard behind it. A poorly specified checklist generates non-compliance findings on reasonable trades. A benchmark calibrated to a branch mix the operator no longer runs classifies ordinary weeks as divergence. A confident wrong verdict is materially worse than no verdict, because it will be acted upon. This is the actual price of the stronger claim, and it means the standards themselves require periodic audit rather than permanent trust.

• Every judgement needs a standard that predates the trade being judged.
• A standard adjusted to fit results cannot contradict them, and is no longer a standard.
• Adjudication imports the weaknesses of its standards and states them confidently.

What this buys

A verdict can be argued with. A report can only be read.

The compensating benefit is specific: a judgement exposes its own reasoning and can therefore be contested on the reasoning rather than on the conclusion. An operator told a week was non-compliant can examine which checklist items failed and whether the items were correctly specified. An operator told the month sat outside the expected envelope can examine whether the envelope still reflects the mix being traded. Neither conversation is available with a report, which offers a number and no surface to push against. Adjudication makes disagreement productive by giving it somewhere to land.

The key idea

Determination is the whole claim, and it is checkable.

The difference between a journal and a governance system is not the number of metrics — a journal with a hundred columns is still reporting. It is whether standards were written before the events they judge, and whether those standards hold when the results are unwelcome. Both properties are inspectable by the buyer without trusting anybody: the checklists, definitions, envelopes and thresholds either exist ahead of the trades or they do not, and the record either shows them being amended by approved change or by disappointing weeks. That is a harder thing to fake than a metric count, and a considerably more useful thing to ask about.