Method

The desk does not predict. It marks the record.

Observed facts, connected evidence, explicit confidence and dated falsifiers. The point is to teach the read — not to pretend certainty.

01 · Observe

Start with what happened.

Market moves, physical data, policy events and hard timestamps — before any interpretation. The gauges count tankers through chokepoints, containers through ports and cargo flights, and publish what they counted alongside the score.

See what is being measured→

02 · Connect

Build the causal chain.

What moved, what did not, and which link actually explains the consequence. Most commentary stops at correlation; the brief is written to name the mechanism, or to say plainly that the mechanism is not yet clear.

Read this morning's chain→

03 · Falsify

Name what would change the reading.

Every strong claim carries a dated condition that would make it weaker, early or wrong. A view without a falsifier is not a view — it is a mood. The condition is written down at the time, not reconstructed afterwards.

Find the falsifier in today’s brief→

04 · Record

Keep the receipts.

The dated call stays exactly as it was written. Nothing is quietly edited once the outcome is known, which is why the archive is worth reading backwards — the mornings we misread are still there, in the words used at the time.

Read the record→

The reading model

Price first. Narrative second. Never the reverse.

A headline explains a move after the fact; the tape records it while it happens. The brief reads the closes that cannot lie — the dollar, the front end, the long end, gold, crude — as one sentence rather than five numbers, then asks which of them disagrees. The disagreement is usually the story.

That is the whole method, and it is teachable. Every edition is written so that the reasoning is visible: you should finish knowing not just what we think, but how the reading was arrived at — and what would change it.

Measured, not surveyed: The gauges count physical activity — vessels, transits, throughput — not sentiment.
Confidence is stated: Each reading publishes the model’s own confidence rather than implying certainty.
Components are shown: The sub-scores behind every gauge are printed, so a score can be checked and not merely believed.
Dated, then left alone: Calls keep their date and their original wording.

Educational and informational only — not investment advice, and never a recommendation to buy or sell. Intelligence, not advice.