Daily, weekly, monthly, quarterly: what each cadence is allowed to conclude.

Why depth scales with period

Evidence accumulates at a fixed rate; conclusions must wait for it.

The rhythm produces sixteen trades a week — a rate that fixes how fast each kind of question becomes answerable. A day's evidence can support tactical observations: cycle P&L direction, exposure state, whether daily EV reads green. A week can support the first real verdicts: branch hit rates, weekly EV, adherence quality. A month can support pattern claims: capital dynamics, regime classification, benchmark placement. A quarter can support structural claims: drift, attribution findings, whether the model itself needs re-examination. Each cadence's conclusion ceiling is just the sample-size arithmetic made procedural — the calendar enforcing what the statistics require.

The map

Each speed, its instruments, and its ceiling.

The daily loop runs the execution instruments — checklist, console, gate state — and may conclude nothing beyond 'the day was operated correctly.' The weekly loop runs the closed-record instruments — Weekly Summary, EV Scorecard, Gate/Brake review — and may adjust tactical posture but not rules. The monthly loop runs the interpretive instruments — Capital Dynamics, monthly metrics, regime read, benchmark comparison — and may direct attention and open diagnostics. The quarterly loop runs the structural instruments — attribution, drift detection, scenario review — and is the first cadence allowed to produce findings that feed the promotion workflow. Nothing at any speed rewrites production rules directly; even quarterly findings route through the sandbox.

The violation patterns

Reading up and reaching down — the two directions of cadence error.

Cadence violations come in two shapes. Reading up: importing a conclusion from a faster loop into a slower one's territory — a red Tuesday treated as an edge question, one weak week treated as grounds to re-weight branches, a single monthly benchmark placement treated as drift. The faster loop's data simply cannot carry the weight, and the deviation-classification doctrine exists precisely to absorb this urge. Reaching down: letting a slow-loop anxiety contaminate fast-loop execution — trading Wednesday differently because the quarterly review is near, or hesitating on an authorized tier because a monthly diagnostic is open. The fast loops are supposed to be boring: execute, capture, respect the gate. Everything interesting has an appointment at a slower speed.

• One green daily EV read cannot override a restricted gate — the manuals state this as a hard rule, and it's the canonical reading-up error.
• Profile comparison explicitly warns: do not change profiles based on one week. That's the same rule, one speed up.
• If a fast-loop observation seems structurally urgent, the correct act is logging it for the slower loop — not concluding early.

The escalation channel

How genuinely urgent things legally skip speeds.

The map has exactly two lawful expressways, and both are narrow. Capital defense: gate transitions, brake states, and System Lock act immediately at any speed, because drawdown authority runs on arithmetic, not calendar — this is not a cadence violation but the one instrument designed to be cadence-free. And compliance breaches: a gate violation, an unlogged override, or a journal-integrity failure is a same-day matter regardless of when it's discovered, because governance has no waiting period. Everything else — however striking — takes the stairs. The test is simple: if the urgency is about capital state or rule compliance, act now; if it's about performance interpretation, it has an appointment.

The key idea

The cadence map is the system's statistics, worn as a schedule.

Underneath the calendar is one principle: every conclusion has a minimum evidence requirement, and the four speeds are just those requirements expressed as appointments. Operators don't have to re-derive the sample-size argument each time a bad day begs for a big conclusion — the map has pre-decided what today is allowed to mean. Run the speeds in their lanes and the system's judgments stay proportionate to their evidence at every horizon simultaneously. That proportionality, maintained for years, is what a governed track record actually is.