How often tier performance gets re-examined — and why not more.

The arithmetic problem

Seven slices divide the sample seven ways.

The live rhythm produces sixteen trades a week — a respectable flow for whole-system metrics. But attribution divides that flow across the ladder, and unevenly: the workhorse zone accumulates evidence at a usable pace, mid-acceleration tiers at a crawl, and T6–T7 at a rate measured in quarters. A weekly attribution read on the thin rungs is therefore mostly a random-number generator wearing a table's clothes: two trades at T5 this week prove nothing in either direction, and a review habit built on proving things from them will generate an intervention urge weekly, forever. The cadence follows directly from the arithmetic — each zone gets examined no faster than its own sample refills.

The tiered rhythm

Weekly glance, monthly read, quarterly verdict.

The system's review architecture already runs on escalating depth, and attribution slots into it. Weekly: a glance, not an analysis — was tier usage gate-authorized, did any cycle deploy outside its resolved tier, is anything anomalous enough to note? No conversion conclusions; the week's job is compliance and anomaly capture. Monthly: the first genuine attribution read — usage distribution against the benchmark's tier tables, workhorse-zone conversion against its own history, directional notes on the acceleration tiers. Conclusions here may direct attention, not action. Quarterly: the verdict layer, where samples in the heavier zones approach usefulness — persistent patterns from the monthly reads get promoted to findings, and findings route onward to diagnosis or the scenario pipeline. Annual: the full-ladder retrospective against the benchmark's year-horizon expectations, the only cadence at which the rare tiers deserve any verdict at all.

Why not faster

Over-review is an intervention machine.

The case against tighter cadence isn't laziness — it's what over-review does downstream. Every examination is an opportunity to find a pattern, and patterns found in noise demand responses: a tweak to when T5 triggers, a hesitation before the next authorized deployment, a quiet distrust of a rung that did nothing wrong. Each response contaminates the very sample that would have exonerated the tier, so the system never accumulates clean evidence about anything — the observer keeps poking the experiment. The benchmark comparison makes the discipline concrete: simulated tier statistics are distributions built from full-year paths, and comparing a fortnight of live usage against a year-horizon distribution isn't conservative, it's dimensionally confused. The cadence isn't slower than ideal. It's as fast as the evidence gets made.

• A finding that would vanish with two more weeks of data was never a finding.
• The urge to check the thin tiers early is precisely the urge the cadence exists to absorb.
• Anomalies are always capturable immediately — cadence governs conclusions, not observation.

The exception protocol

What legitimately breaks cadence — and what doesn't.

The cadence bends for events, not moods. A gate-compliance breach at any tier is a same-day matter — that's governance, not attribution, and governance has no waiting period. A drawdown crossing into diagnostic territory pulls the relevant attribution slices into the formal review immediately, because capital defense outranks calendar. And a benchmark refresh resets the comparison baseline, which naturally triggers a fresh read against the new expectations. What never breaks cadence: a bad week at a favorite tier, a striking pattern two data points old, or the ambient itch to know now. The protocol's honest summary — anything worth breaking cadence for will have announced itself through a channel with more authority than curiosity.

The key idea

Cadence is how the review protects the reviewed.

Attribution's value depends entirely on sample integrity, and sample integrity depends on the reviewer's restraint — the discipline of letting evidence ripen at the pace the ladder's own structure dictates. Weekly compliance, monthly attention, quarterly findings, annual verdicts: the rhythm looks unhurried and is exactly calibrated, because it examines each rung precisely as fast as that rung generates truth. Anything faster reviews the reviewer's patience instead of the ladder's performance.