Radar Perene / Archive / science

From the analog engine to the signal registry: how the two converse

◦ Index methodology v2.2 (working papers with DOI). See the methodology.

Science

Whoever followed the last two trails of this series met two disciplines that, at first glance, live in separate buildings. In one, market signals are only born after maturing across fixed windows and only remain while they clear criteria that know how to retire them. In the other, the present is compared with the past by measured similarity, and the answer is always a fan of outcomes — never a prophecy. Signal registry on one side, historical analogy on the other. The natural question is whether these two buildings talk to each other. They do — and the conversation between them is more interesting than either alone.

A matured signal answers "something noteworthy has happened"; the analog engine answers "of the times something similar happened, here is what followed". The first discipline selects the moment; the second gives it context — and neither usurps the other's function. In the house's production, the two principles operate together; the technical integration between them is bench matter, but the logic of the division is public and explains itself in full.

Two questions that do not substitute for each other

Notice that the two disciplines answer questions of different natures. The signal registry is a filter of temporal relevance: amid the market's continuous noise, it says when a formalized pattern has cleared its criteria — when the present deserves special attention. But it does not say what that moment means: a matured signal is a dated fact, not a narrative.

Historical analogy is the inverse: a context machine with no sense of timing. Queried on an arbitrary day, it returns neighbors and outcomes of an arbitrary day — technically correct, informationally shallow. The fan of outcomes gains weight when the question is asked at a moment that stands out from the noise. A catalog of precedents is worth little without knowing when to open it; an alarm is worth little without knowing what to do when it rings. Each discipline carries exactly the half the other is missing.

The fit: the signal triggers the question, the archive answers

The natural chaining between the two is a one-way sequence. First, a signal clears maturation — a rare event by construction, since most candidates die in the windows or expire in the registry. That event defines a moment and a context: the environment's conditions at the instant the signal was born. Second, that context becomes a query to the archive: which past episodes resemble, by criteria fixed in advance, the environment in which this signal fired — and what came after them? Third, the answer returns in the format this whole trail has defended: a split of outcomes with frequencies in view, or the honest refusal when no valid parallel exists.

The final product of that conversation is not a prediction — it is a sentence with two anchors: "X happened, which is rare; and on occasions resembling this one, what followed split this way". Each half of the sentence was produced by a discipline that could not have produced it whole. How the baton passes — what data flows, in what format, with what safeguards — belongs to the two draft studies and stays there.

Why keep separate what could be a single system

There would be a tempting shortcut: fuse everything into a single system that detects and interprets at once. The house preferred to keep the two disciplines separate and conversing through a narrow interface, and the reasons are the same ones science uses to separate measuring instruments from interpreting instruments. Separate, each can be tested, audited and refuted alone: a defect in the maturation criterion does not contaminate the similarity ruler, and vice versa. Separate, each can die alone — if one fails validation, the other remains standing, and this series' history shows the house retires what fails. And separate, the border between fact and context stays visible in the final product, rather than dissolving into a single answer of inscrutable origin.

That architecture of independent, falsifiable modules is the same principle that organizes the house's production into steps — from hypothesis to study, from study to public text, as described in levels of research: each piece must stand on its own rung before joining any ensemble.

Closing the trail

This trail opened with a conversational phrase — "looks like 2018" — and dismantled it piece by piece: the plural output, the choice of neighbors, the reading by width, the refusal when the regime changes, the signatures that do not transfer, the toll the memory bias collects. The fit with the signal registry closes the circuit: the house's historical analogy is not an oracle consulted out of anxiety; it is context summoned by criterion. The next trail changes subject and scale — it leaves time and enters space: what happens to imported formulas when the market receiving them is too small for them.

Frequently asked questions

Does every signal maturation trigger a query to the analog engine?

The chaining principle is public; the operational rule of when each query happens is a bench decision of the two studies. What matters to the reader is the single direction: the signal summons the context, never the reverse.

Can the analog engine "confirm" a signal?

That is not its role. The fan of outcomes contextualizes the signal — it reports the historical repertoire of the environment in which it was born. Confirmation and refutation of signals happen where they always did: in the fixed windows and the registry's permanence criteria.

Will the two studies be published together?

They are separate studies, each with its own path to maturity, and that is by design — the independence that makes them auditable also makes them publishable one without the other.

Does this arrangement exist at other research houses?

Similar components exist at many; what is rarely seen is the declared border between detection and context, with each side falsifiable separately. It is that border, more than any component, that this trail set out to show.

---

Continue the trail: From 12 to 82 stocks: why American-style market breadth does not fit Brazil →

House reading: the episodes that feed the context are in the Atlas; the day's signals and readings, in the Diário.