Radar Perene / Archive / science

What \"existing\" means for a market signal: deadline, not intuition

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

Science

"I saw that signal last week." The sentence circulates in any market conversation and nobody asks what it means — because the answer would embarrass the conversation. Saw it where? Noted in what? Under which rule? With what date? In most cases, the "seen" signal exists in the same place premonitions exist: in a memory that updates itself as the outcome arrives. Before discussing whether signals work, there is an earlier and harder question: what is it, exactly, for a signal to exist?

In the house's discipline, a signal exists when it meets a formal birth criterion: a condition defined beforehand, verified with data of known date, recorded on the spot with a timestamp — and subjected to the maturation period the rule prescribes. Before meeting the criterion, what there is is an observation: legitimate, useful, and with no right to a certificate.

The boundary between observing and recording

The distinction sounds bureaucratic and is epistemological. An observation is a mental state: someone looked at the market and found something noteworthy. A recorded signal is a public object inside a system: it has a birth date, the rule that generated it, the data that supports it and a place in a registry that accepts no retroactive editing. The observation can be right and the record empty — or the opposite. What the discipline demands is that nobody confuse the two states, because all possible statistics live only in the second.

The practical consequence is hard on folklore: "existing" stops being a matter of conviction and becomes a matter of deadline and paperwork. The signal someone "was sure about" in March, but which was not recorded in March, did not exist in March. What existed was a convinced person — which is something else, and does not enter the count.

The criterion, at the level that can be declared

The house's signal registry operates with an explicit birth criterion, and its shape can be described without numbers: an objective condition on data of known date, defined before any particular case; verification done against the information available on the date, not against the memory of it; the timestamp applied at the moment of recording; and the maturation period — the subject of this trail's first text — running from there. Four requirements, none negotiable case by case. The parameters that fill each requirement are the registry's content and stay on the bench.

What deserves underlining is what the criterion excludes on purpose: intuition as a certificate. Not because intuition is worthless — the house's best objective conditions began as someone's intuition —, but because intuition has no auditable date. The path exists and is a single one: the intuition becomes a declared condition, the condition enters the rule, and the next cases are born under it. Intuition legislates for the future; never backwards.

Why so much ceremony over a word

The short answer: because every performance evaluation depends on it. Whoever debates whether "the signals work" without first fixing what existing means is measuring a population that redraws itself according to the result — the winners enter retroactively, the losers never quite "really existed". No hit rate survives such a population.

There is also a deeper reason, which the reader of The efficient market hypothesis will recognize. Any signal claim is a strong claim: that there was usable information the price did not yet contain. Strong claims call for evidence with a date. A market that is hard to anticipate coexists perfectly well with a crowd of "seen" signals — memory manufactures as many as needed. What it does not manufacture is a record stamped before the outcome. That is why the birth certificate is the only document that matters in this discussion: it is the difference between testing a hypothesis and collecting memories.

The test any reader can apply

Facing any "signal" offered in public, three notary questions: what was the rule before this case? Where is the date stamp? What did this source record that did not work out? Serious sources answer all three without discomfort, because the answers are their product. Folklore sources answer with biography — how long they have operated, how many hits they remember. The biography may be true; it just is not a record.

Frequently asked questions

What is a signal's birth criterion?

It is the formal condition, defined before the cases, that separates observation from recorded signal: an objective condition, data of known date, a timestamp at the moment of recording and a maturation period running from it.

Is an unrecorded observation worth nothing?

It is worth something as raw material: it can inspire a new condition, which applies to the next cases. What it cannot do is be promoted retroactively — the rule it inspired does not reach back to it.

What are the parameters of the house's criterion?

They are the registry's content and stay on the bench. The criterion's shape — the four requirements described above — is public; the values that fill it are working material.

Doesn't this rigor eliminate too many signals?

It eliminates most of them — and that is the function. What remains is little and has a property no large collection of "seen signals" has: it can be counted, audited and refuted. Only on that is any statistic possible.

---

Continue the trail: Pre-registration in practice: declaring the test before running the model →

House reading: the dated records this discipline produces feed the Diário and sustain the patterns of the Atlas.

Writing the birth certificate of a specific signal — the condition, the data, the deadline — is an exercise the house does alongside, when sought out.