Radar Perene / Archive / science

The dossier that decides whether an index survives: backstage of a real decision

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

Science

New versions of indicators are usually born by acclamation: someone improves a formula, the new chart looks prettier than the old one, and the swap happens without anyone recording why — or under which conditions the decision would have gone the other way. Months later, when someone asks the reason for that change, the honest answer is that nobody remembers. The house went through one of those decisions recently, and what made it different was not the outcome: it was the document.

Before a new version of an index goes live, the house assembles a decision dossier: a document that gathers the evidence in favor, actively hunts for the evidence against, records the objections raised, and ends in an explicit verdict — proceed, revise or close. The index does not survive by inertia; it survives in writing.

Why a decision needs a document

The alternative to the dossier is not chaos — it is something worse, because it looks like order: the hallway decision. It is fast, it is reasonable in the moment, and it is unrecoverable afterwards. Without a record, there is no way to audit whether the version swap responded to evidence or to enthusiasm; no way to know which objections were considered and dismissed, and which were never raised; and no way to learn from the decision, because it left no material to study.

The dossier reverses that picture at a modest cost. It forces whoever proposes the change to write the full case — including the uncomfortable part, the evidence that does not cooperate. And it forces the decision to take one of three named forms: the version proceeds as is, the version goes back for revision, or the line closes. There is no fourth form, the market's most popular one: "let's see how it goes".

The recent episode, at the level of process

In the current cycle, a version of one of the house's indices went through exactly this rite. The dossier gathered the comparison between the candidate version and the previous one, confronted the performance of both against the usual rulers and — this is the point worth recording — incorporated a substantive objection raised during review. The objection was not treated as noise: it motivated revision before the final decision.

The numbers of the comparison and the text of the objection stay in the internal document, and there is a methodological reason for that: the public value of this episode lies in the rite, not in the score. What can be said is that the decision came out of the process bearing the three marks that distinguish it from an acclamation — evidence compiled before the verdict, an objection recorded and answered in writing, and a named outcome that any future review can find again and question.

Go/no-go as a phase of research

This rite is the natural continuation of a road that starts much earlier, in the conversion of a loose question into a testable hypothesis — the process we describe in How research is born. The hypothesis that survives testing becomes a study; the study that matures becomes an index; and the index, contrary to what the word suggests, is never finished. Each new version reopens the original question: does this still inform? The dossier is the shape the house gave to that recurring question, so it gets answered with the same rigor as the first time — not with the accumulated affection for an indicator that already has a name, an audience and a history.

One design detail makes a difference: the dossier actively hunts for the evidence against. It is not a matter of attaching whatever unfavorable material happened to appear, but of commissioning the attack — the same adversarial logic of testing against trivial rulers, now applied to the continuity decision. A version that was only defended was not evaluated.

What the reader can take from this

Whoever consumes market indicators rarely gets access to the backstage of versions — but can ask questions whose answers reveal the process. Has this indicator changed versions? What motivated the change? Is there a record of what the previous version did worse? Is there any documented objection, or is the official history a sequence of unanimous improvements? Real processes leave documentary scars. Stories without scars tend to be rewritten stories.

Frequently asked questions

What is a go/no-go decision?

It is a decision with named, mutually exclusive outcomes — proceed, revise or close — taken at a defined point in the process, based on material compiled before the verdict. The opposite of the decision that accumulates on its own, with no owner and no date.

Does the person who writes the dossier also decide?

In the house's rite, the dossier compiles the evidence on both sides and records the objections; the decision is made on top of it, not inside it. The separation between compiling and deciding is part of what makes the document auditable.

Does every new index version go through this?

That is the working standard: a new version without a dossier does not go live. The cost of the document is small next to the cost of a version swap nobody can explain a year later.

Why not publish the whole dossier?

Because it contains the comparative design and the house's working material — the part of the craft that sustains research on demand. The public commitment is to the rite and to the outcome, not to the spreadsheet.

---

Continue the trail: An index under revision: what a not-yet-deposited working paper means →

House reading: the versions that crossed the dossier are the ones producing each day's Diário.

Assembling a go/no-go dossier for a third-party indicator — commissioned attack included — is a service the house provides in conversation.