Radar Perene / Archive / science

The only index with public weights: why the house opened one formula and not the others

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

Science

Anyone leafing through the house's working paper series finds an asymmetry that looks like incoherence: five indices published with concept, DOI and described method — but no weights — and a single one whose formula is fully open, component by component. Either the house believes in transparency or it does not; opening one and closing the others sounds like an unresolved compromise. The asymmetry, however, is the most deliberate part of the design.

The Brazilian Macro Regime Score is the only index in the series with explicit weights in the paper itself — a deliberate exception, not a leak. The criterion separating this case from the others is where the value comes from: when an index's value lies in the reading discipline applied to data anyone can reach, opening the formula strengthens the index; when it lies in the house's own treatment of the inputs, opening the formula would hand over the instrument without handing over what keeps it alive.

Two kinds of index, two contracts with the reader

A macroeconomic regime score works on public raw material: official series any reader downloads for free. What the index adds is not a secret datum — it is an organization: which series enter, with what relative importance, read under what discipline. In that case, hiding the weights would protect little and cost much: the reader would have no way to tell method from opinion. Opening the formula turns the index into a verifiable object — anyone can recalculate it, criticize the choice of components, propose another. The openness is the argument.

The other indices in the series make a different contract. In them, part of the value lives in the treatment the house applies before aggregation — construction decisions that are the accumulated result of attempts, errors and revisions. Publishing concept, question and DOI, but not the weights, is the honest way of saying: what this index measures is public and open to criticism; how it gets there is craft. The case of the risk index with a DOI and no public formula discusses that half of the choice; this article is about the other half.

The difference recalls the levels of research: different materials call for different degrees of exposure, and treating them all with the same ruler — everything open or everything closed — is less transparency than classificatory laziness. Mature transparency is being able to say why each thing sits at the level it sits.

What the exception is not

It is not a market test for opening the rest of the archive. The house does not promise that the other indices will have public weights one day — and resisting that promise is part of the criterion: an exception that becomes an expectation stops being a criterion and becomes a schedule. Nor is it a hierarchy of quality: the open index is not "more scientific" than the closed ones. All pass through the same internal review before reaching the repository; the difference is in the reading contract, not in the rigor.

And it is not, finally, a shield against criticism. The paper with open weights is precisely the most exposed of the series: any reader can recalculate it and disagree component by component. The house considers that exposure a fair price — for that specific index.

What the reader can verify today

The paper is on Zenodo with an active DOI, and the weights appear in the published text — a fact verifiable in minutes, without relying on the house's word. That immediate verifiability is what separates a declared transparency policy from a practiced one: the declaration is in this article; the practice is in the repository.

There is, finally, a pedagogical side effect the house values: the open-weights index works as a sample of the working standard behind the others. The reader who recalculates the open score and finds it sound gains an observable basis for calibrating the trust extended to the indices whose formulas they do not see — not a proof, but a sample; and verifiable samples are exactly what the asymmetry between provider and reader most calls for.

Frequently asked questions

Why not open the weights of every index?

Because their value does not live in the same place. Where the value is the reading discipline over public data, openness strengthens; where it is the house's own treatment of inputs, openness would hand over the instrument without the craft that sustains it. The criterion is per index, not a single principle.

Can an index without public weights be taken seriously?

The better question is: what does it offer instead? Published concept, DOI, preserved versions, open revisions and verifiable data allow substantive criticism even without the formula. What does not stand is the complete black box — no concept, no versioning, no record.

Doesn't opening the formula invite copying?

It does — and for this index, copying is tolerable by design: replicating the calculation does not replicate the maintenance, the revision and the history of decisions that accompany it. The house preferred the gain in verifiability to the cost of the replica, in this specific case.

Where are the weights?

In the text of the working paper itself, under DOI 10.5281/zenodo.21402939. The research page lists the full series.

---

Continue the trail: 16 spreadsheets, one hash: the public data package behind two indices →

House reading: today's reading is in the Diário; the archived episodes, in the Atlas.

Assessing whether a reader's own index or indicator could bear this standard of openness is bench conversation, not article material.