Where MacroRadar's numbers come from

A macro model is only as trustworthy as the data underneath it. MacroRadar uses a single rule: every input must come from a primary, authoritative source that anyone can check. If you have the same public data, you can reproduce every classification MacroRadar publishes.

We publish our own derived outputs — regime classifications, the recession-probability score, the sentiment index, composite indices, and the daily brief — and we link back to the original series rather than re-hosting raw third-party data. The distinction matters, and it is set out in full below.

MacroRadar redistributes its own work, not other people's. What we publish are derived and composite outputs — regime labels, the recession-probability score, the sentiment index, composite indices, and written interpretation — all computed by MacroRadar from the public inputs above.

We do not re-host raw third-party series as a substitute for the source. Where we display an underlying indicator, it is to provide context for our analysis, and the source is named and linked so you can pull the authoritative copy yourself. Indicator metadata — name, description, cadence, source link — is published freely; the canonical values live with the institution that produces them.

If you cite MacroRadar, cite the derived output and link the page. If you need the raw economic series, go to the originating institution in the table above — that is both the correct attribution and the freshest copy.

The pipeline refreshes daily, pulling the latest releases from each source. Because economic data is revised long after first publication, MacroRadar stores every observation with its vintage date — the date that value was first reported — so any value can be read as it stood on a past date. The as-of surfaces read those vintages; the published model histories use the series as later revised.

That point-in-time discipline is documented in full on the methodology page.

The input universe is deliberately compact. A broad indicator set is easy to assemble but hard to interpret; we favor a small set of well-understood series that produces stable, explainable classifications across the full historical sample.

All inputs to the regime and recession models are US economic data — that narrow scope is by design, not a gap awaiting a fix. Broader geographic coverage is a separate, live surface: /breadth publishes cross-country rankings today for two concepts — consumer-price-level (Canada, the euro area, the UK, and the US) and government-bond-yield (Canada, Germany, France, the UK, and Japan) — with policy-rate and production-volume registered across four economies each but currently suppressed because those economies' latest readings don't share a common publication date. The honest scope of the models, and what they cannot do, is set out on the limitations page.