Loading
Geopolitical shocks, policy fights, and market stress are in the news before they are in prices. NOSIBLE Macro Risk Indicators turns that news into 30 daily risk indices your models can read.
Launching August 2026 · not yet available
What each one measures, why it matters, and the research behind it.
Four steps. No keyword lists. No black box.
Each of 100M events in NOSIBLE WORLD is stored as an embedding: a numeric representation of what the text says, in any language.
Each indicator is defined by a handful of plain-English anchor sentences. Events that match are in. No keyword lists, no per-article model calls.
Each deduplicated event counts once, weighted by how many distinct publishers covered it, normalised by a trailing 12-month baseline.
A paired uncertain-versus-resolved axis separates escalation from resolution. A keyword count scores a signed trade deal and a trade war the same way.
Monthly correlation against each published index, reported next to how well the official versions agree with each other.
Every anchor set is published. Grade us against the index you already run.
The published versions count keywords in a handful of English-language newspapers, once a month, with revisions. Ours read the whole web.
Monthly
Keyword counts over a handful of English-language newspapers. Published with a lag. Revised after the fact.
Daily
The same indices, rebuilt from deduplicated world events in 95 languages. Point-in-time, never revised, updated every 15 minutes.
Put 30 risk indices in your model.
Daily feeds, point-in-time history, published anchors. Pilot slots open now.
