# Macro Factor Risk Models

> Build macro risk models that read the open web. NOSIBLE WORLD supplies 100M+ dated events in 95 languages across 30 years for text-derived factor research.

**URL:** https://nosible.com/macro-factor-risk-models

## Hero — Build risk models that read the world

For quants and macro risk teams.

**Build risk models that read the world.**

Millions of stories moved markets this morning. None reached your risk model.

Words move markets. Factors and fundamentals are downstream of text.

CTAs:
- **Start Trial** → [trial hub](https://nosible.com/start-trial)
- **See the research** → [#evidence](https://nosible.com/macro-factor-risk-models#evidence)

## §01 · The open web vs your risk model

A two-column contrast between the open web and a conventional risk model.

### The open web — 100M+
Events from 300K+ sources in 95 languages, dated to the minute and tagged to country, sector, and entity.

| Metric | Value |
| --- | --- |
| Sources | 300K+ |
| Languages | 95 |
| History | 30 years |

### Your risk model — 0.
Prices, fundamentals, and lagged macro releases from a few vendors. Text from the open web is not an input.

| Metric | Value |
| --- | --- |
| Inputs | Price feed |
| Cadence | Lagged |
| Reach | English |

## §02 · Model inputs vs the open web — Your model's inputs, and what they leave out.

Factors and fundamentals are real, but they are downstream of text.

### Left column · Your model (price-only and lagged)
1. **Equity prices** — Updated tick by tick
2. **Volatility surfaces** — Implied, daily
3. **Factor exposures** — Beta on price covariance
4. **Macro releases** — Lagged and revised
5. **Earnings** — Quarterly, footnoted
6. **Yield curves** — Closing snapshots

Price-only and lagged.

### Right column · The open web (dated to the minute, in 95 languages)
1. **Sanctions packages** — Posted hour by hour
2. **Central-bank speak** — 95 languages, all 169 banks
3. **Elections and protests** — Turnout, results, unrest
4. **Supply-chain ruptures** — Ports, strikes, recalls
5. **Regulator and court actions** — Filings the day they land
6. **Monsoon and crop outlooks** — Region by region

Dated to the minute, in 95 languages.

## §03 · Macro risk from text · 2025 — Macro risk, measured from text.

2025 papers showing how text from the open web carries macroeconomic risk, and how to extract it.

**// Index:** Papers 01-02 quantify geopolitical and policy uncertainty from text. Papers 03, 07-08 cover central-bank communication at scale. Papers 04-06 price text-derived shocks in equities and Treasuries.

1. **Measuring Geopolitical Risk using Artificial Intelligence** — Iacoviello & Tong (2025, Fed Board). An LLM reads five million newspapers since 1960 to grade geopolitical risk daily, beating the dictionary GPR. [PDF](https://www.matteoiacoviello.com/research_files/AI_GPR_PAPER.pdf)
2. **Narratives to Numbers: LLMs and Economic Policy Uncertainty** — Hartley (2025, arXiv). One multilingual LLM extends policy uncertainty back to 19th-century newspapers and across countries, beating Baker-Bloom-Davis at its design. [PDF](https://arxiv.org/pdf/2511.17866)
3. **From Text to Quantified Insights: LLM Analysis of Central Bank Communication** — Silva, Moriya & Veyrune (2025, IMF WP). A fine-tuned LLM tags 74,882 documents from 169 central banks since 1884 for topic, stance, sentiment, and audience. [PDF](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025109-print-pdf.pdf)
4. **War Discourse and the Cross Section of Expected Stock Returns** — Hirshleifer, Mai & Pukthuanthong (2025, JF). A war-discourse factor extracted from seven million NYT stories since the 1860s prices 138 anomalies across the cross-section. [NBER](https://www.nber.org/papers/w31348)
5. **Pricing the Global Trade Vulnerability** — Liu & Pan (2025, SSRN). A firm-level country-concentration factor commands a risk premium triggered by US-China tariffs, COVID, and 2025 Liberation Day shocks. [SSRN](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5939939)
6. **Tariff War Shock and the Convenience Yield of US Treasuries** — Acharya & Laarits (2025, NBER). April 2025 tariffs stripped Treasuries of their hedge property; safe-asset investors rotated to short maturities and gold. [NBER](https://www.nber.org/papers/w34640)
7. **Monetary Policy Shocks: A New Hope. LLMs and Central Bank Communication** — Fernandez-Fuertes (2025, Bocconi WP). A multi-agent LLM reads Beige Books and Minutes ahead of every FOMC, producing a cleaner narrative monetary shock. [PDF](https://rubenfernandezfuertes.com/papers/2025/2025__fernandez-fuertes__jmp.pdf)
8. **Mapping the Space of Central Bankers' Ideas** — Park, Perez-Cruz & Shin (2025, BIS WP). Embeds 19,742 speeches from 1,000 officials across more than 100 central banks into a regime-watching vector space. [PDF](https://www.bis.org/publ/work1299.pdf)

**Field note:** "If your risk model has no text, it has half the data."

## §04 · What you actually build — Three things you build.

Each is a query against the same dated event archive.

- **§01 Real-time text GPR factor** — A daily geopolitical risk index across 95 languages, covering sanctions, conflict, regulation, and elections. Feeds your VAR.
- **§02 Country and sector vectors** — Map every event to a country, sector, and named entity. Build long-short baskets that rebalance daily on dated event flow.
- **§03 Event-driven alpha overlay** — Weight risk by event severity instead of binary keyword counts. Decompose realised vol into price-driven and event-driven components.

## §05 · NOSIBLE WORLD · the archive — The dated event archive behind the factor.

Thirty years of dated events. 95 languages. Every event tagged to country, sector, and named entity.

| Stat | Value | Label |
| --- | --- | --- |
| Events | 100M+ | Events |
| Sources | 300K+ | Sources |
| Languages | 95 | Languages |
| Point-in-time | 30 years | Point-in-time |

### Named cases · risk-tagged

| Risk | Place | Date | Event |
| --- | --- | --- | --- |
| Geopolitical | Strait of Hormuz | 2026·03 | Iran declares the Strait of Hormuz closed after joint US and Israel strikes. |
| Fraud | United States | 2015·10 | Theranos exposed by a WSJ investigation flagged years earlier on the open web. |
| Macroeconomic | India | 2023·10 | India bans sugar exports for the first time in seven years; global food prices snap higher. |
| Liquidity | China | 2021·12 | Evergrande defaults after years of warning signs surfaced on the open web. |
| Geopolitical | Ukraine | 2022·02 | Russia invades Ukraine; energy and grain markets reprice across the world within minutes. |
| Trade policy | United States | 2025·04 | US announces reciprocal tariffs on Liberation Day; equities drop ten percent within days. |

## §06 · Get started — Wire NOSIBLE WORLD into your risk factors.

A pilot in 30 days. The same dated event archive the 2025 papers above use.

CTA: **Start Trial** → [trial hub](https://nosible.com/start-trial)

## Related
- [Home](https://nosible.com)
- [NOSIBLE Search API](https://nosible.com/search-api)
- [Research](https://nosible.com/blog)
- [Start Trial](https://nosible.com/start-trial)