> ## Documentation Index
> Fetch the complete documentation index at: https://docs.gdeltcloud.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Market sensitivity
> What `market_sensitivity` measures, what you can do with it, how it is computed, and what it does not claim.
How much market-relevant information this carries for a REASONABLE INVESTOR (the securities-disclosure materiality test) — NOT a predicted price move. 0-1, unsigned.
## What it is for
* Filtering an event feed down to what plausibly attaches to a traded claim — a contract, an obligation, a specific issuer.
* Separating market-relevant news from severe-but-unattached news, which is the distinction a headline volume count cannot make.
* Building an alert that fires on economic consequence rather than on drama.
* Reading a LOW score on a severe event as informative, not as a failure — see the worked examples.
## Filter on it
`market_sensitivity_min` and `market_sensitivity_max` on `GET /api/v2/events`, range `0`–`1`. Published only for the **cameoplus** family — filtering on it excludes every event outside that family.
## Formula
$$
\text{market\_sensitivity} =
\begin{cases}
0 & e \le 0.05 \\[4pt]
\dfrac{e + b}{2} & e > 0.05
\end{cases}
$$
| Symbol | What it is |
| ------ | ------------------------------------------------------------------------------------------------------------------------------------------------------ |
| $e$ | `claim_exposure` — the GATE. The smallest existing claim the event acts on that you can NAME from the article. At or below 0.05 the whole metric is 0. |
| $b$ | `economic_bite` — how far the event has travelled from words to a changed claim, for the claim named by `e`. |
Every sub-factor is clamped to its stated range before it enters the equation, and the result is rounded to **2 decimal places**. Reproducing a served value needs both.
## What the coder scores
The coder reads these off the article and records a reason for each. The scaler above turns them into the value — nothing else feeds it.
### `claim_exposure`
**Gate** — at or below `0.05` the whole metric is `0` regardless of the other sub-factors.
| Anchor | Meaning |
| ------ | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `0.00` | you cannot name even a jurisdiction — no economy stands behind this |
| `0.10` | A JURISDICTION ONLY — the finest claim nameable is the country's own debt, currency or index; the event happens inside an economy but the article states no fact about any particular traded thing (most local crime, protest, migration, outbreak and domestic-political events) |
| `0.35` | A PRICED AGGREGATE — a BASKET covering many products or operators, where no single instrument is the subject. BOTH (a) you name an industry-wide complex, an input, or a system-wide service that priced industries run on (semiconductors, power, transport, ports, shipping, payments, telecoms, agriculture), AND (b) the article states a change to THAT AGGREGATE's supply, output, price, cost or legal market access at its own scale. The aggregate needs no quoted futures price; (b) is the real test. Without (b) the rung is 0.10 |
| `0.60` | A NAMED CLAIM, INCIDENTAL — one specific claim is named in the article and the event touches it, but it is the venue, the setting, one of several, or scenery. A named claim is an issuer, borrower or counterparty OR a single exchange-traded instrument (one commodity, one currency, one sovereign's debt): the split from 0.35 is BASKET vs ONE INSTRUMENT, not company vs commodity |
| `0.80` | A NAMED CLAIM, AS THE SUBJECT — the event acts on it first-order: its holder is the one being fined, charged, bought, seized, approved, blocked or shut, OR the event directly sets the supply, price or availability of that one named instrument |
| `1.00` | THE TERMS OF THE CLAIM — the event redefines what the claim IS: what it pays, who owns it, or whether it may exist |
THE NAMING LADDER. ONE question: what is the SMALLEST existing claim this event acts on that YOU CAN NAME FROM THE ARTICLE? Each rung is a different answer to that one question, so exactly one fits. Take the HIGHEST rung whose test you can actually pass; you may not skip a rung you cannot pass. A claim is broader than a listed share: a listed parent counts (Taco Bell -> Yum! Brands); so do sovereign debt and currencies, exchange-traded commodities, CREDIT AND CONTRACTUAL claims (bankruptcy and litigation claims, an asset freeze, insurance exposure, a large supply contract), and a MAJOR PRIVATE COMPANY with a real claim (active secondary market, private-credit exposure, or a listed backer). Two checks before answering. (1) NAME IT OR DROP A RUNG — 0.35 requires naming the aggregate, 0.60/0.80/1.00 require naming the one claim (an issuer, borrower or counterparty, or one exchange-traded instrument); a `why` containing "not named", "not identified", "unnamed" or "no specific claim" is the reason for a LOWER rung. (2) DO NOT CONSTRUCT A CLAIM THE ARTICLE DOES NOT NAME — an insurer that must exist because a loss was insured, a supplier that must exist because a product is short, an industry that must exist because the victim worked in it: none is a named claim. And a named BODY is not automatically a named claim — a ministry, agency, municipality, municipal utility, public hospital, court or state broadcaster carries no traded claim of its own, so naming one leaves you on 0.10 unless the article states it has traded debt, a traded parent, or a claim held against it.
### `economic_bite`
| Anchor | Meaning |
| ------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `0.00` | NOTHING HAS BEEN DONE TO THE EXPOSED CLAIM — only words about it (statement, accusation, finding, demand, threat, ultimatum, warning, plan, intention, denial), or the action fell on something that is not the exposed claim |
| `0.20` | A PROCESS HAS STARTED — an instrument now exists in a file, docket or register (filed, issued, served, opened, charged, referred, scheduled, proposed, permitted for a distant date); nothing binds and no operation has changed |
| `0.40` | IN FORCE, BUT PARTIAL OR TIME-BOUNDED — covering one site, line, product, region or counterparty, or carrying a BUILT-IN expiry or sunset, or the effect has already ended. An estimated review date is NOT an expiry |
| `0.60` | IN FORCE ACROSS THE WHOLE CLAIM AND CONTINUING — no carve-out and no built-in end |
| `0.80` | 0.60 PLUS A SIZE TEST — large against that claim's core economics and ALREADY REALIZED (it has moved principal revenue, output, cost base or solvency; contracted-but-not-yet-operating is not realized). If "large" cannot be said from the article, 0.60 is the default |
| `1.00` | THE CLAIM IS NOT THE SAME CLAIM — legal status or basic economics changed (insolvency, liquidation, nationalization, licence revoked, default or haircut, peg abandoned, permanent destruction or write-off) |
THE DISTANCE LADDER. ★ THE CLAIM NAMED IN `claim_exposure` IS THE SUBJECT OF THIS SENTENCE — this asks what has ALREADY happened TO THAT CLAIM, not how dramatic, violent, tragic or costly the event was. Value taken, destroyed or lost BY the event is not a bite unless the thing taken IS the exposed claim. When claim\_exposure is 0.10 the only claim in play is the country's debt, currency or index, so absent a nameable sovereign-level act (default, devaluation, capital controls, a rating action) the bite is 0.00-0.20. ONE axis: how far has this travelled from words to a changed claim? Do NOT forecast, do NOT estimate a price move, and do NOT reason about what investors expected or whether it was priced in. "Economics" = cash flows, output or capacity, assets, contracts, licences and legal obligations, or access to and cost of financing.
## The frameworks it adapts
### [The reasonable-investor materiality test](https://supreme.justia.com/cases/federal/us/426/438/)
**What we take from it.** The question itself — whether there is a substantial likelihood a reasonable shareholder would consider this important, judged on the information available now and BEFORE any price move exists to measure. TSC Industries v. Northway, 426 U.S. 438 (1976).
**In current use by** [the SEC's cybersecurity disclosure rule, which makes an issuer apply exactly this test to a non-financial incident on a four-day clock with no price data — Form 8-K Item 1.05](https://www.sec.gov/files/rules/final/2023/33-11216.pdf).
### The full citation trail
The reasonable-investor materiality test, the question a disclosure lawyer answers while reading a document, before any price move exists to measure — TSC Industries v. Northway, 426 U.S. 438 (1976) ("substantial likelihood that a reasonable shareholder would consider it important"; "significantly altered the 'total mix'"); Basic Inc. v. Levinson, 485 U.S. 224 (1988); SEC SAB 99 (1999) and the 2023 cyber-disclosure rule (Form 8-K Item 1.05), where an issuer must judge a non-financial incident's materiality on a four-day clock with no price data — our exact situation; MAR (EU 596/2014) Art. 7(4) and ESMA ("appraise on the basis of the ex ante available information"); ASX Listing Rule 3.1 / Guidance Note 8; IFRS S1 and IFRS Practice Statement 2; and CJEU Lafonta v AMF, C-628/13 (2015), which holds that information can be "precise" WITHOUT indicating the direction of the price change — that is what authorises an unsigned score.
We ADAPT the structure and vocabulary of these frameworks. **None of the bodies named endorses, reviews, or is connected to this metric**, and their use does not make our values empirical measurements.
## Worked examples
Real events from the corpus, scored by the coder and frozen here on **2026-08-19** — check the arithmetic against the formula above. They are snapshots, not live lookups: an event can be re-coded, and an example that changed underneath its own arithmetic would be worse than none.
### Britain seizes Russian tanker Smyrtos in English Channel
*United Kingdom · 2026-08-13*
Near the top of both ladders. One named thing is the subject of the act, and what happened to it changed what it legally is.
| Sub-factor | Value | The coder's reason |
| ---------------- | ----- | ----------------------------------------------------------------------------------------- |
| `claim_exposure` | 0.8 | The named Smyrtos tanker is the specific property seized in the article. |
| `economic_bite` | 1 | The tanker’s legal and operational status changed through reported seizure and detention. |
$$
\frac{e + b}{2} = \frac{0.80 + 1.00}{2} = \mathbf{0.90}
$$
`market_sensitivity` **= 0.9** — what `GET /api/v2/events` returns for this event.
### Vessel owners reduce Strait of Hormuz traffic amid standoff
*Iran · 2026-08-11*
The reading people misread. A chokepoint carrying a fifth of the world’s seaborne oil scores 0.38 — because no single traded claim is named. It is the priced aggregate rung, not the named-claim rung. High systemic importance, modest market sensitivity: that is the metric working.
| Sub-factor | Value | The coder's reason |
| ---------------- | ----- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `claim_exposure` | 0.35 | The article names the Strait of Hormuz as a system-wide oil-shipping chokepoint and reports a material decline in traffic, but does not identify one specific instrument or operator claim. |
| `economic_bite` | 0.4 | Transit has materially declined but is not reported as fully halted; eight vessels still moved through the strait. |
$$
\frac{e + b}{2} = \frac{0.35 + 0.40}{2} = 0.375 \;\longrightarrow\; \mathbf{0.38}
$$
`market_sensitivity` **= 0.38** — what `GET /api/v2/events` returns for this event.
### Prague decides fate of Stalin Center: ban, end or continued operation
*Czech Republic · 2026-08-14*
The gate overriding a non-zero bite. `economic_bite` is 0.40 — something real happened — but `claim_exposure` is 0, so nothing traded is exposed and the answer is 0. Averaging the two would have invented a 0.20.
| Sub-factor | Value | The coder's reason |
| ---------------- | ----- | --------------------------------------------------------------------------------------------------- |
| `claim_exposure` | 0 | No traded, contractual, or investment claim is directly identified. |
| `economic_bite` | 0.4 | Restoration has a bounded operative effect by permitting the cultural center to continue operating. |
$$
e = 0.00 \le 0.05 \;\Longrightarrow\; \text{market\_sensitivity} = \mathbf{0}\quad(b = 0.40\ \text{never enters})
$$
`market_sensitivity` **= 0** — what `GET /api/v2/events` returns for this event.
## When it is `NULL`
Never `null`. A gated 0 is the TRUE value — "nothing traded is exposed" — not an unknown. Instead the metric is ABSENT from the response entirely, in two cases: a conflict-family event (the four CAMEO+ metrics are omitted rather than returned as 0, because a 0 would assert "measured, and it is the minimum"), and an event coded before the `v2-2026-07-24` contract.
These are the readings the definition does not support — building on one of them is building on a promise nobody made.
* NOT a predicted price move, NOT a probability, NOT an expected return, and NOT a magnitude of market reaction. It is how much market-relevant INFORMATION the event carries for a reasonable investor. The field name oversells the construct; treat the value as an ordinal ranking signal.
* NOT signed and NOT directional. It cannot tell you whether the world got better or worse — CJEU Lafonta v AMF (C-628/13, 2015) is precisely the authority for scoring materiality without a direction. Direction is `goldstein_scale`'s job, not this metric's.
* NOT a severity score. It measures how directly the event attaches to a claim that already exists, which is why a national blackout scores low here and high on everything else. Reading it as severity is the single most common mistake made with these metrics.
* A 0 is NOT unknown. `claim_exposure ≤ 0.05` means nothing traded is exposed, so 0 is the TRUE value.
What applies to all four metrics — the frameworks we adapt, how thin the registry coverage is, and the smallest single-event difference worth reading — is on [Limits and error bars](/metrics/limits).