Trending Themes

Solutions

Products

Trending Themes

Emerging media themes, structured as quantitative data

Trending Themes detects statistically significant surges in global media coverage as they develop, and maps each theme to sentiment, relevance, and urgency scores across global entities. Delivered daily via API.

145K+

Mapped Entities

Languages Monitored

1,000+

Curated Media Sources

2005

Historical Coverage From

How It Works

A multi-stage pipeline combining classical NLP with large language models

The pipeline surfaces statistically anomalous narrative clusters and maps them to entities. The encoder-based NLP layer handles statistical detection; a generative AI layer labels and summarizes the output for human readability.

Text Ingestion

Millions of documents published across 1,000+ curated news and social media sources are processed daily through an NLP engine to uncover key phrases, entities, and sentiment scores.

Surge Detection

Three statistical methods identify key phrases with anomalous chatter volume. A phrase is flagged only when at least two methods agree, filtering out low-signal chatter.

LLM Theme Labeling

A large language model reviews the sentences surrounding the surging phrases and assigns a theme name, a narrative summary, as well as relevance and urgency scores.

Entity Mapping

Each theme is linked to the companies, countries, currencies, commodities, and other entities mentioned in its sentences. Every entity receives an individual sentiment score scoped to that theme, along with a buzz count reflecting how prominently it appears in the theme's source sentences.

Delivery via API

A structured output in JSON or CSV format is published daily via RESTful API with historical access available from 2005.

Real-World Examples

Themes detected as they emerged

These examples were detectable in the Trending Themes feed as the media clusters formed around each event.

Negative Contagion

Silicon Valley Bank Collapse, March 2023

As SVB coverage intensified, the feed detected a cluster of sentences co-mentioning SVB with counterparties holding deposits. Linked entities included potential acquirers and firms with similar risk profiles.

SIVB · First Republic · Signature Bank

Market Disruption

DeepSeek GPU Efficiency Narrative, Jan 2025

DeepSeek coverage triggered a statistically significant surge in negative-sentiment sentences associated with S&P 500 semiconductor names. The theme was detected as the media cluster formed, allowing quant models to respond on day one.

NVDA · S&P 500 · SMCI · AMD

Cross-Asset Spread

US Tariff Escalation, 2025

As tariff references in earnings calls reached multi-year highs, Trending Themes mapped the sentiment to individual equities, separating companies with positive associations (domestic manufacturers) from those with negative associations (importers, multinationals).

XOM · NKE · AAPL · USD/CNY

Positive Association

GameStop Short Squeeze, 2021

The GME squeeze produced a secondary theme: media co-mentioning of Evotec SE alongside GameStop created a statistically anomalous positive association. The entity mapping layer captured this cross-asset signal as the cluster emerged.

GME · Evotec SE · AMC · Reddit

Data Format

Two endpoints. Complete coverage.

Every daily delivery consists of two main structured datasets: Themes, one row per theme; Entities, one row per asset per theme. Below, one finds a few of the available fields in each endpoint:

GET /themes — Themes Endpoint

Field	Description	Example
Date	Timestamp of the recorded theme.	2026-03-19T11:00
Theme ID	Persistent alphanumeric ID of the theme.	CMBKXGE26SZQA
Theme	LLM-generated short name for the theme cluster.	US imposed tariffs
Category	Overarching category of the theme.	Economics & Macroeconomics
Summary	LLM-generated narrative summary.	The shortage in GPU is set to…
Emerging Days	Number of days since Theme is trending.	2
Buzz	Sentence count mentioning the theme in the past day.	139
Sentiment	Average sentiment of the sentences, from -1 to +1.	0.32
Relevance	How impactful is the theme to the financial markets.	0.80
Urgency	How quickly the theme demands attention.	0.90

GET /entities — Entities Endpoint

Field	Description	Example
Entity	Entity identifier. Resolves to name/ticker/other IDs.	AAPL
Theme ID	Persistent alphanumeric ID.	CMBKXGE26SZQA
Buzz	Sentence count mentioning this entity within the theme in the past day.	17
Sentiment	Average sentiment of sentences referencing this entity within the theme.	-0.96
Relevance	Relevance of this entity vs other entities within the theme.	0.80

Get Started

Request a data sample or trial access.

Contact us to set up a proof of concept for your asset universe. Sample data files and API access are available on request.

marketpsych.com/contact-us | [email protected]