FinanceLab
Market News Data APIDocs

Market news, already labeled for quantitative research

FinanceLab turns the GDELT news firehose into a clean feed of market-relevant articles, each tagged with domain, impact, region, and FinBERT sentiment. Query it in one call.

Get an API keyRead the docs
GET /api/v1/newsX-API-Key: flk_...

Query

domain=monetary_rates

impact_min=high

start=2026-01-01

limit=20

Relevance

0.91

Impact

high

{
  "articles": [
    {
      "title": "Central bank holds rates steady",
      "source": "reuters.com",
      "domains": ["monetary_rates"],
      "impact": "high",
      "region": "US",
      "sentimentLabel": "neutral",
      "relevance": 0.91
    }
  ]
}

Not a firehose, a clean feed

Raw GDELT is noisy and unlabeled. We add the parts that are expensive to build yourself.

Relevance-gated

A distilled model keeps only market-relevant news. Signal, not the firehose.

Faceted labels

Every article tagged with domain, impact level, and region, so you can slice by what matters.

FinBERT sentiment

Finance-tuned sentiment score and label on each headline.

Point-in-time

Append-only, with labels fixed at enrichment. No look-ahead bias.

Article-level rows

Each result includes source, date, URL, title, labels, sentiment, and relevance.

Research-ready

Build screens, event studies, dashboards, and backtests without a scraping pipeline.

Simple API surface

Filter by date, domain, impact, region, sentiment, limit, and offset.

Tiered history

Start on recent data, then unlock deeper history when the workflow is real.

covers

GDELT

worldwide news source

returns

12

domain facets

serves

JSON

through one endpoint

limits

60/min

per API key

From firehose to usable signal

The product value is in the pipeline: source coverage, relevance filtering, stable labels, and a small API surface that is easy to integrate.

1

Ingest

We pull worldwide news from GDELT in near-real-time.

2

Label

A distilled student model gates relevance and tags domain, impact, and region; FinBERT scores sentiment on the headline.

3

Serve

Each labeled article is queryable from /api/v1/news. Filter by date, domain, impact, region, or sentiment.

Simple pricing

Start free. Upgrade for full point-in-time history.

Free

For trying it out.

$0/ month
Get started

For first integrations and recent-news experiments.

  • Last 30 days of history
  • 1,000 requests / month
  • All labels + FinBERT sentiment
  • 60 requests / min

Pro

Point-in-time history for production.

$49/ month

For production workflows that need historical depth.

  • Full point-in-time history
  • Unlimited requests
  • All labels + FinBERT sentiment
  • 60 requests / min

FAQ

Where does the data come from?

GDELT, which catalogs worldwide news in near-real-time. We gate it for market relevance and add our own labels and FinBERT sentiment.

How fresh is it?

New articles are ingested and labeled continuously. Each row carries its publish date, so you always know when a story landed.

What are the rate limits?

60 requests per minute per API key. Exceeding it returns a 429 with a Retry-After header.

What's the difference between Free and Pro?

Free queries the last 30 days and includes 1,000 requests per month; Pro unlocks full point-in-time history and unlimited monthly requests. Both include every label and sentiment field.

Start building with labeled news

Create a key and make your first call in two minutes.

Get an API keyRead the docs