FinanceLabArticle-level labels for financial news
Backtest and research signals on a labeled news feed, every article tagged by domain, impact, region, and tone.
Query
domain=monetary_rates
impact_min=high
region=US
start=2026-01-01
end=2026-01-15
sentiment=neutral
limit=20
Relevance score
0.91
Sentiment
neutral
{
"articles": [
{
"date": "2026-01-15",
"url": "https://example.com/central-bank-rates",
"title": "Central bank holds rates steady",
"source": "example-news.com",
"domains": ["monetary_rates"],
"impact": "high",
"region": "US",
"relevance": 0.91,
"tone": 1.8,
"sentiment": 0.04,
"sentimentLabel": "neutral"
}
]
}Structured news fields for research and backtesting
Use labeled article records instead of building a news-labeling pipeline from unstructured articles.
Screen news by label
Filter by date, domain, impact, region, and sentiment for research screens or monitoring views.
Feed monitoring dashboards
Use labeled article records without building your own news-labeling pipeline.
Build event-study datasets
Pull historical article records with relevance scores, labels, tone, and sentiment for event-window analysis.
Create backtest inputs
Compare research rules against labeled historical news using the same endpoint and response format as recent-news workflows.
Query from research notebooks
Pull compact article records for exploratory analysis and repeatable data pulls.
Sentiment as a queryable field
Each record includes a FinBERT sentiment score and label for filtering, grouping, and comparison.
Query with standard filters
Use date range, domain, impact_min, region, sentiment, limit, and offset to return only the records you need.
Start with recent records
Build against the last 30 days on Free, then upgrade to access the available historical archive.
Endpoint
GET /api/v1/news
Free window
days of recent news
Free requests
per month
Rate limit
per API key
From GDELT-sourced news to labeled records
Filter by date, domain, impact, region, and sentiment to pull exactly the articles your models need.
Ingest
GDELT-sourced global news is ingested regularly.
Label
Each article is enriched with domain, impact, and region labels, a relevance score, GDELT tone, and FinBERT sentiment.
Query
Call /api/v1/news with filters for date range, domain, impact minimum, region, sentiment, limit, and offset.
Use
Drop records into screens, dashboards, monitoring views, event-study datasets, backtest inputs, or research notebooks.
Start with recent records, upgrade for historical access
Both plans use the same endpoint, filters, and labeled article response format.
Free
For testing the API, prototyping, and adding recent news to an app.
- Last 30 days of news
- 1,000 requests / month
- 60 requests / min
- Article-level labels and relevance scores
- GDELT tone and FinBERT sentiment
Pro
For historical research, event studies, dashboards, and backtests.
- The available historical archive
- No monthly request cap
- 60 requests / min
- Article-level labels and relevance scores
- GDELT tone and FinBERT sentiment
FAQ
What is FinanceLab?
FinanceLab is a labeled news API for GDELT-sourced global news, served through /api/v1/news.
What fields are included?
Each record includes source, date, URL, title, domain, impact, and region labels, a relevance score, GDELT tone score, and FinBERT sentiment score and label.
What can I filter by?
You can filter by date range, domain, impact_min, region, sentiment, limit, and offset.
Is this real-time market data?
No. FinanceLab provides regularly ingested labeled news from GDELT, not real-time market data or price feeds.
What is included in Free?
Free includes the last 30 days of news, 1,000 requests per month, and 60 requests per minute.
What does Pro add?
Pro adds access to the available historical archive and no monthly request cap, with the same 60 requests-per-minute rate limit.
What can I use it for?
Use it for screens, dashboards, monitoring views, event-study datasets, backtest inputs, and research notebooks.
Add labeled news records to your research stack
Get a free API key and query recent labeled news in minutes.