Historical news data: querying a news archive by date
Real-time is only half of news data — the other half is history. Backtesting a signal, building a training set, or studying how a story unfolded all need a clean, timestamped archive you can query by date. Here's how the archive works.
A clean, timestamped archive
Every article is stored with a precise fetched_at timestamp in UTC, alongside its full enrichment — urgency, political lean, topic/country tags and event cluster_id. The archive has grown continuously since it started, so historical records carry the same analytical fields as live ones. That consistency is what makes it usable for research and backtesting.
Querying a time window
The days parameter narrows any query to a lookback window, so you can pull "the last 7 days of high-urgency Ukraine defense news" in one call:
curl -H "X-API-Key: YOUR_KEY" \ "https://api.newsagentdata.com/v1/feed?country=ua&topic=defense&min_score=6&days=7"
Combine it with country, topic, language and lean filters to carve out exactly the slice you need — no full-dump-then-filter.
What you can build with it
- Backtesting — replay how an urgency threshold would have fired against past events (deterministic scoring means past scores are stable — see the methodology).
- Datasets — assemble a labelled corpus by topic/country/lean for analysis or model training.
- Narrative studies — pull a whole event by
cluster_idand watch coverage and framing evolve over the window.
History depth by tier
The free tier exposes a 1-day window (great for live prototyping); paid tiers open up longer lookback for archive work. The honest note: this is a growing archive, not decades of backfill — depth increases every day. Full parameters in the API docs, and per-country slices via the by-country guide.