Atlas

Appearance

Atlas for researchers ​

This guide is for researchers — academic, policy, think-tank — using Atlas for analyses where reproducibility, provenance, and citation discipline matter as much as the finding itself. The emphasis differs from the trader guide (urgency, intraday) and the investor guide (decision quality on a portfolio): researchers prioritize building defensible, durable artifacts.

ZonePanelWhy
Main canvasTimeline & eventsThe chronological backbone for any longitudinal study
Left railNewsSaved screens with stable filter sets
Right railCharts & graphsReusable, versioned visualizations
Pop-outSecurity dashboardDeep-dive when a single name anchors the study

Save as a custom dashboard called "Research" so it loads in one click.

Core principles ​

1. Save everything; reference by name ​

Every search, screen, graph, and dashboard layout has a save mechanism. Use it. A saved object has a stable name and a permalink that survives sessions — a one-off filter applied at 3 PM is gone by 4 PM.

2. Capture provenance at the moment of use ​

For any datum you'll cite:

  • Symbol / entity: as Atlas displayed it (Atlas symbol, MMSI, callsign).
  • Timestamp of capture: the wall-clock when you read the value, not the upstream timestamp (which Atlas also shows).
  • Upstream provider: Finnhub / EODHD / OpenSky / MarineTraffic / news source name.
  • Atlas version: from the Changelog — the active release.
  • Permalink to the panel state.

Atlas is a tool for finding and cross-referencing. The citation in your paper is to the upstream provider, with Atlas as the access method.

3. Beware revisions ​

Upstream providers revise data. EODHD parses fundamentals on its own cadence; AIS positions can be retroactively cleaned by MarineTraffic; news outlets sometimes silently edit articles after publication. Snapshot what you cite at the moment of citation — screenshot, downloaded JSON, or a permalink with a captured timestamp.

Research workflows ​

Longitudinal study (months to years) ​

Goal: track a phenomenon over a long window — e.g., dark-fleet AIS behavior under sanctions; sectoral capital flows during a regulatory change.

  1. Define the entity set up front. Save it as a screen (Advanced search for symbols; Globe → saved subset for vessels/aircraft).
  2. Set up dated checkpoints. At each checkpoint, capture the same panels with the same filters.
  3. Use Charts & graphs to build series that update at each checkpoint.
  4. Don't rely on raw "what's the value now" — relies on availability today; use stored snapshots.

Event study (focused window) ​

Goal: characterize what happened around a single event — e.g., the price reaction to a regulatory announcement, the AIS behavior during a port closure.

  1. Pin the event date in Timeline & events.
  2. Build the event window (T-N to T+N) and capture the panels at each end.
  3. Cross-reference Security dashboard → Disclosures for filings near the event.
  4. Use AI chat to summarize commentary around the event date — but treat the summary as a research note, not a primary source.

Comparative study ​

Goal: compare entities (countries, sectors, vessels of a fleet, aircraft of an operator) on the same metric over the same window.

  1. Build comparison series in Charts & graphs — multi-symbol price, multi-metric fundamentals, or operator-grouped tracks.
  2. Standardize on a single normalization (rebase to 100, log scale, percentage change) and document it.
  3. Save the graph and reference its permalink in the writeup.

Exports and packaging ​

What you can export ​

  • Per-panel PDF — model portfolios and Security dashboard sub-panels generate self-contained PDFs.
  • Permalinks — for any panel state.
  • Screenshots — out-of-band; use the OS screenshot tool against the active panel.
  • Graph permalinks — MyGraphs survive sessions.

What's not yet exportable ​

  • Raw underlying data (CSV / JSON) is not surfaced through the UI for most panels. For programmatic access, use the Atlas API (see API reference) — it returns the same JSON the panels render.
  • Bulk timeline export is roadmapped but not shipped.

For a deeply data-driven paper, plan to use the API alongside the UI: UI for exploration, API for the captured numbers in your tables.

Citation patterns ​

Suggested citation forms (adapt to your style guide):

Vessel position data: MarineTraffic AIS feed, accessed via Atlas v0.1.0, captured 2026-04-12 14:32 UTC. MMSI 123456789. Atlas permalink: https://atlas-terminal.io/?….

Equity quote: Finnhub real-time feed, accessed via Atlas v0.1.0, captured 2026-04-12 14:35 UTC. AAPL last $193.42. Atlas permalink: <…>.

Aggregate analysis: Author's calculations from Atlas Charts & graphs, MyGraphs ID study-2026-fleet-velocity, captured 2026-04-12. Underlying data: OpenSky ADS-B feed.

Atlas is the access method, not the source of authority. The provider remains the primary attribution.

Operational security ​

Even academic research can attract attention from the entities you study. Read OSINT operational security before publishing on sensitive subjects — separate Atlas accounts per investigation, 2FA, the network-level OPSEC checklist.

For multi-author research, prefer permalinks over shared accounts. Permalink recipients see the view, not your credentials.

Reproducibility checklist ​

Before submitting:

  • [ ] Every datum has a captured timestamp, upstream provider, and Atlas version.
  • [ ] Every saved screen / graph / dashboard mentioned in the paper has a stable name.
  • [ ] Every chart has a documented normalization.
  • [ ] All AI summaries are flagged as such; quotes from transcripts/articles cite the original source.
  • [ ] Any figure derived from a permalink has a screenshot stored offline (the underlying data may shift on later loads).

Released under the project license. Public sources only — no proprietary or restricted data.

Atlas