Move from scan setup to labels, trust review, and workbench follow-up across the signed-in research surface.
Use this guide to understand how the current research UI moves from context selection into scans, labels, explainability, and deeper follow-up surfaces.
Start in /research with watchlist, instrument, interval, date range, and quality tier set before you run a scan.
Treat the trust panel, explainability drawer, and regime context as part of the research loop, not post-processing extras.
Escalate into workbench, pipeline, experiments, and specialist research tools only after the primary scan and labels are coherent.
These research surfaces are the current signed-in workflow backbone exposed in the product.
Set context, run scans, review trust posture, inspect factor discovery, and build labels from the primary workspace.
Deepen the analysis after labels exist and you need a more focused review or follow-up workflow.
Use options, sizing, calibration, and other method-specific surfaces when the hypothesis needs a narrower tool.
Track reproducibility, experiment flow, and staged research work once the initial scan is worth preserving.
Fundamentals, macro events, news, and datasets provide the supporting context that shapes the scan and label review.
Follow this sequence so the research surface stays interpretable and the downstream evidence chain is clear.
Choose the watchlist, instrument, interval, date range, data tier, and optional news features before you run anything.
Start with the scan, then immediately inspect trust posture, regime context, and explainability so the output is credible.
Only build labels after factor and pattern outputs make sense for the current context and quality tier.
Move into workbench for deeper iteration or into backtests when the idea has enough evidence for validation.
The top half of the research page is where context, trust, and scan posture are established.
Use the form rows and context rail together so the current research run is anchored to a specific watchlist, instrument, interval, time range, and quality posture.
The trust panel and explainability drawer are part of the research loop, not reporting extras. The regime panel adds the macro backdrop before you over-read a scan result.
The draft callout is the handoff checkpoint between scanning and operational action. It is where a scan becomes a reviewable watchlist proposal.
Research mode changes what the page exposes. Advanced options, KR presets, and simple-mode summaries explain what the current workspace is allowed to do.
The lower half of the research page turns scan context into reusable outputs.
Primary actions gate whether the current context is ready to scan, build labels, or move into workbench follow-through.
Factor discovery keeps the current hypothesis tied to asset class, country, market segment, and symbol instead of turning into a generic score table.
The outputs panels are where factors, patterns, labels, feature matrices, and saved-view context are reviewed before anything is treated as durable research.
CSV or parquet exports, watchlist-draft application, and workbench or backtest handoffs should happen only after the outputs panel tells a coherent story.
Do not treat scan output as stable until the trust panel and explainability context are both acceptable.
Use the selected quality tier and news-feature setting deliberately because they change what the workflow is allowed to conclude.
Preserve watchlist, instrument, and range context when you escalate into pipeline, experiments, or workbench surfaces.
Prefer a clean handoff into backtests over re-running research with drifting context.
Run the signed-in research workflow with the current scan, trust, and label surfaces.
Return to the upstream discovery surface if the market context still needs refinement.
Move into run creation, inspection, compare, and evidence review once the hypothesis is ready.
Mar 24, 2026
Report unclear guidance, stale contracts, missing coverage, or broken docs UI on this page.
Open feedback issueJump to the section you need without losing your place.
Mar 24, 2026
Report unclear guidance, stale contracts, missing coverage, or broken docs UI on this page.
Open feedback issue