Analysis Lab

What this surface is

Analysis Lab is the premium async analysis workspace — “Saved deep-analysis history, benchmarking, and exports.”

The distinction that matters: Quick Analysis runs in your browser and gives you an answer now. Deep Analysis is queued server-side, runs at a higher sample count, and its result is saved as a run record. Analysis Lab is where those saved runs live, so you can compare a deck against its own past selves rather than only against its current state.

Where to find it in the app

Open a deck in Deck Lab, then use the deep-link /decks?deck=<id>&tab=analysis-lab.

Attention

Analysis Lab has no tab-bar entry. The Deck Lab Analytics dropdown lists Hand Draw, Analysis, Suggestions and Optimizer — Analysis Lab is not among them, even though the tab bar knows how to render an active state for it. Today the URL above is the only way in. This is an underexposure gap, not an intentional restriction.

With no deck selected the surface says so explicitly: “Select a deck from the Select Deck tab to open Analysis Lab.”

Entitlement

Deep Analysis is gated: “Deep Analysis is available on Slifer, Ra, and Obelisk bundles.” Without one of those bundles the surface renders its gate message rather than the run history.

Layout

Region	What it holds
Run History	Every saved run for this deck, newest first
Run Details	The selected run’s snapshot and metadata
Client Compare	A baseline run against a target run, metric by metric
Historical Benchmark	The selected run measured against a baseline run

An empty history says: “No deep-analysis runs yet. Queue your first server-side benchmark to populate Analysis Lab.”

Metrics

The compare and benchmark tables report the same six metrics plus a sample count. Rate metrics render as whole percentages; counts render as integers.

Metric	Meaning
Starter Access	Share of sampled hands that open with a starter
Dead Hand	Share of sampled hands with no viable line
Playable Hand	Share of sampled hands that can act
Starter + Extender	Share of hands with both a starter and an extender
Engine Online	Share of hands that get the engine going
Interaction	Share of hands holding interaction
Samples	How many hands the run sampled

Deltas in the compare table are signed — a leading + on an improvement.

Run lifecycle

A run moves through four statuses: queued, running, succeeded, failed. Each row in the history shows the deck version it ran against, the run date, the tier, the sample count, and the starter-access figure.

Actions

Action	Effect
Run Deep Analysis	Queues a new server-side run for the current deck
Rerun	Queues a fresh run against an existing run’s configuration
Export Report	Downloads the selected run’s report

Export reports “No report is available for this run yet” when the run has not produced one — a queued or failed run has nothing to export.

Related workflows

Analyze decks and scores Quick Analysis first; Deep Analysis when the answer needs to persist.

Related references

• Quick Analysis — the in-browser analysis surface
• Suggestions — turning a diagnosis into moves
• Deck Lab — the workspace Analysis Lab sits inside
• Scoring model — how the metrics are defined