SIGNAL ATLAS / RESEARCH DESK

The evidence workbench.

Two small tools for exposing gaps and testing assumptions. No account, live data or submissions.

01 / CLAIM AUDIT

What is missing from the claim?

Check only what your evidence packet actually contains. This is an editorial checklist, not a reliability score.

A defined population, measure and time windowThe original source, not just a citation to coverageAn independently originated cross-checkA counter-explanation or disconfirming observationA check that time-sensitive facts still holdA decision threshold and review trigger

0 of 6 checks recorded. Start with scope and source.

Build the full decision brief · Check source independence

02 / ASSUMPTION LAB

How much depends on your probabilities?

A three-branch teaching model. Enter your own probabilities and impacts in one consistent unit. Defaults are hypothetical, not forecasts or investment recommendations.

BranchProbability %Impact / same unit

Downside

Base

Upside

Hypothetical weighted impact: 10 units.

Calculation: Σ(probability ÷ 100 × impact). Probabilities must sum to 100%; branches must be mutually exclusive and collectively exhaustive. An average is not an outcome. This omits tail risk, dependencies, utility, costs and timing.

Choose evidence that would change a branch

What these tools do not do

Nothing is sent or saved. Inputs reset when the page reloads. The checklist does not verify your sources, and the scenario calculator does not estimate probabilities or recommend a decision. Use both to surface assumptions before reading the underlying evidence.