Canvas

Canvas is the workbench. You build — or open — a graph: data legs, analytical operators and simulated institutions wired together with edges that carry the reasoning. One canvas holds the full causal chain from raw inputs to a portfolio-level readout.

The four-step flow

Every analysis follows the same spine, shown as the workflow bar at the top:

1. Design — load a template from the Component Library (left) or build from blank. Templates are complete, runnable graphs: FOMC debating gold, today's UK gilt network, the AI financing web.
2. Results — run the graph and read each node's output as a card. The context strip shows how many nodes executed and the run's Audit trail chip.
3. Backtest — ask how this machinery behaved on history: strategy tearsheets for trading graphs, historical readings for structural ones.
4. Build App — publish the finished graph as a locked, daily-refreshed Application.

The Component Library

Modules are organised in layers — data legs at the bottom, composite indicators and signal generators in the middle, strategy and allocation vessels on top. Every module has an ⓘ help entry and a reference card in these docs (see Modules).

Evidence

Every run is sealed into a tamper-evident record. The green Audit trail ✓ chip opens per-node fingerprints and a one-click export — see Trust & Provenance.