Trust & security

2,300+ engine tests, every commit

Forecasting math is the product. We treat the engine the way an aviation team treats avionics: regression-tested every commit, with an explicit performance budget that fails the build if a change is unexpectedly slow.

• 2,300+ backend tests across the forecast engine, tax calculators (federal + 51-state + locality), authentication, Plaid integration, estate entities, and Monte Carlo trial determinism.
• 470+ frontend unit tests + 8 Playwright e2e specs covering the dashboard, accessibility on every Pro-feature page, CSP enforcement, and onboarding flows.
• Engine performance budget enforced in CI: a 480-month deterministic forecast must complete within a fixed time envelope per representative profile.
• Net-worth invariant test: balance-sheet identity (assets − liabilities = net worth) verified at every month of every forecast in every test.
• Mock-profile coverage: 22+ realistic household scenarios (retiree, sandwich generation, high-earner, single parent, fresh grad, immigrant tech executive, etc.) drive integration tests that exercise the full month-by-month loop.