Introducing Arche
Financial statements, modeled correctly
Arche is Protos Systems’ canonical financial data platform. It models financial statements as versioned, time-aware assertions to preserve restatements, revisions and provenance instead of overwriting history. The result is point-in-time accurate financial data you can trust in research, analytics and production systems.
- Point-in-time correctness.
- Query financial data exactly as it was known at any moment. No look-ahead bias, no retroactive overwrites, no silent drift.
- Explicit provenance.
- Every number is traceable to a specific filing, version, and effective date. Auditable by design, not bolted on later.
- Versioned financial primitives.
- Statements, metrics, and classifications are modeled as evolving assertions, preserving revisions and restatements without breaking downstream systems.
Restated
Temporal
Reconciled
⍺rche
Foundation, not feeds
A foundation layer, not another dataset.
Protos Systems provides infrastructure primitives for financial data, the way databases provide primitives for applications. Instead of mutable rows, Arche models financial truth as structured, time-aware assertions, making downstream analytics correct by construction, not by convention.
- Assertions, not rows.
- Financial statements are modeled as explicit assertions rather than mutable records. Values exist because they were stated, not because they overwrote history.
- Time-aware by design.
- Every assertion is bound to effective and knowledge dates, enabling precise as-of queries without look-ahead bias.
- Revisions preserved.
- Amendments and restatements are captured as first-class versions, allowing true before-and-after comparison.
- Provenance built in.
- Each number remains traceable to a specific filing, version, and source, making reconciliation and auditability inherent.
- Canonical primitives.
- Statements, metrics, and classifications are expressed as stable primitives that downstream systems can depend on.
- Composable by default.
- The model is designed to integrate cleanly into analytics, research, and production systems without downstream hacks.