Nodo Sur research
NEXUS SUR, verifiable decision support
A modular research layer combining source consensus, market state, equilibrium, signals, simulation, rival models and Policy Guard.
01
From data to decision
Each stage preserves its provenance and can fail explicitly.
- 01Independent and recent observations.
- 02Robust consensus excluding invalid, stale and outlier values.
- 03Normalized state and regime.
- 04Analytical equilibrium with confidence and band.
- 05Policy Guard and circuit breaker before any recommendation.
02
Learning without taking authority
Models are compared on equal terms, but the engine does not promote winners by itself.
- 01Synthetic or anonymized datasets depending on the phase.
- 02Reproducible metrics and minimum samples.
- 03Recommendation subject to human review.
03
Neural Engine integration
The existing Neural layer consumes NEXUS as a deterministic core.
- 01Input validated at the web boundary.
- 02Declared input provenance.
- 03Output always marked as decision support.
Common constitution
- No model signs, transfers, holds funds or executes operations.
- A recommendation never changes the monetary constitution.
- Every model promotion requires evidence and human review.
- Insufficient, stale or conflicting data produce a fail-closed response.
- Simulation, managed state and mainnet operation retain separate evidence.
Document status: verifiable internal research. Scenario values, where present, do not represent quotes or market availability.

