Adaptive Regime Aggregation (ARA): A Design-First Architecture for Structured Balancing under Limited Information

Adaptive Regime Aggregation (ARA) is a design-first formal architecture for structured balancing under limited or conflicting information. Version 3 presents the consolidated mathematical specification of the ARA architecture, including formal convexity guarantees, regime-parameterized weighting, affine equivariance under global metric transformations, and explicit structural domain validity conditions. The method addresses decision-making environments in which multiple subsystems (e.g., data-driven models and structured intuition) must be coordinated within a shared metric framework. Rather than eliminating subjectivity, ARA structures it through explicitly declared regimes and controlled weighting sets. This version includes: • Strict convexity and uniqueness guarantees• Regime-parameterized weighting via constrained α-sets• Affine invariance properties• Explicit structural validity constraints• Two worked examples (probabilistic forecasting and multi-criteria aggregation)• Monte Carlo validation assessing robustness, flip-rate stability, and regret behavior under controlled perturbations All simulation scripts and CSV outputs required for reproducibility are included in this record. ARA is not proposed as a universal optimizer, but as a structurally transparent coordination architecture suitable for institutional and complex decision environments.