For the ML evaluation layer, a leakage-safe procedure for combining multiple model outputs into a single ranking was selected and independently audited before adoption. The exact combination itself stays closed by default; what is public is the discipline used to choose it.
Research
How the Layer Combines Model Outputs, Decided Carefully
A leakage-safe, independently audited procedure was adopted for selecting how multiple model outputs are combined into a single directional ranking.
← Back to Build History