Metrics

What ASA measures, and why it matters

Core Measurement Domains
Metrics Philosophy

ASA evaluates progress using a small set of interpretable metrics designed to capture trajectory quality rather than point performance. The system assumes noisy data, delayed feedback, and evolving objectives. Metrics are therefore constructed to emphasize stability, directional alignment, and recovery over time — prioritizing robustness and long-horizon signal over short-term optimization.

❋ Trajectory Stability Index (TSI)

Measures variance and coherence in inferred latent state over time.

  • Captures whether behavior is consolidating or oscillating
  • Penalizes reactive volatility and short-term reversals
  • Improves only through sustained directional consistency
❋ Directional Alignment Score (DAS)

Estimates alignment between observed behavior and stated long-horizon objectives.

  • Evaluated over rolling windows, not point decisions
  • Tolerates short-term deviations if trajectory improves
  • Emphasizes trend over precision
❋ Persistence & Recovery Ratio (PRR)

Quantifies follow-through after disruption.

  • Measures how quickly behavior returns to baseline after shocks
  • Separates structural failure from temporary deviation
  • Strong predictor of long-term outcomes
❋ Evidence Confidence Weight

Tracks the quality of inference rather than the state itself.

  • Accounts for data sparsity, bias, and inconsistency
  • Down-weights unreliable signals automatically
  • Prevents overconfidence early in observation