Backtesting and Calibration

The methodology is designed to be empirical and adaptive.

• Initial thresholds and weights are calibrated using historical protocols and known incidents.

• We monitor:

  • How AAA/AA/A protocols perform vs. BB/CCC when stress events or exploits occur

  • Incidents per 10 protocols per rating band (or similar normalized metrics)

• We periodically adjust the framework to better align with real-world outcomes.

We commit to publishing annual calibration reports summarizing:

• Rating distributions and changes over time

• Observed incidents and losses by rating band

• Methodology adjustments and their rationale