Understanding Parameters

Understanding parameters

This guide gives practical intuition for common CRML parameters.

For the portable semantics and document shapes, see:

• CRML Specification (Overview)
• Scenario schema

---

Frequency (lambda)

For a Poisson frequency model, lambda is the expected number of events per year in the configured basis.

• If basis is per_organization_per_year, lambda is the org-wide expected annual event count.
• If basis is per_asset_unit_per_year, the effective annual rate is scaled by bound exposure $E$.

See: Runtime (Frequency)

---

Severity (lognormal)

Common parameters:

• median: median single-event loss
• sigma: log-space spread
• currency: currency code

See: Runtime (Severity)

---

Simulation runs

Percentile estimates (VaR) stabilize more slowly than means.

• Use smaller --runs while iterating.
• Increase runs for reporting.