Risk

Conceptual

Exposure measurement, scenario analysis and enforced limits.

Risk work is where confident-sounding output is most dangerous. Pimsy separates what is measured from what is assumed, and enforces limits in the runtime rather than in prose.

Analysis types#

Analysis	Produces	Key caveat
Exposure decomposition	Position weights by factor, sector, chain, counterparty	Classification choices change the answer
Scenario analysis	Outcome under named shocks	Scenarios are assumptions, not forecasts
Correlation structure	Pairwise and rolling correlations	Correlations converge in stress
Liquidity assessment	Cost to exit at size within a horizon	Depth data is venue-specific and moves fast
Counterparty mapping	Concentration across custodians and protocols	Off-chain relationships may be invisible

Runtime-enforced limits#

yaml

risk_policy:
  max_single_asset_weight: class="tok-num">0.15
  max_protocol_exposure_usd: class="tok-num">250000
  require_simulation_above_usd: class="tok-num">1000
  daily_notional_cap_usd: class="tok-num">100000
  forbidden:
    - unaudited_contracts
    - unverified_source_contracts
  escalate_to_human:
    - any_new_counterparty
    - drawdown_exceeds: class="tok-num">0.08

Last updated 2026-09-08