Model Risk & Governance
Models are wrong in structured ways, so validation and governance carry the quant’s duty.
Every number above comes from a model, and models fail in structured ways: bad data, wrong distributions, miscalibrated parameters, or use outside the validated domain. Model-risk management builds the defense. Independent validation challenges assumptions and replicates outputs; challenger models run in parallel to expose blind spots; an inventory with change control keeps anything unvalidated off the desk; and governance in the spirit of Federal Reserve SR 11-7 sets ownership and escalation paths. The quant’s duty does not stop at correct code; it includes saying plainly when a model is stretched past its evidence. VaR, expected shortfall, stress tests, and CVA are only as trustworthy as the governance around them.