Simply closer
to financial markets
Model Risk Management
An independent partner for model risk across pricing and risk models — and, in principle, every model in scope of SR 11-7 / SR 26-2, including AI / ML. From inventory and governance to validation, backtesting, stress testing, monitoring and the dialogue with your regulator, we help market participants run model risk as a managed lifecycle rather than a series of one-off projects, combining academic rigour with front-office and validation experience.
Our clients are banks, asset managers, insurers, funds and public organisations.
Model inventory
One authoritative register across pricing, risk and AI / ML models — owners, tiering, materiality, status, dependencies and validation history.
Documentation
Development and validation documentation produced to the standard supervisors require.
Independent validation
Independent review and challenge of pricing and risk models — soundness, implementation and limitations.
Stress testing
Scenario design and stress testing across the portfolio, feeding risk appetite and capital.
Risk-not-in-model
Identification and quantification of risks not captured by the model (RNIV / RNIME) and add-ons.
Audit tracking
Findings, action points and reserves captured in a register — classified, owned and tracked to closure.
Governance, roles & tiering
Three-lines-of-defence responsibilities, tiering and materiality frameworks, validation policy and approval workflows.
Data quality
Validation of market data, calibration inputs and risk-factor data — conventions, surfaces and curves, reproducible and auditable.
Backtesting & monitoring
Ongoing performance monitoring and backtesting of risk models (e.g. VaR / ES), with breach analysis.
Reserves & capital buffers
Model reserves, prudent-value adjustments and model-risk capital buffers (AVA, RWA add-ons).
Management reporting
Dashboards rolling validation status, open findings, reserves, limits and breaches up to a committee view.
Regulatory interaction
Submissions, responses to findings, expert reports and the ongoing supervisory dialogue.
AUTOMATION
High-quality, compliant-by-design workflows
A deterministic core plus agentic support cuts the manual effort across the lifecycle by 30–40% — every result traceable, the validator in control.
Worked Example
Model validation for FX pricing models
A concrete breakdown of an independent and highly automated validation — from theoretical soundness through to live-portfolio behaviour.
Get in touch!
Dr. Jürgen Linde / Jörg Zinnegger
modelriskmanagement@mathfinance.com
Address
Kaiserstr. 50
60329 Frankfurt am Main, Germany
