Overview
Independent Validation for Quantitative Models
Shasat's model validation services provide rigorous, independent assessment of financial institutions' quantitative models — ensuring accuracy, appropriateness and regulatory compliance. Our quantitative experts apply a structured twelve-step methodology to validate ECL and CECL models, actuarial models, stress testing frameworks, pricing models and Python-based implementations.
Regulators including the EBA, ECB, Basel Committee, PRA and Federal Reserve require that material financial models are independently validated. Model risk — the risk of adverse consequences from decisions based on incorrect or misused models — is a significant operational and regulatory risk. Shasat's independent validation identifies model weaknesses, confirms mathematical integrity, assesses data quality and tests performance, giving institutions and their boards confidence in their quantitative frameworks.
Scope of coverage
Models We Validate
Shasat validates the full range of quantitative models used in financial reporting, risk management and regulatory capital calculation.
ECL Models — IFRS 9
PD, LGD and EAD models, staging models, forward-looking macroeconomic scenario models, collective and individual assessment models, and expected credit loss calculation engines.
CECL Models — ASC 326
Lifetime expected credit loss models under US GAAP including DCF, loss-rate, roll-rate and vintage analysis approaches. Challenger model build for peer benchmarking.
Actuarial Models — IFRS 17
Fulfilment cash flow models, risk adjustment calculation models, CSM unlock models and stochastic liability models under the General Measurement Model, PAA and Variable Fee Approach.
Stress Testing Models
Internal stress testing frameworks, regulatory stress test models (EBA, ECB, PRA), ICAAP models and reverse stress testing methodologies across credit, market and operational risk.
Pricing and Valuation Models
Derivative pricing models, fair value models, Level 2 and Level 3 valuation models, interest rate models (Hull-White, Black-Scholes), and credit risk pricing models.
Credit Scoring Models
Application and behavioural scorecards, internal rating models, probability of default models and credit risk classification systems for retail, SME and corporate portfolios.
Validation methodology
A Structured Twelve-Step Process
Shasat's model validation follows a comprehensive, structured methodology — aligned with regulatory expectations under Basel SR 11-7, EBA Guidelines on internal models, ECB TRIM and Solvency II internal model requirements.
01
Scope Definition
We define the validation scope, outline objectives, identify all relevant models, and map the applicable regulatory criteria — establishing clear boundaries for the engagement before work begins.
02
Model Documentation Review
We review model documentation exhaustively to verify completeness — assessing theoretical foundations, underlying assumptions, data inputs, model limitations and governance records to confirm the documentation supports the intended model purpose.
03
Independent Testing
We conduct independent testing to evaluate mathematical integrity, algorithmic appropriateness and implementation accuracy — including back-testing against observed outcomes, sensitivity analysis and stress testing to confirm model reliability under different conditions.
04
Data Integrity Assessment
We assess the quality and integrity of all data inputs — verifying relevance, completeness, consistency, accuracy and appropriate processing, and identifying data gaps that could affect model reliability or regulatory acceptability.
05
Comparative Analysis
We compare model outputs against benchmarks, industry standards and challenger models to validate performance, assess conservatism and identify material deviations from expected outputs.
06
Outcome Analysis
We scrutinise the predictive power and decision-making utility of the model — identifying and understanding deviations from expected outcomes, assessing discriminatory power, calibration accuracy and stability over time.
07
Governance and Use Assessment
We evaluate the governance framework surrounding model use — including approval processes, user training, model inventory documentation, change management procedures and the model's role in strategic and operational decisions.
08
Regulatory Compliance Review
Our validation adheres to regulatory standards including Basel Committee SR 11-7, EBA Guidelines on internal models, ECB TRIM framework, Solvency II internal model requirements, and PRA supervisory statements on model risk.
09
Python-Based Model Validation
We provide specialised validation for Python-coded models — verifying algorithmic implementations, testing coding practices, reviewing use of Python libraries (Pandas, NumPy, Scikit-learn, SciPy) and ensuring model alignment with regulatory and institutional requirements.
10
Comprehensive Validation Report
We prepare a detailed validation report summarising the entire process, findings, identified weaknesses, their materiality rating and specific improvement recommendations — with an overall model risk assessment suitable for board and regulatory review.
11
Recommendations and Remediation Support
We provide actionable advice for addressing all identified issues and actively support the institution through the remediation process — ensuring required adjustments are implemented correctly and in a timeframe consistent with regulatory expectations.
12
Ongoing Monitoring and Periodic Review
We establish regular review and continuous monitoring regimens to maintain model performance and compliance — ensuring models remain accurate, relevant and regulatory-compliant as market conditions, portfolio composition and regulatory requirements evolve.
Data validation
Data Validation and Data Services
Shasat provides specialised data validation and data services, ensuring accuracy and integrity in financial institutions' data management processes. Our approach addresses data governance, compliance and quality optimisation — the critical foundations on which reliable quantitative models depend.
Rigorous Data Validation
We verify the accuracy, completeness and consistency of data inputs — identifying and rectifying discrepancies in source data, transformation logic and aggregation processes to ensure models operate on reliable foundations.
Enhanced Data Services
We provide access to high-quality financial data sourced from trusted providers, including market data, pricing information, credit reference data and tailored datasets appropriate to specific modelling requirements.
Compliance Assurance
We align data management practices with regulatory standards — including BCBS 239 data principles, EBA data quality guidelines and internal risk data aggregation requirements — mitigating data-related compliance risks.
Strategic Data Insights
We provide insights to transform data into strategic assets — identifying patterns, gaps and improvement opportunities that support more reliable credit risk assessment, impairment estimation and regulatory reporting.
Continuous Monitoring and Updates
We establish ongoing data monitoring and periodic review processes to keep data relevant, accurate and effective — ensuring model inputs remain fit for purpose as portfolios, market conditions and regulatory requirements change.
Challenger Model Build
We build independent challenger models to benchmark primary model outputs, identify conservatism or weakness in key assumptions, and provide a quantitative basis for the model risk rating in the validation report.
Regulatory alignment
Accounting and Regulatory Frameworks
Shasat's model and data validation services support both IFRS and US GAAP accounting standards, and comply with the major regulatory frameworks governing model risk across banking, insurance and investment management.
Basel SR 11-7
Federal Reserve model risk guidance
EBA Internal Models
EBA guidelines on ECL and IRB models
ECB TRIM
Targeted review of internal models
Solvency II
Internal model approval and validation
PRA SS 1/23
Model risk management expectations
BCBS 239
Risk data aggregation principles
IFRS 9
ECL model compliance
ASC 326
CECL model compliance
IFRS 17
Actuarial model compliance
Technology
Python-Based Model Validation
Shasat provides specialised validation for Python-coded quantitative models — verifying algorithmic implementations, testing coding logic, reviewing library usage and confirming alignment with regulatory and institutional standards.
Python Core language
Pandas and NumPy Data processing
Scikit-learn ML validation
SciPy Statistical tests
Back-testing Historical performance
Sensitivity analysis Parameter stability
Stress testing Scenario robustness
Challenger models Benchmarking
FAQs
Common Questions
Independent model validation is a structured review of a quantitative model by experts who were not involved in building it. Regulators including the EBA, ECB, PRA and Federal Reserve require that material financial models — including ECL, stress testing and pricing models — are independently validated to identify weaknesses, confirm mathematical integrity, assess data quality and test performance under different scenarios. The Basel Committee SR 11-7 guidance sets out the supervisory expectation for robust model risk management including independent validation as a core component.
Shasat validates ECL models under IFRS 9 (PD, LGD, EAD, staging, scenario and calculation models), CECL models under ASC 326, actuarial models under IFRS 17 (GMM, PAA, VFA), stress testing models, credit scoring models, pricing and fair value models, risk parameter models, and Python-based quantitative models across all asset classes and product types for financial institutions globally.
A challenger model is an alternative quantitative model built independently of the primary model being validated, used to benchmark and test the primary model's outputs. By comparing challenger and primary model results across a range of scenarios and portfolio segments, validators can identify material weaknesses, biases or undue conservatism in the primary model's assumptions and methodology. Shasat builds challenger models as a standard component of its independent ECL and CECL validation engagements.
Key regulatory frameworks include the Basel Committee SR 11-7 guidance on model risk management (applied by the Federal Reserve and adopted broadly), EBA Guidelines on internal models and ECL estimation, ECB TRIM guidance for significant institutions, Solvency II internal model approval requirements for insurers, PRA Supervisory Statement SS 1/23 on model risk management, and BCBS 239 principles on risk data aggregation. Shasat's validation methodology is aligned with all of these frameworks.
Yes. Shasat provides specialised validation for Python-coded quantitative models — verifying algorithmic implementations against mathematical specifications, testing coding practices and logic, reviewing use of Python libraries (Pandas, NumPy, Scikit-learn, SciPy), assessing code documentation and version control, and confirming that the implementation correctly reflects the approved model design. Python validation is increasingly requested by regulators who require evidence that model code faithfully implements the approved mathematical framework.
Also from Shasat
Related Services
ECL and CECL Modelling
IFRS 9 and ASC 326 model build, PD, LGD, EAD, scenario analysis
Actuarial Risk Modelling
IFRS 17 liability valuation, risk adjustment and Solvency II
Credit Risk Valuation
CVA, DVA, FVA and XVA for OTC derivatives
IFRS 9 Consulting
Classification, measurement, ECL implementation and hedge accounting