Quantitative Analysis
Numbers that support decisions
Models, scenarios and risk metrics that the board understands and the supervisor accepts — combining robust statistical methods with artificial intelligence where it adds value.
Loss distribution
Illustrative simulation
Low losses
Tail
VaR 99%
Maximum expected loss
Expected Shortfall
Average tail loss
Stress scenarios
Plausible shocks
Português
What we do
Where quantitative analysis makes the difference
01
Stress testing and scenarios
Macroeconomic and idiosyncratic scenarios, reverse stress tests and sensitivity analysis, with impact on capital and liquidity.
02
Credit risk
PD, LGD and EAD models, IFRS 9 expected credit losses, scoring and internal ratings, and portfolio monitoring.
03
Market and liquidity risk
VaR, Expected Shortfall, ALM, interest rate risk and liquidity metrics for management and reporting.
04
Capital and solvency
Support for ICAAP, ILAAP and ORSA: risk quantification, capital planning and documentation.
05
Operational risk
Scenario-based quantification, loss data and indicators, linked to risk appetite.
06
Risk-adjusted profitability
Pricing, RAROC and capital allocation to guide commercial and investment decisions.
Quantitative + artificial intelligence
More predictive power, without losing control
AI can detect patterns that traditional models miss. We integrate it responsibly: benchmarked against classical methods, explainable and subject to validation.
Prediction
Early warning signals
Machine learning models that anticipate the deterioration of clients and portfolios before it turns into default.
✓
Credit early warning
✓
Liquidity and cash-flow forecasting
Detection
Anomalies and fraud
Automatic identification of transactions, losses or behaviours outside the norm, prioritised for review.
✓
Near real-time monitoring
✓
Fewer false positives
Scenarios
Scenarios and market signals
Scenario generation and text analysis (news, reports) to enrich stress tests.
✓
Richer scenario narratives
✓
Sentiment indicators
Principles:
Explainability
Benchmarking against classical models
Independent validation
Human oversight
Methodology
From business question to decision
Phase 1
· 1–2 weeks
Framing
We define the decision to support, the relevant metrics and the success criteria.
Phase 2
· 2–4 weeks
Data
Collection, cleaning and quality testing; we document sources and assumptions.
Phase 3
· 4–8 weeks
Modelling
Development, calibration and testing — classical and AI side by side.
Phase 4
· Ongoing
Integration
Models in production, dashboards, performance monitoring and knowledge transfer.
Indicative timelines, depending on model complexity and data availability.
Results
What you gain from sound quantitative analysis
Evidence-based decisions
Credit, investment and capital choices backed by evidence, not intuition.
Well-allocated capital
Knowing where risk pays off — and where it consumes capital without return.
Anticipation
Early alerts that give you time to act before losses materialise.
Credibility
Documented, defensible models for auditors and supervisors.
Frequently asked questions
Before you start
Do we need an in-house data science team?
+
Do supervisors accept AI models?
+
What data do we need to start?
+
Which decision would you like to make with more confidence?
Tell us about the problem. We will propose the simplest quantitative approach that solves it.
Book a diagnostic →
© 2026 Gryphon Consulting
Lisbon, Portugal · Europe