The audit is two weeks away. The regulator has asked why the Stage 2 provisions jumped last quarter. The ECL model was built by an external vendor three years ago, the analyst who understood it has left, and nobody in the room can explain how the PD curve was calibrated.
If this sounds familiar, you are not alone. Many banks and NBFCs run critical risk models they do not fully understand. In a world of tighter supervision, faster-changing portfolios and AI-driven lending, that is no longer just an operational weakness — it is a risk in itself.
This post looks at why owning your models matters, which capabilities a modern risk function needs, and a practical roadmap to build them in-house.
The hidden cost of outsourced, black-box models
Vendor models and one-time consulting builds solve a deadline. They rarely build capability. Over time, institutions run into the same problems:
- Explainability gaps – the team can run the model but cannot defend its assumptions to auditors, the board or the regulator.
- Slow change – every recalibration, new product or overlay needs the vendor back, at a cost and on their timeline.
- Key-person risk – knowledge sits with one or two people; when they move, the model becomes a black box.
- Weak validation – independent validation is hard when no one internally understands the build.
- Missed business value – models built only for compliance are not used for pricing, limits or strategy.
What a modern risk function needs to own
| Area | What the team must be able to do |
|---|---|
| Credit risk & IFRS 9 / ECL | Build and recalibrate PD, LGD and EAD models, staging logic, scorecards, forward-looking scenarios, low-default portfolio approaches |
| Capital & ICAAP | Assess material risks, run stress tests, link capital planning to risk appetite |
| Liquidity & ILAAP | Liquidity statements, LCR and NSFR, behavioural modelling of deposits and products |
| IRRBB & ALM | Interest-rate risk in the banking book, repricing gaps, fund transfer pricing, hedging |
| Market & counterparty risk | VaR and Expected Shortfall, FRTB, sensitivities, SA-CCR and CVA |
| Model risk management | Independent validation, back-testing, documentation and governance |
| AI & machine learning | Default prediction, early-warning signals, anomaly detection — with responsible-AI governance |
No single hire covers all of this. Capability has to be built across the team.
A practical 4-step roadmap
Step 1 – Map the gap
List every model you use (ECL, scorecards, ALM, ICAAP stress tests, pricing), who owns it, who can explain it and when it was last validated. The red cells on that map are your training priorities.
Step 2 – Build a common foundation
Give analysts, credit officers and treasury staff a shared language: risk concepts, regulation and hands-on modelling in Excel and Python. Transparent Excel builds help teams see every step before they automate it in Python.
Step 3 – Go deep where it matters
Send specialists through domain-depth programs — credit risk modelling, ALM and liquidity, market and counterparty risk, quant and AI — that end with them building models, not just watching slides.
Step 4 – Move from training to ownership
The last mile is rebuilding or validating your own models with expert guidance on your own data, so the team ends with working models, documentation and the confidence to defend them.
How Peaks2Tails partners with institutions
Peaks2Tails was founded by risk practitioners Satyapriya Ojha and Karan Aggarwal, both FRM® and CQF®, with one goal: bridge the gap between risk theory and real-world implementation. The firm works across Training, Solutioning and Mentoring, and has trained 5,000+ professionals across Asia. It has received the Best Credit Risk Training Program and Liquidity Risk Solution of the Year awards.
Institutions that have engaged Peaks2Tails for training include Axis Finance, CIBIL, ICICI Bank, PNB, Synchrony and Union Bank of India.
Flexible engagement models
| Format | Typical duration | Best for |
|---|---|---|
| Physical training | 2–5 days, on-site | Rapid upskilling of a whole team; unlimited participants |
| Self-paced licence | ~200 hours of modules | Large or distributed teams who learn at their own pace |
| Hybrid program | 3–6 months | Live workshops plus self-paced depth for a core group |
| Mentoring | 6 months – 1 year | Building or validating your own models on your own data, under NDA |
Every program is customisable to the team's level — junior, mid-level or senior management — and includes live instructor-led sessions, hands-on exercises, assessments with certification and post-training support for applying skills on live projects.
Focus areas
Basel · IFRS 9 · ICAAP · ILAAP · IRRBB · Model Risk · Market Risk · Valuations · Credit Analysis · Machine Learning
Proprietary tools that speed up the journey
- ECL SMART – a full-suite IFRS 9 / CECL solution for model development, validation and audit
- CLEAR OPS – an integrated tool for ICAAP, ILAAP and IRRBB modelling
- CREDIT SPARK – a loan pricing engine for dynamic lending strategies
- RISK RAINBOW – a visual framework to define risk appetite, tolerance and capacity
- MENTOS & CHAMPS – mentoring-led programs to build risk, quant and model capability
Solutioning engagements come with end-to-end ownership, modular Excel and Python builds, full documentation and governance packs — so what is delivered can be understood and maintained internally.
The payoff
Institutions that own their models gain more than compliance:
- Faster responses to auditors and regulators
- Lower dependence on external vendors for every change
- Better decisions — models used for pricing, limits and portfolio strategy, not just reporting
- Retained talent — people stay where they grow
- Audit readiness with clear documentation and validation trails
Start the conversation
- 📥 Download the Peaks2Tails Training Catalogue from the Corporate Engagements page
- 📞 Talk to the team about a customised program for your risk, treasury or credit function
- ▶️ Watch the Risk Rainbow walkthrough and other sessions on the Peaks2Tails YouTube channel
Peaks2Tails — Training | Solutioning | Mentoring. Turning risk into opportunity.
FAQs
Can the training be customised to our portfolio and policies? Yes. Curriculum is tailored to learner level and objectives, and mentoring engagements can work on the institution's own data under NDA.
Do you train in Excel or Python? Both. Programs typically start with transparent Excel builds and then implement the same models in Python.
Who should attend? Credit risk, market risk, model validation, ALM/treasury, finance and regulatory reporting teams — from analysts to senior risk officers.
Is there certification? Yes. Participants are certified after structured assessments, and certificates can be verified online.
Can you help us build or validate a specific model? Yes. Beyond training, Peaks2Tails offers solutioning and long-term mentoring for model development, validation and audit readiness.